/*
This file is part of Magnum .
Copyright © 2010 , 2011 , 2012 , 2013 , 2014 , 2015 , 2016 , 2017 , 2018 , 2019 ,
2020 , 2021 , 2022 , 2023 , 2024
Vladimír Vondruš < mosra @ centrum . cz >
Permission is hereby granted , free of charge , to any person obtaining a
copy of this software and associated documentation files ( the " Software " ) ,
to deal in the Software without restriction , including without limitation
the rights to use , copy , modify , merge , publish , distribute , sublicense ,
and / or sell copies of the Software , and to permit persons to whom the
Software is furnished to do so , subject to the following conditions :
The above copyright notice and this permission notice shall be included
in all copies or substantial portions of the Software .
THE SOFTWARE IS PROVIDED " AS IS " , WITHOUT WARRANTY OF ANY KIND , EXPRESS OR
IMPLIED , INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY ,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT . IN NO EVENT SHALL
THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM , DAMAGES OR OTHER
LIABILITY , WHETHER IN AN ACTION OF CONTRACT , TORT OR OTHERWISE , ARISING
FROM , OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
DEALINGS IN THE SOFTWARE .
*/
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
# include <Corrade/Containers/Optional.h>
# include <Corrade/Containers/StridedArrayView.h>
# include <Corrade/Containers/String.h>
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
# include <Corrade/Containers/Triple.h>
# include <Corrade/TestSuite/Tester.h>
# include <Corrade/TestSuite/Compare/Container.h>
# include <Corrade/TestSuite/Compare/String.h>
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
# include <Corrade/Utility/Algorithms.h>
# include "Magnum/Image.h"
# include "Magnum/ImageView.h"
# include "Magnum/PixelFormat.h"
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
# include "Magnum/DebugTools/CompareImage.h"
# include "Magnum/Math/Range.h"
# include "Magnum/Text/AbstractGlyphCache.h"
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
# include "Magnum/TextureTools/Atlas.h"
# ifdef MAGNUM_BUILD_DEPRECATED
# include <vector>
# include <Corrade/Containers/ArrayViewStl.h>
# include <Corrade/Utility/DebugStl.h> /* std::pair */
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
# endif
namespace Magnum { namespace Text { namespace Test { namespace {
struct AbstractGlyphCacheTest : TestSuite : : Tester {
explicit AbstractGlyphCacheTest ( ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void debugFeature ( ) ;
void debugFeatures ( ) ;
void debugFeaturesSupersets ( ) ;
void construct ( ) ;
void constructNoPadding ( ) ;
void construct2D ( ) ;
void construct2DNoPadding ( ) ;
void constructProcessed ( ) ;
void constructProcessedNoPadding ( ) ;
void constructProcessed2D ( ) ;
void constructProcessed2DNoPadding ( ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
# ifdef MAGNUM_BUILD_DEPRECATED
void constructDeprecated ( ) ;
void constructDeprecatedNoPadding ( ) ;
# endif
void constructImageRowPadding ( ) ;
void constructZeroSize ( ) ;
void constructNoCreate ( ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void constructCopy ( ) ;
void constructMove ( ) ;
void features ( ) ;
# ifdef MAGNUM_BUILD_DEPRECATED
void textureSizeNot2D ( ) ;
# endif
void setInvalidGlyph ( ) ;
void setInvalidGlyph2D ( ) ;
void setInvalidGlyphOutOfRange ( ) ;
void setInvalidGlyphOutOfRangePadded ( ) ;
void setInvalidGlyph2DNot2D ( ) ;
void addFont ( ) ;
void addFontDuplicatePointer ( ) ;
void fontOutOfRange ( ) ;
# ifdef MAGNUM_BUILD_DEPRECATED
void reserve ( ) ;
void reserveIncremental ( ) ;
void reserveTooSmall ( ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void reserveNot2D ( ) ;
# endif
void addGlyph ( ) ;
void addGlyph2D ( ) ;
void addGlyphIndexOutOfRange ( ) ;
void addGlyphAlreadyAdded ( ) ;
void addGlyphOutOfRange ( ) ;
void addGlyphOutOfRangePadded ( ) ;
void addGlyphTooMany ( ) ;
void addGlyph2DNot2D ( ) ;
# ifdef MAGNUM_BUILD_DEPRECATED
void insert ( ) ;
void insertNot2D ( ) ;
void insertMultiFont ( ) ;
# endif
void flushImage ( ) ;
void flushImageWholeArea ( ) ;
void flushImageLayer ( ) ;
void flushImage2D ( ) ;
void flushImage2DPassthrough2D ( ) ;
void flushImageNotImplemented ( ) ;
void flushImagePassthrough2DNotImplemented ( ) ;
void flushImageOutOfRange ( ) ;
void flushImage2DNot2D ( ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
# ifdef MAGNUM_BUILD_DEPRECATED
void setImage ( ) ;
void setImageOutOfRange ( ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void setImageInvalidFormat ( ) ;
void setImageNot2D ( ) ;
# endif
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void processedImage ( ) ;
void processedImageNotSupported ( ) ;
void processedImageNotImplemented ( ) ;
void setProcessedImage ( ) ;
void setProcessedImage2D ( ) ;
void setProcessedImage2DPassthrough2D ( ) ;
void setProcessedImageNotImplemented ( ) ;
void setProcessedImagePassthrough2DNotImplemented ( ) ;
void setProcessedImageOutOfRange ( ) ;
void setProcessedImageInvalidFormat ( ) ;
void setProcessedImage2DNot2D ( ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void access ( ) ;
void accessBatch ( ) ;
void accessInvalid ( ) ;
void accessBatchInvalid ( ) ;
# ifdef MAGNUM_BUILD_DEPRECATED
void accessDeprecated ( ) ;
void accessDeprecatedNot2D ( ) ;
# endif
} ;
const struct {
const char * name ;
Vector2i padding ;
bool differentProcessedFormatSize ;
} FlushImageData [ ] {
{ " " , { } , false } ,
{ " with padding " , { 2 , 3 } , false } ,
{ " with different processed format and size " , { } , false }
} ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
const struct {
const char * name ;
GlyphCacheFeatures features ;
} ProcessedImageNotSupportedData [ ] {
{ " no processing " , { } } ,
{ " no processed image download " , GlyphCacheFeature : : ImageProcessing } ,
} ;
const struct {
const char * name ;
Vector2i padding ;
} SetProcessedImageOutOfRangeData [ ] {
{ " " , { } } ,
{ " with padding " , { 2 , 3 } } ,
} ;
AbstractGlyphCacheTest : : AbstractGlyphCacheTest ( ) {
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
addTests ( { & AbstractGlyphCacheTest : : debugFeature ,
& AbstractGlyphCacheTest : : debugFeatures ,
& AbstractGlyphCacheTest : : debugFeaturesSupersets ,
& AbstractGlyphCacheTest : : construct ,
& AbstractGlyphCacheTest : : constructNoPadding ,
& AbstractGlyphCacheTest : : construct2D ,
& AbstractGlyphCacheTest : : construct2DNoPadding ,
& AbstractGlyphCacheTest : : constructProcessed ,
& AbstractGlyphCacheTest : : constructProcessedNoPadding ,
& AbstractGlyphCacheTest : : constructProcessed2D ,
& AbstractGlyphCacheTest : : constructProcessed2DNoPadding ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
# ifdef MAGNUM_BUILD_DEPRECATED
& AbstractGlyphCacheTest : : constructDeprecated ,
& AbstractGlyphCacheTest : : constructDeprecatedNoPadding ,
# endif
& AbstractGlyphCacheTest : : constructImageRowPadding ,
& AbstractGlyphCacheTest : : constructZeroSize ,
& AbstractGlyphCacheTest : : constructNoCreate ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
& AbstractGlyphCacheTest : : constructCopy ,
& AbstractGlyphCacheTest : : constructMove ,
& AbstractGlyphCacheTest : : features ,
# ifdef MAGNUM_BUILD_DEPRECATED
& AbstractGlyphCacheTest : : textureSizeNot2D ,
# endif
& AbstractGlyphCacheTest : : setInvalidGlyph ,
& AbstractGlyphCacheTest : : setInvalidGlyph2D ,
& AbstractGlyphCacheTest : : setInvalidGlyphOutOfRange ,
& AbstractGlyphCacheTest : : setInvalidGlyphOutOfRangePadded ,
& AbstractGlyphCacheTest : : setInvalidGlyph2DNot2D ,
& AbstractGlyphCacheTest : : addFont ,
& AbstractGlyphCacheTest : : addFontDuplicatePointer ,
& AbstractGlyphCacheTest : : fontOutOfRange ,
# ifdef MAGNUM_BUILD_DEPRECATED
& AbstractGlyphCacheTest : : reserve ,
& AbstractGlyphCacheTest : : reserveIncremental ,
& AbstractGlyphCacheTest : : reserveTooSmall ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
& AbstractGlyphCacheTest : : reserveNot2D ,
# endif
& AbstractGlyphCacheTest : : addGlyph ,
& AbstractGlyphCacheTest : : addGlyph2D ,
& AbstractGlyphCacheTest : : addGlyphIndexOutOfRange ,
& AbstractGlyphCacheTest : : addGlyphAlreadyAdded ,
& AbstractGlyphCacheTest : : addGlyphOutOfRange ,
& AbstractGlyphCacheTest : : addGlyphOutOfRangePadded ,
& AbstractGlyphCacheTest : : addGlyphTooMany ,
& AbstractGlyphCacheTest : : addGlyph2DNot2D ,
# ifdef MAGNUM_BUILD_DEPRECATED
& AbstractGlyphCacheTest : : insert ,
& AbstractGlyphCacheTest : : insertNot2D ,
& AbstractGlyphCacheTest : : insertMultiFont ,
# endif
} ) ;
addInstancedTests ( { & AbstractGlyphCacheTest : : flushImage ,
& AbstractGlyphCacheTest : : flushImageWholeArea ,
& AbstractGlyphCacheTest : : flushImageLayer ,
& AbstractGlyphCacheTest : : flushImage2D ,
& AbstractGlyphCacheTest : : flushImage2DPassthrough2D } ,
Containers : : arraySize ( FlushImageData ) ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
addTests ( { & AbstractGlyphCacheTest : : flushImageNotImplemented ,
& AbstractGlyphCacheTest : : flushImagePassthrough2DNotImplemented } ) ;
addInstancedTests ( { & AbstractGlyphCacheTest : : flushImageOutOfRange } ,
Containers : : arraySize ( FlushImageData ) ) ;
addTests ( { & AbstractGlyphCacheTest : : flushImage2DNot2D } ) ;
# ifdef MAGNUM_BUILD_DEPRECATED
addInstancedTests ( { & AbstractGlyphCacheTest : : setImage } ,
Containers : : arraySize ( FlushImageData ) ) ;
addTests ( { & AbstractGlyphCacheTest : : setImageOutOfRange ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
& AbstractGlyphCacheTest : : setImageInvalidFormat ,
& AbstractGlyphCacheTest : : setImageNot2D } ) ;
# endif
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
addTests ( { & AbstractGlyphCacheTest : : processedImage } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
addInstancedTests ( { & AbstractGlyphCacheTest : : processedImageNotSupported } ,
Containers : : arraySize ( ProcessedImageNotSupportedData ) ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
addTests ( { & AbstractGlyphCacheTest : : processedImageNotImplemented ,
& AbstractGlyphCacheTest : : setProcessedImage ,
& AbstractGlyphCacheTest : : setProcessedImage2D ,
& AbstractGlyphCacheTest : : setProcessedImage2DPassthrough2D ,
& AbstractGlyphCacheTest : : setProcessedImageNotImplemented ,
& AbstractGlyphCacheTest : : setProcessedImagePassthrough2DNotImplemented } ) ;
addInstancedTests ( { & AbstractGlyphCacheTest : : setProcessedImageOutOfRange } ,
Containers : : arraySize ( SetProcessedImageOutOfRangeData ) ) ;
addTests ( { & AbstractGlyphCacheTest : : setProcessedImageInvalidFormat ,
& AbstractGlyphCacheTest : : setProcessedImage2DNot2D ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
& AbstractGlyphCacheTest : : access ,
& AbstractGlyphCacheTest : : accessBatch ,
& AbstractGlyphCacheTest : : accessInvalid ,
& AbstractGlyphCacheTest : : accessBatchInvalid ,
# ifdef MAGNUM_BUILD_DEPRECATED
& AbstractGlyphCacheTest : : accessDeprecated ,
& AbstractGlyphCacheTest : : accessDeprecatedNot2D ,
# endif
} ) ;
}
void AbstractGlyphCacheTest : : debugFeature ( ) {
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Debug { & out } < < GlyphCacheFeature : : ImageProcessing < < GlyphCacheFeature ( 0xca ) ;
CORRADE_COMPARE ( out , " Text::GlyphCacheFeature::ImageProcessing Text::GlyphCacheFeature(0xca) \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
void AbstractGlyphCacheTest : : debugFeatures ( ) {
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Debug { & out } < < ( GlyphCacheFeature : : ImageProcessing | GlyphCacheFeature ( 0xf0 ) ) < < GlyphCacheFeatures { } ;
CORRADE_COMPARE ( out , " Text::GlyphCacheFeature::ImageProcessing|Text::GlyphCacheFeature(0xf0) Text::GlyphCacheFeatures{} \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
void AbstractGlyphCacheTest : : debugFeaturesSupersets ( ) {
/* ProcessedImageDownload is a superset of ImageProcessing, only one should
be printed */
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Debug { & out } < < ( GlyphCacheFeature : : ImageProcessing | GlyphCacheFeature : : ProcessedImageDownload ) ;
CORRADE_COMPARE ( out , " Text::GlyphCacheFeature::ProcessedImageDownload \n " ) ;
}
struct DummyGlyphCache : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override { return { } ; }
void doSetImage ( const Vector2i & , const ImageView2D & ) override { }
} ;
struct DummyProcessingGlyphCache : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override {
return GlyphCacheFeature : : ImageProcessing ;
}
void doSetImage ( const Vector2i & , const ImageView2D & ) override { }
} ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void AbstractGlyphCacheTest : : construct ( ) {
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , { 2 , 5 } } ;
CORRADE_COMPARE ( cache . format ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
CORRADE_COMPARE ( cache . processedFormat ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . processedSize ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_COMPARE ( cache . padding ( ) , ( Vector2i { 2 , 5 } ) ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 0 ) ;
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . atlas ( ) . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . filledSize ( ) , ( Vector3i { 1024 , 512 , 0 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . flags ( ) , TextureTools : : AtlasLandfillFlag : : WidestFirst ) ;
CORRADE_COMPARE ( cache . atlas ( ) . padding ( ) , ( Vector2i { 2 , 5 } ) ) ;
CORRADE_COMPARE ( cache . image ( ) . format ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . image ( ) . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* Invalid glyph is always present */
CORRADE_COMPARE ( cache . glyph ( 0 ) , Containers : : triple ( Vector2i { } , 0 , Range2Di { } ) ) ;
CORRADE_COMPARE_AS ( cache . glyphOffsets ( ) , Containers : : arrayView ( {
Vector2i { }
} ) , TestSuite : : Compare : : Container ) ;
CORRADE_COMPARE_AS ( cache . glyphLayers ( ) , Containers : : arrayView ( {
0
} ) , TestSuite : : Compare : : Container ) ;
CORRADE_COMPARE_AS ( cache . glyphRectangles ( ) , Containers : : arrayView ( {
Range2Di { }
} ) , TestSuite : : Compare : : Container ) ;
/* Const overloads */
const DummyGlyphCache & ccache = cache ;
CORRADE_COMPARE ( ccache . atlas ( ) . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
CORRADE_COMPARE ( ccache . atlas ( ) . filledSize ( ) , ( Vector3i { 1024 , 512 , 0 } ) ) ;
CORRADE_COMPARE ( ccache . atlas ( ) . flags ( ) , TextureTools : : AtlasLandfillFlag : : WidestFirst ) ;
CORRADE_COMPARE ( ccache . atlas ( ) . padding ( ) , ( Vector2i { 2 , 5 } ) ) ;
CORRADE_COMPARE ( ccache . image ( ) . format ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( ccache . image ( ) . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void AbstractGlyphCacheTest : : constructNoPadding ( ) {
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } } ;
CORRADE_COMPARE ( cache . format ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
CORRADE_COMPARE ( cache . processedFormat ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . processedSize ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
/* 1 by default to avoid artifacts */
CORRADE_COMPARE ( cache . padding ( ) , Vector2i { 1 } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_COMPARE ( cache . fontCount ( ) , 0 ) ;
/* Invalid glyph is always present */
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . glyph ( 0 ) , Containers : : triple ( Vector2i { } , 0 , Range2Di { } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . filledSize ( ) , ( Vector3i { 1024 , 512 , 0 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . flags ( ) , TextureTools : : AtlasLandfillFlag : : WidestFirst ) ;
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
CORRADE_COMPARE ( cache . atlas ( ) . padding ( ) , Vector2i { 1 } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_COMPARE ( cache . image ( ) . format ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . image ( ) . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
/* Invalid glyph is always present, has zero size in this case as well */
CORRADE_COMPARE ( cache . glyph ( 0 ) , Containers : : triple ( Vector2i { } , 0 , Range2Di { } ) ) ;
CORRADE_COMPARE_AS ( cache . glyphOffsets ( ) , Containers : : arrayView ( {
Vector2i { }
} ) , TestSuite : : Compare : : Container ) ;
CORRADE_COMPARE_AS ( cache . glyphLayers ( ) , Containers : : arrayView ( {
0
} ) , TestSuite : : Compare : : Container ) ;
CORRADE_COMPARE_AS ( cache . glyphRectangles ( ) , Containers : : arrayView ( {
Range2Di { }
} ) , TestSuite : : Compare : : Container ) ;
/* The rest shouldn't be any different */
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void AbstractGlyphCacheTest : : construct2D ( ) {
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 } , { 2 , 5 } } ;
CORRADE_COMPARE ( cache . format ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . size ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
CORRADE_COMPARE ( cache . processedFormat ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . processedSize ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_COMPARE ( cache . padding ( ) , ( Vector2i { 2 , 5 } ) ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 0 ) ;
/* Invalid glyph is always present */
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_COMPARE ( cache . atlas ( ) . size ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . filledSize ( ) , ( Vector3i { 1024 , 0 , 1 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . flags ( ) , TextureTools : : AtlasLandfillFlag : : WidestFirst ) ;
CORRADE_COMPARE ( cache . atlas ( ) . padding ( ) , ( Vector2i { 2 , 5 } ) ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* The rest shouldn't be any different */
}
void AbstractGlyphCacheTest : : construct2DNoPadding ( ) {
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 } } ;
CORRADE_COMPARE ( cache . format ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . size ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
CORRADE_COMPARE ( cache . processedFormat ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . processedSize ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
/* 1 by default to avoid artifacts */
CORRADE_COMPARE ( cache . padding ( ) , Vector2i { 1 } ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 0 ) ;
/* Invalid glyph is always present */
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . atlas ( ) . size ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . filledSize ( ) , ( Vector3i { 1024 , 0 , 1 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . flags ( ) , TextureTools : : AtlasLandfillFlag : : WidestFirst ) ;
CORRADE_COMPARE ( cache . atlas ( ) . padding ( ) , Vector2i { 1 } ) ;
/* The rest shouldn't be any different */
}
void AbstractGlyphCacheTest : : constructProcessed ( ) {
DummyProcessingGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , PixelFormat : : R16F , { 256 , 128 } , { 2 , 5 } } ;
CORRADE_COMPARE ( cache . format ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
CORRADE_COMPARE ( cache . processedFormat ( ) , PixelFormat : : R16F ) ;
CORRADE_COMPARE ( cache . processedSize ( ) , ( Vector3i { 256 , 128 , 3 } ) ) ;
CORRADE_COMPARE ( cache . padding ( ) , ( Vector2i { 2 , 5 } ) ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 0 ) ;
/* Invalid glyph is always present */
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . atlas ( ) . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . filledSize ( ) , ( Vector3i { 1024 , 512 , 0 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . flags ( ) , TextureTools : : AtlasLandfillFlag : : WidestFirst ) ;
CORRADE_COMPARE ( cache . atlas ( ) . padding ( ) , ( Vector2i { 2 , 5 } ) ) ;
/* The rest shouldn't be any different */
}
void AbstractGlyphCacheTest : : constructProcessedNoPadding ( ) {
DummyProcessingGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , PixelFormat : : R16F , { 256 , 128 } } ;
CORRADE_COMPARE ( cache . format ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
CORRADE_COMPARE ( cache . processedFormat ( ) , PixelFormat : : R16F ) ;
CORRADE_COMPARE ( cache . processedSize ( ) , ( Vector3i { 256 , 128 , 3 } ) ) ;
/* 1 by default to avoid artifacts */
CORRADE_COMPARE ( cache . padding ( ) , Vector2i { 1 } ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 0 ) ;
/* Invalid glyph is always present */
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . atlas ( ) . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . filledSize ( ) , ( Vector3i { 1024 , 512 , 0 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . flags ( ) , TextureTools : : AtlasLandfillFlag : : WidestFirst ) ;
CORRADE_COMPARE ( cache . atlas ( ) . padding ( ) , Vector2i { 1 } ) ;
/* The rest shouldn't be any different */
}
void AbstractGlyphCacheTest : : constructProcessed2D ( ) {
DummyProcessingGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 } , PixelFormat : : R16F , { 256 , 128 } , { 2 , 5 } } ;
CORRADE_COMPARE ( cache . format ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . size ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
CORRADE_COMPARE ( cache . processedFormat ( ) , PixelFormat : : R16F ) ;
CORRADE_COMPARE ( cache . processedSize ( ) , ( Vector3i { 256 , 128 , 1 } ) ) ;
CORRADE_COMPARE ( cache . padding ( ) , ( Vector2i { 2 , 5 } ) ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 0 ) ;
/* Invalid glyph is always present */
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . atlas ( ) . size ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . filledSize ( ) , ( Vector3i { 1024 , 0 , 1 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . flags ( ) , TextureTools : : AtlasLandfillFlag : : WidestFirst ) ;
CORRADE_COMPARE ( cache . atlas ( ) . padding ( ) , ( Vector2i { 2 , 5 } ) ) ;
/* The rest shouldn't be any different */
}
void AbstractGlyphCacheTest : : constructProcessed2DNoPadding ( ) {
DummyProcessingGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 } , PixelFormat : : R16F , { 256 , 128 } } ;
CORRADE_COMPARE ( cache . format ( ) , PixelFormat : : R32F ) ;
CORRADE_COMPARE ( cache . size ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
CORRADE_COMPARE ( cache . processedFormat ( ) , PixelFormat : : R16F ) ;
CORRADE_COMPARE ( cache . processedSize ( ) , ( Vector3i { 256 , 128 , 1 } ) ) ;
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
/* 1 by default to avoid artifacts */
CORRADE_COMPARE ( cache . padding ( ) , Vector2i { 1 } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_COMPARE ( cache . fontCount ( ) , 0 ) ;
/* Invalid glyph is always present */
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_COMPARE ( cache . atlas ( ) . size ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . filledSize ( ) , ( Vector3i { 1024 , 0 , 1 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . flags ( ) , TextureTools : : AtlasLandfillFlag : : WidestFirst ) ;
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
CORRADE_COMPARE ( cache . atlas ( ) . padding ( ) , Vector2i { 1 } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* The rest shouldn't be any different */
}
# ifdef MAGNUM_BUILD_DEPRECATED
void AbstractGlyphCacheTest : : constructDeprecated ( ) {
/* Testing just the minimal set of getters as the deprecated constructor
should delegate */
CORRADE_IGNORE_DEPRECATED_PUSH
/* Not using the DummyGlyphCache as it'd warn about the deprecated
constructor even with the IGNORE macros */
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override { return { } ; }
void doSetImage ( const Vector2i & , const ImageView2D & ) override { }
} cache { { 1024 , 512 } , { 2 , 5 } } ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( cache . format ( ) , PixelFormat : : R8Unorm ) ;
CORRADE_COMPARE ( cache . size ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
CORRADE_IGNORE_DEPRECATED_PUSH
CORRADE_COMPARE ( cache . textureSize ( ) , ( Vector2i { 1024 , 512 } ) ) ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( cache . padding ( ) , ( Vector2i { 2 , 5 } ) ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 0 ) ;
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . atlas ( ) . size ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . filledSize ( ) , ( Vector3i { 1024 , 0 , 1 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . flags ( ) , TextureTools : : AtlasLandfillFlag : : WidestFirst ) ;
CORRADE_COMPARE ( cache . atlas ( ) . padding ( ) , ( Vector2i { 2 , 5 } ) ) ;
}
void AbstractGlyphCacheTest : : constructDeprecatedNoPadding ( ) {
/* Testing just the minimal set of getters as the deprecated constructor
should delegate */
CORRADE_IGNORE_DEPRECATED_PUSH
/* Not using the DummyGlyphCache as it'd warn about the deprecated
constructor even with the IGNORE macros */
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override { return { } ; }
void doSetImage ( const Vector2i & , const ImageView2D & ) override { }
} cache { { 1024 , 512 } } ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( cache . format ( ) , PixelFormat : : R8Unorm ) ;
CORRADE_COMPARE ( cache . size ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
CORRADE_IGNORE_DEPRECATED_PUSH
CORRADE_COMPARE ( cache . textureSize ( ) , ( Vector2i { 1024 , 512 } ) ) ;
CORRADE_IGNORE_DEPRECATED_POP
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
/* 1 by default to avoid artifacts */
CORRADE_COMPARE ( cache . padding ( ) , Vector2i { 1 } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_COMPARE ( cache . fontCount ( ) , 0 ) ;
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . atlas ( ) . size ( ) , ( Vector3i { 1024 , 512 , 1 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . filledSize ( ) , ( Vector3i { 1024 , 0 , 1 } ) ) ;
CORRADE_COMPARE ( cache . atlas ( ) . flags ( ) , TextureTools : : AtlasLandfillFlag : : WidestFirst ) ;
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
CORRADE_COMPARE ( cache . atlas ( ) . padding ( ) , Vector2i { 1 } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
# endif
void AbstractGlyphCacheTest : : constructImageRowPadding ( ) {
/* This shouldn't assert due to the data for the image being too small */
DummyGlyphCache cache { PixelFormat : : RGB8Unorm , { 2 , 3 , 5 } } ;
CORRADE_COMPARE ( cache . size ( ) , ( Vector3i { 2 , 3 , 5 } ) ) ;
CORRADE_COMPARE ( cache . image ( ) . format ( ) , PixelFormat : : RGB8Unorm ) ;
CORRADE_COMPARE ( cache . image ( ) . size ( ) , ( Vector3i { 2 , 3 , 5 } ) ) ;
CORRADE_COMPARE ( cache . image ( ) . data ( ) . size ( ) , 8 * 3 * 5 ) ; /* not 6*3*5 */
}
void AbstractGlyphCacheTest : : constructZeroSize ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
DummyGlyphCache { PixelFormat : : R8Unorm , { 2 , 0 , 1 } } ;
DummyGlyphCache { PixelFormat : : R8Unorm , { 0 , 2 , 1 } } ;
DummyGlyphCache { PixelFormat : : R8Unorm , { 2 , 2 , 0 } } ;
DummyGlyphCache { PixelFormat : : R8Unorm , { 2 , 2 } , PixelFormat : : R8Unorm , { 2 , 0 } } ;
DummyGlyphCache { PixelFormat : : R8Unorm , { 2 , 2 } , PixelFormat : : R8Unorm , { 0 , 2 } } ;
CORRADE_COMPARE ( out ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
" Text::AbstractGlyphCache: expected non-zero size, got {2, 0, 1} \n "
" Text::AbstractGlyphCache: expected non-zero size, got {0, 2, 1} \n "
" Text::AbstractGlyphCache: expected non-zero size, got {2, 2, 0} \n "
" Text::AbstractGlyphCache: expected non-zero processed size, got {2, 0} \n "
" Text::AbstractGlyphCache: expected non-zero processed size, got {0, 2} \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
void AbstractGlyphCacheTest : : constructNoCreate ( ) {
DummyGlyphCache cache { NoCreate } ;
/* Shouldn't crash */
CORRADE_VERIFY ( true ) ;
/* Implicit construction is not allowed */
CORRADE_VERIFY ( ! std : : is_convertible < NoCreateT , DummyGlyphCache > : : value ) ;
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void AbstractGlyphCacheTest : : constructCopy ( ) {
CORRADE_VERIFY ( ! std : : is_copy_constructible < DummyGlyphCache > { } ) ;
CORRADE_VERIFY ( ! std : : is_copy_assignable < DummyGlyphCache > { } ) ;
}
void AbstractGlyphCacheTest : : constructMove ( ) {
DummyGlyphCache a { PixelFormat : : R16F , { 1024 , 512 , 3 } , { 2 , 5 } } ;
DummyGlyphCache b = Utility : : move ( a ) ;
CORRADE_COMPARE ( b . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
DummyGlyphCache c { PixelFormat : : R8Unorm , { 2 , 3 } } ;
c = Utility : : move ( b ) ;
CORRADE_COMPARE ( c . size ( ) , ( Vector3i { 1024 , 512 , 3 } ) ) ;
CORRADE_VERIFY ( std : : is_nothrow_move_constructible < DummyGlyphCache > : : value ) ;
CORRADE_VERIFY ( std : : is_nothrow_move_assignable < DummyGlyphCache > : : value ) ;
}
void AbstractGlyphCacheTest : : features ( ) {
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override { return GlyphCacheFeature : : ImageProcessing ; }
void doSetImage ( const Vector2i & , const ImageView2D & ) override { }
} cache { PixelFormat : : R8Unorm , { 2 , 3 } } ;
CORRADE_COMPARE ( cache . features ( ) , GlyphCacheFeature : : ImageProcessing ) ;
}
# ifdef MAGNUM_BUILD_DEPRECATED
void AbstractGlyphCacheTest : : textureSizeNot2D ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } } ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
CORRADE_IGNORE_DEPRECATED_PUSH
cache . textureSize ( ) ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::textureSize(): can't be used on an array glyph cache \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
# endif
void AbstractGlyphCacheTest : : setInvalidGlyph ( ) {
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , { 2 , 3 } } ;
cache . setInvalidGlyph ( { 3 , 5 } , 2 , { { 15 , 30 } , { 45 , 35 } } ) ;
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . glyph ( 0 ) , Containers : : triple (
Vector2i { 1 , 2 } ,
2 ,
Range2Di { { 13 , 27 } , { 47 , 38 } } ) ) ;
/* Invalid glyph spanning the whole area (with padding) shouldn't assert */
cache . setInvalidGlyph ( { 3 , 5 } , 2 , { { 2 , 3 } , { 1022 , 509 } } ) ;
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . glyph ( 0 ) , Containers : : triple (
Vector2i { 1 , 2 } ,
2 ,
Range2Di { { } , { 1024 , 512 } } ) ) ;
}
void AbstractGlyphCacheTest : : setInvalidGlyph2D ( ) {
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 } , { 2 , 3 } } ;
cache . setInvalidGlyph ( { 3 , 5 } , { { 15 , 30 } , { 45 , 35 } } ) ;
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . glyph ( 0 ) , Containers : : triple (
Vector2i { 1 , 2 } ,
0 ,
Range2Di { { 13 , 27 } , { 47 , 38 } } ) ) ;
/* Invalid glyph spanning the whole area is tested above already */
}
void AbstractGlyphCacheTest : : setInvalidGlyphOutOfRange ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
/* Default padding is 1, test that it works for zero as well */
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , { } } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . setInvalidGlyph ( { } , - 1 , { { 15 , 30 } , { 45 , 35 } } ) ;
cache . setInvalidGlyph ( { } , 3 , { { 15 , 30 } , { 45 , 35 } } ) ;
cache . setInvalidGlyph ( { } , 0 , { { 15 , - 1 } , { 45 , 35 } } ) ;
cache . setInvalidGlyph ( { } , 0 , { { - 1 , 30 } , { 45 , 35 } } ) ;
cache . setInvalidGlyph ( { } , 0 , { { 15 , 30 } , { 1025 , 35 } } ) ;
cache . setInvalidGlyph ( { } , 0 , { { 15 , 30 } , { 45 , 513 } } ) ;
/* Negative rect size */
cache . setInvalidGlyph ( { } , 0 , { { 45 , 30 } , { 15 , 35 } } ) ;
cache . setInvalidGlyph ( { } , 0 , { { 15 , 35 } , { 45 , 30 } } ) ;
CORRADE_COMPARE_AS ( out ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
" Text::AbstractGlyphCache::setInvalidGlyph(): layer -1 and rectangle {{15, 30}, {45, 35}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 3 and rectangle {{15, 30}, {45, 35}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 0 and rectangle {{15, -1}, {45, 35}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 0 and rectangle {{-1, 30}, {45, 35}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 0 and rectangle {{15, 30}, {1025, 35}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 0 and rectangle {{15, 30}, {45, 513}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 0 and rectangle {{45, 30}, {15, 35}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 0 and rectangle {{15, 35}, {45, 30}} out of range for size {1024, 512, 3} and padding {0, 0} \n " ,
TestSuite : : Compare : : String ) ;
}
void AbstractGlyphCacheTest : : setInvalidGlyphOutOfRangePadded ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , { 2 , 3 } } ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
/* Padding has no effect on layers */
cache . setInvalidGlyph ( { } , - 1 , { { 15 , 30 } , { 45 , 35 } } ) ;
cache . setInvalidGlyph ( { } , 3 , { { 15 , 30 } , { 45 , 35 } } ) ;
/* These four pass if padding is not included in the check */
cache . setInvalidGlyph ( { } , 0 , { { 15 , 1 } , { 45 , 35 } } ) ;
cache . setInvalidGlyph ( { } , 0 , { { 1 , 30 } , { 45 , 35 } } ) ;
cache . setInvalidGlyph ( { } , 0 , { { 15 , 30 } , { 1023 , 35 } } ) ;
cache . setInvalidGlyph ( { } , 0 , { { 15 , 30 } , { 45 , 510 } } ) ;
/* Negative rect size. The second would pass if it was checked with
padding included . */
cache . setInvalidGlyph ( { } , 0 , { { 45 , 30 } , { 15 , 35 } } ) ;
cache . setInvalidGlyph ( { } , 0 , { { 15 , 35 } , { 45 , 30 } } ) ;
CORRADE_COMPARE_AS ( out ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
" Text::AbstractGlyphCache::setInvalidGlyph(): layer -1 and rectangle {{15, 30}, {45, 35}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 3 and rectangle {{15, 30}, {45, 35}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 0 and rectangle {{15, 1}, {45, 35}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 0 and rectangle {{1, 30}, {45, 35}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 0 and rectangle {{15, 30}, {1023, 35}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 0 and rectangle {{15, 30}, {45, 510}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 0 and rectangle {{45, 30}, {15, 35}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::setInvalidGlyph(): layer 0 and rectangle {{15, 35}, {45, 30}} out of range for size {1024, 512, 3} and padding {2, 3} \n " ,
TestSuite : : Compare : : String ) ;
}
void AbstractGlyphCacheTest : : setInvalidGlyph2DNot2D ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } } ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . setInvalidGlyph ( { } , { } ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::setInvalidGlyph(): use the layer overload for an array glyph cache \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
void AbstractGlyphCacheTest : : addFont ( ) {
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 } } ;
const AbstractFont * font = reinterpret_cast < const AbstractFont * > ( std : : size_t { 0xdeadbeef } ) ;
CORRADE_COMPARE ( cache . findFont ( * font ) , Containers : : NullOpt ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_COMPARE ( cache . addFont ( 35 , nullptr ) , 0 ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . fontGlyphCount ( 0 ) , 35 ) ;
CORRADE_COMPARE ( cache . fontPointer ( 0 ) , nullptr ) ;
CORRADE_COMPARE ( cache . findFont ( * font ) , Containers : : NullOpt ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_COMPARE ( cache . addFont ( 12 , font ) , 1 ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 2 ) ;
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . fontGlyphCount ( 1 ) , 12 ) ;
CORRADE_COMPARE ( cache . fontPointer ( 1 ) , font ) ;
CORRADE_COMPARE ( cache . findFont ( * font ) , 1 ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
void AbstractGlyphCacheTest : : addFontDuplicatePointer ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 } } ;
cache . addFont ( 7 , nullptr ) ;
const AbstractFont * font = reinterpret_cast < const AbstractFont * > ( std : : size_t { 0xdeadbeef } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
cache . addFont ( 35 , font ) ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . addFont ( 12 , font ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::addFont(): pointer 0xdeadbeef already used for font 1 \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
void AbstractGlyphCacheTest : : fontOutOfRange ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 } } ;
const AbstractFont * font = reinterpret_cast < const AbstractFont * > ( std : : size_t { 0xdeadbeef } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
cache . addFont ( 35 , nullptr ) ;
cache . addFont ( 12 , font ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 2 ) ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . fontGlyphCount ( 2 ) ;
cache . fontPointer ( 2 ) ;
CORRADE_COMPARE ( out ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
" Text::AbstractGlyphCache::fontGlyphCount(): index 2 out of range for 2 fonts \n "
" Text::AbstractGlyphCache::fontPointer(): index 2 out of range for 2 fonts \n " ) ;
}
# ifdef MAGNUM_BUILD_DEPRECATED
void AbstractGlyphCacheTest : : reserve ( ) {
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
DummyGlyphCache cache { PixelFormat : : R8Unorm , { 24 , 20 } , { 1 , 2 } } ;
/* Padding should get applied to all */
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_PUSH
std : : vector < Range2Di > out = cache . reserve ( {
{ 5 , 3 } ,
/* Landscape glyphs shouldn't get rotated */
{ 12 , 6 } ,
{ 10 , 5 } ,
/* Zero-sized glyphs should get preserved */
{ 0 , 1 } ,
{ 3 , 0 } ,
} ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE_AS ( Containers : : arrayView ( out ) , Containers : : arrayView < Range2Di > ( {
Range2Di : : fromSize ( { 6 , 12 } , { 5 , 3 } ) ,
Range2Di : : fromSize ( { 1 , 2 } , { 12 , 6 } ) ,
Range2Di : : fromSize ( { 13 , 12 } , { 10 , 5 } ) ,
Range2Di : : fromSize ( { 4 , 12 } , { 0 , 1 } ) ,
Range2Di : : fromSize ( { 1 , 17 } , { 3 , 0 } ) ,
} ) , TestSuite : : Compare : : Container ) ;
}
void AbstractGlyphCacheTest : : reserveIncremental ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
DummyGlyphCache cache { PixelFormat : : R8Unorm , { 24 , 20 } , { 1 , 2 } } ;
/* insert() is what triggers the assert, not reserve() alone */
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_PUSH
cache . insert ( 34 , { 3 , 5 } , { { 10 , 10 } , { 23 , 10 } } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_POP
Containers : : String out ;
Error redirectError { & out } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_PUSH
cache . reserve ( { { 12 , 6 } } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::reserve(): reserving space in non-empty cache is not yet implemented \n " ) ;
}
void AbstractGlyphCacheTest : : reserveTooSmall ( ) {
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
DummyGlyphCache cache { PixelFormat : : R8Unorm , { 24 , 18 } , { 1 , 2 } } ;
Containers : : String out ;
Error redirectError { & out } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_PUSH
CORRADE_VERIFY ( cache . reserve ( { { 5 , 3 } , { 12 , 6 } , { 10 , 5 } } ) . empty ( ) ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::reserve(): requested atlas size Vector(24, 18) is too small to fit 3 textures. Generated atlas will be empty. \n " ) ;
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void AbstractGlyphCacheTest : : reserveNot2D ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } } ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
CORRADE_IGNORE_DEPRECATED_PUSH
cache . reserve ( { } ) ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::reserve(): can't be used on an array glyph cache \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
# endif
void AbstractGlyphCacheTest : : addGlyph ( ) {
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , { 2 , 3 } } ;
UnsignedInt font9 = cache . addFont ( 9 ) ;
UnsignedInt font3 = cache . addFont ( 3 ) ;
/* The queried values are with padding applied */
UnsignedInt font9Glyph6 = cache . addGlyph ( font9 , 6 , { 3 , 4 } , 2 , { { 15 , 30 } , { 45 , 35 } } ) ;
CORRADE_COMPARE ( font9Glyph6 , 1 ) ;
CORRADE_COMPARE ( cache . glyph ( font9Glyph6 ) , Containers : : triple (
Vector2i { 1 , 1 } ,
2 ,
Range2Di { { 13 , 27 } , { 47 , 38 } } ) ) ;
/* Glyph in another font */
UnsignedInt font3Glyph1 = cache . addGlyph ( font3 , 1 , { 5 , 6 } , 1 , { { 10 , 15 } , { 25 , 30 } } ) ;
CORRADE_COMPARE ( font3Glyph1 , 2 ) ;
CORRADE_COMPARE ( cache . glyph ( font3Glyph1 ) , Containers : : triple (
Vector2i { 3 , 3 } ,
1 ,
Range2Di { { 8 , 12 } , { 27 , 33 } } ) ) ;
/* Glyph adding order shouldn't matter; glyph spanning the whole area (with
padding ) shouldn ' t assert */
UnsignedInt font3Glyph0 = cache . addGlyph ( font3 , 0 , { 3 , 5 } , 2 , { { 2 , 3 } , { 1022 , 509 } } ) ;
CORRADE_COMPARE ( font3Glyph0 , 3 ) ;
CORRADE_COMPARE ( cache . glyph ( font3Glyph0 ) , Containers : : triple (
Vector2i { 1 , 2 } ,
2 ,
Range2Di { { } , { 1024 , 512 } } ) ) ;
/* Another glyph in an earlier font */
UnsignedInt font9Glyph3 = cache . addGlyph ( font9 , 3 , { 5 , 7 } , 0 , { { 5 , 10 } , { 15 , 30 } } ) ;
CORRADE_COMPARE ( font9Glyph3 , 4 ) ;
CORRADE_COMPARE ( cache . glyph ( font9Glyph3 ) , Containers : : triple (
Vector2i { 3 , 4 } ,
0 ,
Range2Di { { 3 , 7 } , { 17 , 33 } } ) ) ;
}
void AbstractGlyphCacheTest : : addGlyph2D ( ) {
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 } , { 2 , 3 } } ;
cache . addFont ( 9 ) ;
UnsignedInt fontId = cache . addFont ( 3 ) ;
CORRADE_COMPARE ( cache . addGlyph ( fontId , 2 , { 3 , 5 } , { { 15 , 30 } , { 45 , 35 } } ) , 1 ) ;
CORRADE_COMPARE ( cache . glyphCount ( ) , 2 ) ;
CORRADE_COMPARE ( cache . glyph ( 1 ) , Containers : : triple (
Vector2i { 1 , 2 } ,
0 ,
Range2Di { { 13 , 27 } , { 47 , 38 } } ) ) ;
}
void AbstractGlyphCacheTest : : addGlyphIndexOutOfRange ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } } ;
cache . addFont ( 9 ) ;
UnsignedInt fontId = cache . addFont ( 3 ) ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . addGlyph ( cache . fontCount ( ) , 0 , { } , 2 , { } ) ;
cache . addGlyph ( fontId , cache . fontGlyphCount ( fontId ) , { } , 2 , { } ) ;
CORRADE_COMPARE ( out ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
" Text::AbstractGlyphCache::addGlyph(): index 2 out of range for 2 fonts \n "
" Text::AbstractGlyphCache::addGlyph(): index 3 out of range for 3 glyphs in font 1 \n " ) ;
}
void AbstractGlyphCacheTest : : addGlyphAlreadyAdded ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
/* Default padding of 1 makes it impossible to add a glyph at zero offset
as it ' s out of range . Don ' t want to bother with that here so resetting
it to 0. */
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , { } } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
cache . addFont ( 9 ) ;
UnsignedInt fontId = cache . addFont ( 3 ) ;
cache . addGlyph ( fontId , 0 , { } , 2 , { } ) ;
cache . addGlyph ( fontId , 1 , { } , 2 , { } ) ;
cache . addGlyph ( fontId , 2 , { } , 2 , { } ) ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . addGlyph ( fontId , 2 , { } , 2 , { } ) ;
CORRADE_COMPARE ( out ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
" Text::AbstractGlyphCache::addGlyph(): glyph 2 in font 1 already added at index 3 \n " ) ;
}
void AbstractGlyphCacheTest : : addGlyphOutOfRange ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
/* Default padding is 1, test that it works for zero as well */
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , { } } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
UnsignedInt fontId = cache . addFont ( 9 ) ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . addGlyph ( fontId , 1 , { } , - 1 , { { 15 , 30 } , { 45 , 35 } } ) ;
cache . addGlyph ( fontId , 2 , { } , 3 , { { 15 , 30 } , { 45 , 35 } } ) ;
cache . addGlyph ( fontId , 3 , { } , 0 , { { 15 , - 1 } , { 45 , 35 } } ) ;
cache . addGlyph ( fontId , 4 , { } , 0 , { { - 1 , 30 } , { 45 , 35 } } ) ;
cache . addGlyph ( fontId , 5 , { } , 0 , { { 15 , 30 } , { 1025 , 35 } } ) ;
cache . addGlyph ( fontId , 6 , { } , 0 , { { 15 , 30 } , { 45 , 513 } } ) ;
/* Negative rect size */
cache . addGlyph ( fontId , 8 , { } , 0 , { { 45 , 30 } , { 15 , 35 } } ) ;
cache . addGlyph ( fontId , 7 , { } , 0 , { { 15 , 35 } , { 45 , 30 } } ) ;
CORRADE_COMPARE_AS ( out ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
" Text::AbstractGlyphCache::addGlyph(): layer -1 and rectangle {{15, 30}, {45, 35}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 3 and rectangle {{15, 30}, {45, 35}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 0 and rectangle {{15, -1}, {45, 35}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 0 and rectangle {{-1, 30}, {45, 35}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 0 and rectangle {{15, 30}, {1025, 35}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 0 and rectangle {{15, 30}, {45, 513}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 0 and rectangle {{45, 30}, {15, 35}} out of range for size {1024, 512, 3} and padding {0, 0} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 0 and rectangle {{15, 35}, {45, 30}} out of range for size {1024, 512, 3} and padding {0, 0} \n " ,
TestSuite : : Compare : : String ) ;
}
void AbstractGlyphCacheTest : : addGlyphOutOfRangePadded ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , { 2 , 3 } } ;
UnsignedInt fontId = cache . addFont ( 9 ) ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
/* Padding has no effect on layers */
cache . addGlyph ( fontId , 1 , { } , - 1 , { { 15 , 30 } , { 45 , 35 } } ) ;
cache . addGlyph ( fontId , 2 , { } , 3 , { { 15 , 30 } , { 45 , 35 } } ) ;
/* These four pass if padding is not included in the check */
cache . addGlyph ( fontId , 3 , { } , 0 , { { 15 , 1 } , { 45 , 35 } } ) ;
cache . addGlyph ( fontId , 4 , { } , 0 , { { 1 , 30 } , { 45 , 35 } } ) ;
cache . addGlyph ( fontId , 5 , { } , 0 , { { 15 , 30 } , { 1023 , 35 } } ) ;
cache . addGlyph ( fontId , 6 , { } , 0 , { { 15 , 30 } , { 45 , 510 } } ) ;
/* Negative rect size. The second would pass if it was checked with
padding included . */
cache . addGlyph ( fontId , 8 , { } , 0 , { { 45 , 30 } , { 15 , 35 } } ) ;
cache . addGlyph ( fontId , 7 , { } , 0 , { { 15 , 35 } , { 45 , 30 } } ) ;
CORRADE_COMPARE_AS ( out ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
" Text::AbstractGlyphCache::addGlyph(): layer -1 and rectangle {{15, 30}, {45, 35}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 3 and rectangle {{15, 30}, {45, 35}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 0 and rectangle {{15, 1}, {45, 35}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 0 and rectangle {{1, 30}, {45, 35}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 0 and rectangle {{15, 30}, {1023, 35}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 0 and rectangle {{15, 30}, {45, 510}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 0 and rectangle {{45, 30}, {15, 35}} out of range for size {1024, 512, 3} and padding {2, 3} \n "
" Text::AbstractGlyphCache::addGlyph(): layer 0 and rectangle {{15, 35}, {45, 30}} out of range for size {1024, 512, 3} and padding {2, 3} \n " ,
TestSuite : : Compare : : String ) ;
}
void AbstractGlyphCacheTest : : addGlyphTooMany ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
/* Default padding of 1 makes it impossible to add a glyph at zero offset
as it ' s out of range . Don ' t want to bother with that here so resetting
it to 0. */
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 } , { } } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* Adding a font with over 65k potential glyphs is okay */
UnsignedInt fontId = cache . addFont ( 100000 ) ;
for ( UnsignedInt i = 0 ; i ! = 65535 ; + + i )
cache . addGlyph ( fontId , i , { } , { } ) ;
CORRADE_COMPARE ( cache . glyphCount ( ) , 65536 ) ;
/* But adding 65k actual glyphs isn't */
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . addGlyph ( fontId , 65536 , { } , { } ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::addGlyph(): only at most 65536 glyphs can be added \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
void AbstractGlyphCacheTest : : addGlyph2DNot2D ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } } ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . addGlyph ( 0 , 0 , { } , { } ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::addGlyph(): use the layer overload for an array glyph cache \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
# ifdef MAGNUM_BUILD_DEPRECATED
void AbstractGlyphCacheTest : : insert ( ) {
DummyGlyphCache cache { PixelFormat : : R8Unorm , { 100 , 200 } , { 2 , 3 } } ;
/* Overwriting the "Not Found" glyph. Shouldn't result in any font or glyph
being added . */
CORRADE_IGNORE_DEPRECATED_PUSH
cache . insert ( 0 , { 3 , 5 } , { { 10 , 10 } , { 23 , 45 } } ) ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( cache . glyphCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 0 ) ;
CORRADE_COMPARE ( cache . glyph ( 0 ) , Containers : : triple (
Vector2i { 1 , 2 } ,
0 ,
Range2Di { { 8 , 7 } , { 25 , 48 } } ) ) ;
/* Adding a new glyph adds the first font if not there yet, setting its
glyph count to fit the glyph ID */
CORRADE_IGNORE_DEPRECATED_PUSH
cache . insert ( 25 , { 3 , 4 } , { { 15 , 30 } , { 45 , 35 } } ) ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( cache . glyphCount ( ) , 2 ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . fontGlyphCount ( 0 ) , 26 ) ;
CORRADE_COMPARE ( cache . glyph ( 0 , 25 ) , Containers : : triple (
Vector2i { 1 , 1 } ,
0 ,
Range2Di { { 13 , 27 } , { 47 , 38 } } ) ) ;
/* Adding another glyph with a lower ID doesn't change the font in any
way */
CORRADE_IGNORE_DEPRECATED_PUSH
cache . insert ( 5 , { 5 , 6 } , { { 10 , 15 } , { 25 , 30 } } ) ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( cache . glyphCount ( ) , 3 ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . fontGlyphCount ( 0 ) , 26 ) ;
CORRADE_COMPARE ( cache . glyph ( 0 , 5 ) , Containers : : triple (
Vector2i { 3 , 3 } ,
0 ,
Range2Di { { 8 , 12 } , { 27 , 33 } } ) ) ;
/* Adding a glyph with greater ID expands the font glyph count again */
CORRADE_IGNORE_DEPRECATED_PUSH
cache . insert ( 35 , { 5 , 7 } , { { 5 , 10 } , { 15 , 30 } } ) ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( cache . glyphCount ( ) , 4 ) ;
CORRADE_COMPARE ( cache . fontCount ( ) , 1 ) ;
CORRADE_COMPARE ( cache . fontGlyphCount ( 0 ) , 36 ) ;
CORRADE_COMPARE ( cache . glyph ( 0 , 35 ) , Containers : : triple (
Vector2i { 3 , 4 } ,
0 ,
Range2Di { { 3 , 7 } , { 17 , 33 } } ) ) ;
}
void AbstractGlyphCacheTest : : insertNot2D ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } } ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
CORRADE_IGNORE_DEPRECATED_PUSH
cache . insert ( 0 , { } , { } ) ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::insert(): can't be used on an array glyph cache \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
void AbstractGlyphCacheTest : : insertMultiFont ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 } } ;
cache . addFont ( 15 ) ;
cache . addFont ( 35 ) ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
CORRADE_IGNORE_DEPRECATED_PUSH
cache . insert ( 0 , { } , { } ) ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::insert(): can't be used on a multi-font glyph cache \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
# endif
void AbstractGlyphCacheTest : : flushImage ( ) {
auto & & data = FlushImageData [ testCaseInstanceId ( ) ] ;
setTestCaseDescription ( data . name ) ;
struct Cache : AbstractGlyphCache {
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override { return { } ; }
void doSetImage ( const Vector3i & offset , const ImageView3D & image ) override {
called = true ;
CORRADE_COMPARE ( offset , ( Vector3i { 15 , 30 , 3 } - Vector3i { padding ( ) , 0 } ) ) ;
CORRADE_COMPARE ( image . size ( ) , ( Vector3i { 3 , 2 , 2 } + Vector3i { 2 * padding ( ) , 0 } ) ) ;
if ( padding ( ) . isZero ( ) ) {
char pixels0 [ ] {
' a ' , ' b ' , ' c ' , 0 ,
' d ' , ' e ' , ' f ' , 0 ,
} ;
char pixels1 [ ] {
' 0 ' , ' 1 ' , ' 2 ' , 0 ,
' 3 ' , ' 4 ' , ' 5 ' , 0 ,
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 0 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels0 } ) ,
DebugTools : : CompareImage ) ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 1 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels1 } ) ,
DebugTools : : CompareImage ) ;
} else {
char pixels0 [ ] {
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , ' a ' , ' b ' , ' c ' , 0 , 0 , 0 ,
0 , 0 , ' d ' , ' e ' , ' f ' , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0
} ;
char pixels1 [ ] {
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , ' 0 ' , ' 1 ' , ' 2 ' , 0 , 0 , 0 ,
0 , 0 , ' 3 ' , ' 4 ' , ' 5 ' , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 0 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 7 , 8 } , pixels0 } ) ,
DebugTools : : CompareImage ) ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 1 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 7 , 8 } , pixels1 } ) ,
DebugTools : : CompareImage ) ;
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
bool called = false ;
} cache { NoCreate } ;
if ( data . differentProcessedFormatSize )
cache = Cache { PixelFormat : : R8Snorm , { 45 , 35 , 5 } , PixelFormat : : RG32F , { 12 , 34 } } ;
else
cache = Cache { PixelFormat : : R8Snorm , { 45 , 35 , 5 } , data . padding } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* Capture correct function name */
CORRADE_VERIFY ( true ) ;
/* Copy two slices of the image */
char pixels [ ] {
' a ' , ' b ' , ' c ' ,
' d ' , ' e ' , ' f ' ,
' 0 ' , ' 1 ' , ' 2 ' ,
' 3 ' , ' 4 ' , ' 5 ' ,
} ;
Utility : : copy (
Containers : : StridedArrayView3D < char > { pixels , { 2 , 2 , 3 } } ,
cache . image ( ) . pixels < char > ( ) . sliceSize ( { 3 , 30 , 15 } , { 2 , 2 , 3 } ) ) ;
cache . flushImage ( Range3Di : : fromSize ( { 15 , 30 , 3 } , { 3 , 2 , 2 } ) ) ;
CORRADE_VERIFY ( cache . called ) ;
}
void AbstractGlyphCacheTest : : flushImageWholeArea ( ) {
auto & & data = FlushImageData [ testCaseInstanceId ( ) ] ;
setTestCaseDescription ( data . name ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* Like above, but calling flushImage() with the whole size to test bounds
checking . The padding doesn ' t affect the call in this case - - the actual
range is always the whole image . */
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
struct Cache : AbstractGlyphCache {
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override { return { } ; }
void doSetImage ( const Vector3i & offset , const ImageView3D & image ) override {
called = true ;
char pixels0 [ ] {
' a ' , ' b ' , ' c ' , 0 ,
' d ' , ' e ' , ' f ' , 0 ,
} ;
char pixels1 [ ] {
' 0 ' , ' 1 ' , ' 2 ' , 0 ,
' 3 ' , ' 4 ' , ' 5 ' , 0 ,
} ;
CORRADE_COMPARE ( offset , Vector3i { } ) ;
CORRADE_COMPARE ( image . size ( ) , ( Vector3i { 3 , 2 , 2 } ) ) ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 0 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels0 } ) ,
DebugTools : : CompareImage ) ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 1 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels1 } ) ,
DebugTools : : CompareImage ) ;
}
bool called = false ;
} cache { NoCreate } ;
if ( data . differentProcessedFormatSize )
cache = Cache { PixelFormat : : R8Snorm , { 3 , 2 , 5 } , PixelFormat : : RG32F , { 1 , 1 } } ;
else
cache = Cache { PixelFormat : : R8Snorm , { 3 , 2 , 2 } , data . padding } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* Capture correct function name */
CORRADE_VERIFY ( true ) ;
/* Copy two slices of the image */
char pixels [ ] {
' a ' , ' b ' , ' c ' ,
' d ' , ' e ' , ' f ' ,
' 0 ' , ' 1 ' , ' 2 ' ,
' 3 ' , ' 4 ' , ' 5 ' ,
} ;
Utility : : copy (
Containers : : StridedArrayView3D < char > { pixels , { 2 , 2 , 3 } } ,
cache . image ( ) . pixels < char > ( ) ) ;
cache . flushImage ( { { } , { 3 , 2 , 2 } } ) ;
CORRADE_VERIFY ( cache . called ) ;
}
void AbstractGlyphCacheTest : : flushImageLayer ( ) {
auto & & data = FlushImageData [ testCaseInstanceId ( ) ] ;
setTestCaseDescription ( data . name ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* Single slice subset of flushImage() */
struct Cache : AbstractGlyphCache {
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override { return { } ; }
void doSetImage ( const Vector3i & offset , const ImageView3D & image ) override {
called = true ;
CORRADE_COMPARE ( offset , ( Vector3i { 15 , 30 , 3 } - Vector3i { padding ( ) , 0 } ) ) ;
CORRADE_COMPARE ( image . size ( ) , ( Vector3i { 3 , 2 , 1 } + Vector3i { 2 * padding ( ) , 0 } ) ) ;
if ( padding ( ) . isZero ( ) ) {
char pixels [ ] {
' a ' , ' b ' , ' c ' , 0 ,
' d ' , ' e ' , ' f ' , 0 ,
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 0 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels } ) ,
DebugTools : : CompareImage ) ;
} else {
char pixels [ ] {
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , ' a ' , ' b ' , ' c ' , 0 , 0 , 0 ,
0 , 0 , ' d ' , ' e ' , ' f ' , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 0 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 7 , 8 } , pixels } ) ,
DebugTools : : CompareImage ) ;
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
bool called = false ;
} cache { NoCreate } ;
if ( data . differentProcessedFormatSize )
cache = Cache { PixelFormat : : R8Snorm , { 45 , 35 , 5 } , PixelFormat : : RG32F , { 12 , 34 } } ;
else
cache = Cache { PixelFormat : : R8Snorm , { 45 , 35 , 5 } , data . padding } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* Capture correct function name */
CORRADE_VERIFY ( true ) ;
char pixels [ ] {
' a ' , ' b ' , ' c ' ,
' d ' , ' e ' , ' f ' ,
} ;
Utility : : copy (
Containers : : StridedArrayView3D < char > { pixels , { 1 , 2 , 3 } } ,
cache . image ( ) . pixels < char > ( ) . sliceSize ( { 3 , 30 , 15 } , { 1 , 2 , 3 } ) ) ;
cache . flushImage ( 3 , Range2Di : : fromSize ( { 15 , 30 } , { 3 , 2 } ) ) ;
CORRADE_VERIFY ( cache . called ) ;
}
void AbstractGlyphCacheTest : : flushImage2D ( ) {
auto & & data = FlushImageData [ testCaseInstanceId ( ) ] ;
setTestCaseDescription ( data . name ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* Like flushImageLayer() but reduced to two dimensions */
struct Cache : AbstractGlyphCache {
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override { return { } ; }
void doSetImage ( const Vector3i & offset , const ImageView3D & image ) override {
called = true ;
CORRADE_COMPARE ( offset , ( Vector3i { 15 , 30 , 0 } - Vector3i { padding ( ) , 0 } ) ) ;
CORRADE_COMPARE ( image . size ( ) , ( Vector3i { 3 , 2 , 1 } + Vector3i { 2 * padding ( ) , 0 } ) ) ;
if ( padding ( ) . isZero ( ) ) {
char pixels [ ] {
' a ' , ' b ' , ' c ' , 0 ,
' d ' , ' e ' , ' f ' , 0 ,
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 0 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels } ) ,
DebugTools : : CompareImage ) ;
} else {
char pixels [ ] {
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , ' a ' , ' b ' , ' c ' , 0 , 0 , 0 ,
0 , 0 , ' d ' , ' e ' , ' f ' , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 0 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 7 , 8 } , pixels } ) ,
DebugTools : : CompareImage ) ;
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
bool called = false ;
} cache { NoCreate } ;
if ( data . differentProcessedFormatSize )
cache = Cache { PixelFormat : : R8Snorm , { 45 , 35 } , PixelFormat : : RG32F , { 12 , 34 } } ;
else
cache = Cache { PixelFormat : : R8Snorm , { 45 , 35 } , data . padding } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* Capture correct function name */
CORRADE_VERIFY ( true ) ;
char pixels [ ] {
' a ' , ' b ' , ' c ' ,
' d ' , ' e ' , ' f ' ,
} ;
Utility : : copy (
Containers : : StridedArrayView2D < char > { pixels , { 2 , 3 } } ,
cache . image ( ) . pixels < char > ( ) [ 0 ] . sliceSize ( { 30 , 15 } , { 2 , 3 } ) ) ;
cache . flushImage ( Range2Di : : fromSize ( { 15 , 30 } , { 3 , 2 } ) ) ;
CORRADE_VERIFY ( cache . called ) ;
}
void AbstractGlyphCacheTest : : flushImage2DPassthrough2D ( ) {
auto & & data = FlushImageData [ testCaseInstanceId ( ) ] ;
setTestCaseDescription ( data . name ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* Like flushImage2D() but with 2D doSetImage() */
struct Cache : AbstractGlyphCache {
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override { return { } ; }
void doSetImage ( const Vector2i & offset , const ImageView2D & image ) override {
called = true ;
CORRADE_COMPARE ( offset , ( Vector2i { 15 , 30 } ) - padding ( ) ) ;
CORRADE_COMPARE ( image . size ( ) , ( Vector2i { 3 , 2 } ) + 2 * padding ( ) ) ;
if ( padding ( ) . isZero ( ) ) {
char pixels [ ] {
' a ' , ' b ' , ' c ' , 0 ,
' d ' , ' e ' , ' f ' , 0 ,
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) ,
( ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels } ) ,
DebugTools : : CompareImage ) ;
} else {
char pixels [ ] {
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , ' a ' , ' b ' , ' c ' , 0 , 0 , 0 ,
0 , 0 , ' d ' , ' e ' , ' f ' , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) ,
( ImageView2D { PixelFormat : : R8Snorm , { 7 , 8 } , pixels } ) ,
DebugTools : : CompareImage ) ;
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
bool called = false ;
} cache { NoCreate } ;
if ( data . differentProcessedFormatSize )
cache = Cache { PixelFormat : : R8Snorm , { 45 , 35 } , PixelFormat : : RG32F , { 12 , 34 } } ;
else
cache = Cache { PixelFormat : : R8Snorm , { 45 , 35 } , data . padding } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* Capture correct function name */
CORRADE_VERIFY ( true ) ;
char pixels [ ] {
' a ' , ' b ' , ' c ' ,
' d ' , ' e ' , ' f ' ,
} ;
Utility : : copy (
Containers : : StridedArrayView2D < char > { pixels , { 2 , 3 } } ,
cache . image ( ) . pixels < char > ( ) [ 0 ] . sliceSize ( { 30 , 15 } , { 2 , 3 } ) ) ;
cache . flushImage ( Range2Di : : fromSize ( { 15 , 30 } , { 3 , 2 } ) ) ;
CORRADE_VERIFY ( cache . called ) ;
}
void AbstractGlyphCacheTest : : flushImageNotImplemented ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override { return { } ; }
/* The 2D variant shouldn't be called on an array cache */
void doSetImage ( const Vector2i & , const ImageView2D & ) override {
CORRADE_FAIL ( " This should not be called " ) ;
}
} cache { PixelFormat : : R32F , { 1024 , 512 , 8 } } ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . flushImage ( 0 , { } ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::image(): not implemented by derived class \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
void AbstractGlyphCacheTest : : flushImagePassthrough2DNotImplemented ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override { return { } ; }
/* Should try the 3D variant, and from that one call into the 2D where
it ' d assert */
} cache { PixelFormat : : R32F , { 1024 , 512 } } ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . flushImage ( 0 , { } ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::image(): not implemented by derived class \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
void AbstractGlyphCacheTest : : flushImageOutOfRange ( ) {
auto & & data = FlushImageData [ testCaseInstanceId ( ) ] ;
setTestCaseDescription ( data . name ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { NoCreate } ;
/* Neither the padding nor the processed size should not have any effect on
the check */
if ( data . differentProcessedFormatSize )
cache = DummyGlyphCache { PixelFormat : : R32F , { 1024 , 512 , 8 } , PixelFormat : : R8Snorm , { 1536 , 768 } } ;
else
cache = DummyGlyphCache { PixelFormat : : R32F , { 1024 , 512 , 8 } , data . padding } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
/* Negative min X, Y, layer */
cache . flushImage ( { { - 1 , 30 , 4 } , { 45 , 35 , 6 } } ) ;
cache . flushImage ( { { 15 , - 1 , 4 } , { 45 , 35 , 6 } } ) ;
cache . flushImage ( { { 15 , 30 , - 1 } , { 45 , 35 , 6 } } ) ;
cache . flushImage ( - 1 , { { 15 , 30 } , { 45 , 35 } } ) ;
/* Too large max X, Y, layer */
cache . flushImage ( { { 15 , 30 , 4 } , { 1025 , 35 , 6 } } ) ;
cache . flushImage ( { { 15 , 30 , 4 } , { 45 , 513 , 6 } } ) ;
cache . flushImage ( { { 15 , 30 , 4 } , { 45 , 35 , 9 } } ) ;
cache . flushImage ( 8 , { { 15 , 30 } , { 45 , 35 } } ) ;
/* Negative range size on X, Y, layer */
cache . flushImage ( { { 45 , 30 , 4 } , { 15 , 35 , 6 } } ) ;
cache . flushImage ( { { 15 , 35 , 4 } , { 45 , 30 , 6 } } ) ;
cache . flushImage ( { { 15 , 30 , 6 } , { 45 , 35 , 4 } } ) ;
CORRADE_COMPARE_AS ( out ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
" Text::AbstractGlyphCache::flushImage(): {{-1, 30, 4}, {45, 35, 6}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::flushImage(): {{15, -1, 4}, {45, 35, 6}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::flushImage(): {{15, 30, -1}, {45, 35, 6}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::flushImage(): {{15, 30, -1}, {45, 35, 0}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::flushImage(): {{15, 30, 4}, {1025, 35, 6}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::flushImage(): {{15, 30, 4}, {45, 513, 6}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::flushImage(): {{15, 30, 4}, {45, 35, 9}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::flushImage(): {{15, 30, 8}, {45, 35, 9}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::flushImage(): {{45, 30, 4}, {15, 35, 6}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::flushImage(): {{15, 35, 4}, {45, 30, 6}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::flushImage(): {{15, 30, 6}, {45, 35, 4}} out of range for size {1024, 512, 8} \n " ,
TestSuite : : Compare : : String ) ;
}
void AbstractGlyphCacheTest : : flushImage2DNot2D ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } } ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . flushImage ( Range2Di { } ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::flushImage(): use the 3D or layer overload for an array glyph cache \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
# ifdef MAGNUM_BUILD_DEPRECATED
void AbstractGlyphCacheTest : : setImage ( ) {
auto & & data = FlushImageData [ testCaseInstanceId ( ) ] ;
setTestCaseDescription ( data . name ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_PUSH
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override { return { } ; }
void doSetImage ( const Vector2i & offset , const ImageView2D & image ) override {
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
called = true ;
CORRADE_COMPARE ( offset , ( Vector2i { 15 , 30 } ) - padding ( ) ) ;
CORRADE_COMPARE ( image . size ( ) , ( Vector2i { 3 , 2 } ) + 2 * padding ( ) ) ;
if ( padding ( ) . isZero ( ) ) {
char pixels [ ] {
' a ' , ' b ' , ' c ' , 0 ,
' d ' , ' e ' , ' f ' , 0 ,
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) ,
( ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels } ) ,
DebugTools : : CompareImage ) ;
} else {
char pixels [ ] {
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , ' a ' , ' b ' , ' c ' , 0 , 0 , 0 ,
0 , 0 , ' d ' , ' e ' , ' f ' , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 ,
0 , 0 , 0 , 0 , 0 , 0 , 0 , 0
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) ,
( ImageView2D { PixelFormat : : R8Snorm , { 7 , 8 } , pixels } ) ,
DebugTools : : CompareImage ) ;
}
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
bool called = false ;
/* Deliberately using the deprecated PixelFormat-less constructor to verify
that passing a R8Unorm image " just works " */
} cache { { 45 , 35 } , data . padding } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_POP
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
/* Capture correct function name */
CORRADE_VERIFY ( true ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
char pixels [ ] {
0 , 0 , 0 , 0 , 0 ,
0 , ' a ' , ' b ' , ' c ' , 0 ,
0 , ' d ' , ' e ' , ' f ' , 0 ,
0 , 0 , 0 , 0 , 0
} ;
/* Testing with a custom PixelStorage to verify the right area gets copied
to the internal image */
CORRADE_IGNORE_DEPRECATED_PUSH
cache . setImage ( { 15 , 30 } , ImageView2D {
PixelStorage { }
. setAlignment ( 1 )
. setRowLength ( 5 )
. setSkip ( { 1 , 1 , 0 } ) ,
PixelFormat : : R8Unorm ,
{ 3 , 2 } ,
pixels } ) ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_VERIFY ( cache . called ) ;
}
void AbstractGlyphCacheTest : : setImageOutOfRange ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
DummyGlyphCache cache { PixelFormat : : R8Unorm , { 100 , 200 } } ;
/* This is fine */
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_PUSH
cache . setImage ( { 80 , 175 } , ImageView2D { PixelFormat : : R8Unorm , { 20 , 25 } } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_POP
Containers : : String out ;
Error redirectError { & out } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_PUSH
cache . setImage ( { 81 , 175 } , ImageView2D { PixelFormat : : R8Unorm , { 20 , 25 } } ) ;
cache . setImage ( { 80 , 176 } , ImageView2D { PixelFormat : : R8Unorm , { 20 , 25 } } ) ;
cache . setImage ( { - 1 , 175 } , ImageView2D { PixelFormat : : R8Unorm , { 20 , 25 } } ) ;
cache . setImage ( { 80 , - 1 } , ImageView2D { PixelFormat : : R8Unorm , { 20 , 25 } } ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE_AS ( out ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
" Text::AbstractGlyphCache::setImage(): Range({81, 175}, {101, 200}) out of range for glyph cache of size Vector(100, 200) \n "
" Text::AbstractGlyphCache::setImage(): Range({80, 176}, {100, 201}) out of range for glyph cache of size Vector(100, 200) \n "
" Text::AbstractGlyphCache::setImage(): Range({-1, 175}, {19, 200}) out of range for glyph cache of size Vector(100, 200) \n "
" Text::AbstractGlyphCache::setImage(): Range({80, -1}, {100, 24}) out of range for glyph cache of size Vector(100, 200) \n " ,
TestSuite : : Compare : : String ) ;
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void AbstractGlyphCacheTest : : setImageInvalidFormat ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 } } ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
CORRADE_IGNORE_DEPRECATED_PUSH
cache . setImage ( { 15 , 30 } , ImageView2D { PixelFormat : : R8Unorm , { 45 , 35 } } ) ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::setImage(): expected PixelFormat::R32F but got PixelFormat::R8Unorm \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
void AbstractGlyphCacheTest : : setImageNot2D ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } } ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
CORRADE_IGNORE_DEPRECATED_PUSH
cache . setImage ( { } , ImageView2D { PixelFormat : : R32F , { } , nullptr } ) ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::setImage(): can't be used on an array glyph cache \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
# endif
void AbstractGlyphCacheTest : : processedImage ( ) {
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
GlyphCacheFeatures doFeatures ( ) const override {
return GlyphCacheFeature : : ProcessedImageDownload ;
}
void doSetImage ( const Vector2i & , const ImageView2D & ) override { }
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Image3D doProcessedImage ( ) override { return Image3D { PixelFormat : : RG8Unorm , { 2 , 3 , 1 } , Containers : : Array < char > { NoInit , 6 * 2 } } ; }
/* Using a different format or size for the source image shouldn't cause
any problem */
} cache { PixelFormat : : RG8Srgb , { 200 , 300 } } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Image3D image = cache . processedImage ( ) ;
CORRADE_COMPARE ( image . format ( ) , PixelFormat : : RG8Unorm ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_COMPARE ( image . size ( ) , ( Vector3i { 2 , 3 , 1 } ) ) ;
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void AbstractGlyphCacheTest : : processedImageNotSupported ( ) {
auto & & data = ProcessedImageNotSupportedData [ testCaseInstanceId ( ) ] ;
setTestCaseDescription ( data . name ) ;
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
struct Cache : AbstractGlyphCache {
explicit Cache ( const Vector2i & size , GlyphCacheFeatures features ) : AbstractGlyphCache { PixelFormat : : R8Unorm , size } , _features { features } { }
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
GlyphCacheFeatures doFeatures ( ) const override { return _features ; }
void doSetImage ( const Vector2i & , const ImageView2D & ) override { }
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
GlyphCacheFeatures _features ;
} cache { { 200 , 300 } , data . features } ;
Containers : : String out ;
Error redirectError { & out } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
cache . processedImage ( ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::processedImage(): feature not supported \n " ) ;
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void AbstractGlyphCacheTest : : processedImageNotImplemented ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
GlyphCacheFeatures doFeatures ( ) const override {
return GlyphCacheFeature : : ProcessedImageDownload ;
}
void doSetImage ( const Vector2i & , const ImageView2D & ) override { }
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
} cache { PixelFormat : : R8Unorm , { 200 , 300 } } ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . processedImage ( ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::processedImage(): feature advertised but not implemented \n " ) ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
}
void AbstractGlyphCacheTest : : setProcessedImage ( ) {
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override {
return GlyphCacheFeature : : ImageProcessing ;
}
void doSetProcessedImage ( const Vector3i & offset , const ImageView3D & image ) override {
called = true ;
CORRADE_COMPARE ( offset , ( Vector3i { 15 , 30 , 3 } ) ) ;
CORRADE_COMPARE ( image . size ( ) , ( Vector3i { 3 , 2 , 2 } ) ) ;
char pixels0 [ ] {
' a ' , ' b ' , ' c ' , 0 ,
' d ' , ' e ' , ' f ' , 0 ,
} ;
char pixels1 [ ] {
' 0 ' , ' 1 ' , ' 2 ' , 0 ,
' 3 ' , ' 4 ' , ' 5 ' , 0 ,
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 0 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels0 } ) ,
DebugTools : : CompareImage ) ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 1 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels1 } ) ,
DebugTools : : CompareImage ) ;
}
bool called = false ;
} cache { PixelFormat : : RGB16Unorm , { 4 , 3 , 5 } , PixelFormat : : R8Snorm , { 45 , 35 } } ;
/* Capture correct function name */
CORRADE_VERIFY ( true ) ;
char pixels [ ] {
' a ' , ' b ' , ' c ' , 0 ,
' d ' , ' e ' , ' f ' , 0 ,
' 0 ' , ' 1 ' , ' 2 ' , 0 ,
' 3 ' , ' 4 ' , ' 5 ' , 0 ,
} ;
cache . setProcessedImage ( { 15 , 30 , 3 } , ImageView3D { PixelFormat : : R8Snorm , { 3 , 2 , 2 } , pixels } ) ;
CORRADE_VERIFY ( cache . called ) ;
}
void AbstractGlyphCacheTest : : setProcessedImage2D ( ) {
/* Like setProcessedImage() but reduced to two dimensions */
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override {
return GlyphCacheFeature : : ImageProcessing ;
}
void doSetProcessedImage ( const Vector3i & offset , const ImageView3D & image ) override {
called = true ;
CORRADE_COMPARE ( offset , ( Vector3i { 15 , 30 , 0 } ) ) ;
CORRADE_COMPARE ( image . size ( ) , ( Vector3i { 3 , 2 , 1 } ) ) ;
char pixels0 [ ] {
' a ' , ' b ' , ' c ' , 0 ,
' d ' , ' e ' , ' f ' , 0 ,
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) [ 0 ] ,
( ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels0 } ) ,
DebugTools : : CompareImage ) ;
}
bool called = false ;
} cache { PixelFormat : : RGB16Unorm , { 4 , 3 } , PixelFormat : : R8Snorm , { 45 , 35 } } ;
/* Capture correct function name */
CORRADE_VERIFY ( true ) ;
char pixels [ ] {
' a ' , ' b ' , ' c ' , 0 ,
' d ' , ' e ' , ' f ' , 0 ,
} ;
cache . setProcessedImage ( { 15 , 30 } , ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels } ) ;
CORRADE_VERIFY ( cache . called ) ;
}
void AbstractGlyphCacheTest : : setProcessedImage2DPassthrough2D ( ) {
/* Like setProcessedImage2D() but with 2D doSetProcessedImage() */
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override {
return GlyphCacheFeature : : ImageProcessing ;
}
void doSetProcessedImage ( const Vector2i & offset , const ImageView2D & image ) override {
called = true ;
CORRADE_COMPARE ( offset , ( Vector2i { 15 , 30 } ) ) ;
CORRADE_COMPARE ( image . size ( ) , ( Vector2i { 3 , 2 } ) ) ;
char pixels0 [ ] {
' a ' , ' b ' , ' c ' , 0 ,
' d ' , ' e ' , ' f ' , 0 ,
} ;
CORRADE_COMPARE_AS ( image . pixels < Byte > ( ) ,
( ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels0 } ) ,
DebugTools : : CompareImage ) ;
}
bool called = false ;
} cache { PixelFormat : : RGB16Unorm , { 4 , 3 } , PixelFormat : : R8Snorm , { 45 , 35 } } ;
/* Capture correct function name */
CORRADE_VERIFY ( true ) ;
char pixels [ ] {
' a ' , ' b ' , ' c ' , 0 ,
' d ' , ' e ' , ' f ' , 0 ,
} ;
cache . setProcessedImage ( { 15 , 30 } , ImageView2D { PixelFormat : : R8Snorm , { 3 , 2 } , pixels } ) ;
CORRADE_VERIFY ( cache . called ) ;
}
void AbstractGlyphCacheTest : : setProcessedImageNotImplemented ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override {
return GlyphCacheFeature : : ImageProcessing ;
}
/* The 2D variant shouldn't be called on an array cache */
void doSetProcessedImage ( const Vector2i & , const ImageView2D & ) override {
CORRADE_FAIL ( " This should not be called " ) ;
}
} cache { PixelFormat : : R32F , { 1024 , 512 , 8 } } ;
Containers : : String out ;
Error redirectError { & out } ;
cache . setProcessedImage ( { } , ImageView3D { PixelFormat : : R32F , { } } ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::setProcessedImage(): feature advertised but not implemented \n " ) ;
}
void AbstractGlyphCacheTest : : setProcessedImagePassthrough2DNotImplemented ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override {
return GlyphCacheFeature : : ImageProcessing ;
}
/* The 2D variant shouldn't be called on an array cache */
void doSetProcessedImage ( const Vector2i & , const ImageView2D & ) override {
CORRADE_FAIL ( " This should not be called " ) ;
}
} cache { PixelFormat : : R32F , { 1024 , 512 , 8 } } ;
Containers : : String out ;
Error redirectError { & out } ;
cache . setProcessedImage ( { } , ImageView3D { PixelFormat : : R32F , { } } ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::setProcessedImage(): feature advertised but not implemented \n " ) ;
}
void AbstractGlyphCacheTest : : setProcessedImageOutOfRange ( ) {
auto & & data = SetProcessedImageOutOfRangeData [ testCaseInstanceId ( ) ] ;
setTestCaseDescription ( data . name ) ;
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
/* Like flushImage(), but for setProcessedImage() */
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override {
return GlyphCacheFeature : : ImageProcessing ;
}
/* The source size and padding should not have any effect on the check */
} cache { PixelFormat : : RGBA32F , { 1536 , 768 , 8 } , PixelFormat : : R8Snorm , { 1024 , 512 } , data . padding } ;
/* Large enough data to fit in all cases below, 4-byte aligned rows */
Containers : : Array < char > image { NoInit , 1012 * 5 * 2 } ;
Containers : : String out ;
Error redirectError { & out } ;
/* Negative min X, Y, layer */
cache . setProcessedImage ( { - 1 , 30 , 4 } ,
ImageView3D { PixelFormat : : R8Snorm , { 46 , 5 , 2 } , image } ) ;
cache . setProcessedImage ( { 15 , - 1 , 4 } ,
ImageView3D { PixelFormat : : R8Snorm , { 30 , 36 , 2 } , image } ) ;
cache . setProcessedImage ( { 15 , 30 , - 1 } ,
ImageView3D { PixelFormat : : R8Snorm , { 30 , 5 , 7 } , image } ) ;
/* Too large max X, Y, layer */
cache . setProcessedImage ( { 15 , 30 , 4 } ,
ImageView3D { PixelFormat : : R8Snorm , { 1010 , 5 , 2 } } ) ;
cache . setProcessedImage ( { 15 , 30 , 4 } ,
ImageView3D { PixelFormat : : R8Snorm , { 30 , 483 , 2 } } ) ;
cache . setProcessedImage ( { 15 , 30 , 4 } ,
ImageView3D { PixelFormat : : R8Snorm , { 30 , 5 , 5 } } ) ;
CORRADE_COMPARE_AS ( out ,
" Text::AbstractGlyphCache::setProcessedImage(): {{-1, 30, 4}, {45, 35, 6}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::setProcessedImage(): {{15, -1, 4}, {45, 35, 6}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::setProcessedImage(): {{15, 30, -1}, {45, 35, 6}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::setProcessedImage(): {{15, 30, 4}, {1025, 35, 6}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::setProcessedImage(): {{15, 30, 4}, {45, 513, 6}} out of range for size {1024, 512, 8} \n "
" Text::AbstractGlyphCache::setProcessedImage(): {{15, 30, 4}, {45, 35, 9}} out of range for size {1024, 512, 8} \n " ,
TestSuite : : Compare : : String ) ;
}
void AbstractGlyphCacheTest : : setProcessedImageInvalidFormat ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override {
return GlyphCacheFeature : : ImageProcessing ;
}
/* The source format should not have any effect on the check */
} cache { PixelFormat : : RGBA32F , { 1024 , 512 , 8 } , PixelFormat : : R8Snorm , { 3 , 2 } } ;
Containers : : String out ;
Error redirectError { & out } ;
cache . setProcessedImage ( { } , ImageView3D { PixelFormat : : R8Unorm , { 3 , 2 , 1 } , " abcdefgh " } ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::setProcessedImage(): expected PixelFormat::R8Snorm but got PixelFormat::R8Unorm \n " ) ;
}
void AbstractGlyphCacheTest : : setProcessedImage2DNot2D ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
struct : AbstractGlyphCache {
using AbstractGlyphCache : : AbstractGlyphCache ;
GlyphCacheFeatures doFeatures ( ) const override {
return GlyphCacheFeature : : ImageProcessing ;
}
} cache { PixelFormat : : R8Unorm , { 3 , 2 , 8 } } ;
Containers : : String out ;
Error redirectError { & out } ;
cache . setProcessedImage ( { } , ImageView2D { PixelFormat : : R8Unorm , { 3 , 2 } , " abcdefgh " } ) ;
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::setProcessedImage(): use the 3D overload for an array glyph cache \n " ) ;
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
void AbstractGlyphCacheTest : : access ( ) {
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
/* Padding tested well enough in addGlyph(), resetting it back to 0 here */
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , { } } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
cache . setInvalidGlyph ( { 5 , 7 } , 2 , { { 5 , 10 } , { 15 , 30 } } ) ;
UnsignedInt font9 = cache . addFont ( 9 ) ;
UnsignedInt font3 = cache . addFont ( 3 ) ;
UnsignedInt font9Glyph6 = cache . addGlyph ( font9 , 6 , { 3 , 4 } , 0 , { { 15 , 30 } , { 45 , 35 } } ) ;
UnsignedInt font3Glyph1 = cache . addGlyph ( font3 , 1 , { 5 , 6 } , 1 , { { 10 , 15 } , { 25 , 30 } } ) ;
UnsignedInt font9Glyph3 = cache . addGlyph ( font9 , 3 , { 6 , 9 } , 2 , { { 10 , 5 } , { 25 , 10 } } ) ;
CORRADE_COMPARE ( font9Glyph6 , 1 ) ;
CORRADE_COMPARE ( font3Glyph1 , 2 ) ;
CORRADE_COMPARE ( font9Glyph3 , 3 ) ;
/* Mapping to the global glyph ID */
CORRADE_COMPARE ( cache . glyphId ( font9 , 6 ) , font9Glyph6 ) ;
CORRADE_COMPARE ( cache . glyphId ( font3 , 1 ) , font3Glyph1 ) ;
CORRADE_COMPARE ( cache . glyphId ( font9 , 3 ) , font9Glyph3 ) ;
/* Both overloads should return the same */
CORRADE_COMPARE ( cache . glyph ( font9 , 6 ) , Containers : : triple (
Vector2i { 3 , 4 } ,
0 ,
Range2Di { { 15 , 30 } , { 45 , 35 } } ) ) ;
CORRADE_COMPARE ( cache . glyph ( font9Glyph6 ) , Containers : : triple (
Vector2i { 3 , 4 } ,
0 ,
Range2Di { { 15 , 30 } , { 45 , 35 } } ) ) ;
CORRADE_COMPARE ( cache . glyph ( font3 , 1 ) , Containers : : triple (
Vector2i { 5 , 6 } ,
1 ,
Range2Di { { 10 , 15 } , { 25 , 30 } } ) ) ;
CORRADE_COMPARE ( cache . glyph ( font3Glyph1 ) , Containers : : triple (
Vector2i { 5 , 6 } ,
1 ,
Range2Di { { 10 , 15 } , { 25 , 30 } } ) ) ;
CORRADE_COMPARE ( cache . glyph ( font9 , 3 ) , Containers : : triple (
Vector2i { 6 , 9 } ,
2 ,
Range2Di { { 10 , 5 } , { 25 , 10 } } ) ) ;
CORRADE_COMPARE ( cache . glyph ( font9Glyph3 ) , Containers : : triple (
Vector2i { 6 , 9 } ,
2 ,
Range2Di { { 10 , 5 } , { 25 , 10 } } ) ) ;
/* Mapping to the invalid glyph ID if it hasn't been added yet */
CORRADE_COMPARE ( cache . glyphId ( font9 , 5 ) , 0 ) ;
CORRADE_COMPARE ( cache . glyphId ( font3 , 2 ) , 0 ) ;
/* Querying glyphs that haven't been added yet gives back the invalid
glyph properties */
CORRADE_COMPARE ( cache . glyph ( font9 , 5 ) , Containers : : triple (
Vector2i { 5 , 7 } ,
2 ,
Range2Di { { 5 , 10 } , { 15 , 30 } } ) ) ;
CORRADE_COMPARE ( cache . glyph ( font3 , 2 ) , Containers : : triple (
Vector2i { 5 , 7 } ,
2 ,
Range2Di { { 5 , 10 } , { 15 , 30 } } ) ) ;
CORRADE_COMPARE ( cache . glyph ( 0 ) , Containers : : triple (
Vector2i { 5 , 7 } ,
2 ,
Range2Di { { 5 , 10 } , { 15 , 30 } } ) ) ;
}
void AbstractGlyphCacheTest : : accessBatch ( ) {
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
/* Padding tested well enough in addGlyph(), resetting it back to 0 here */
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , { } } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
cache . setInvalidGlyph ( { 5 , 7 } , 2 , { { 5 , 10 } , { 15 , 30 } } ) ;
UnsignedInt font9 = cache . addFont ( 9 ) ;
UnsignedInt font3 = cache . addFont ( 3 ) ;
UnsignedInt font9Glyph6 = cache . addGlyph ( font9 , 6 , { 3 , 4 } , 0 , { { 15 , 30 } , { 45 , 35 } } ) ;
UnsignedInt font3Glyph1 = cache . addGlyph ( font3 , 1 , { 5 , 6 } , 1 , { { 10 , 15 } , { 25 , 30 } } ) ;
UnsignedInt font9Glyph3 = cache . addGlyph ( font9 , 3 , { 6 , 9 } , 2 , { { 10 , 5 } , { 25 , 10 } } ) ;
CORRADE_COMPARE ( font9Glyph6 , 1 ) ;
CORRADE_COMPARE ( font3Glyph1 , 2 ) ;
CORRADE_COMPARE ( font9Glyph3 , 3 ) ;
/* Direct data access */
CORRADE_COMPARE_AS ( cache . glyphOffsets ( ) , Containers : : arrayView < Vector2i > ( {
{ 5 , 7 } ,
{ 3 , 4 } ,
{ 5 , 6 } ,
{ 6 , 9 } ,
} ) , TestSuite : : Compare : : Container ) ;
CORRADE_COMPARE_AS ( cache . glyphLayers ( ) , Containers : : arrayView ( {
2 ,
0 ,
1 ,
2
} ) , TestSuite : : Compare : : Container ) ;
CORRADE_COMPARE_AS ( cache . glyphRectangles ( ) , Containers : : arrayView < Range2Di > ( {
{ { 5 , 10 } , { 15 , 30 } } ,
{ { 15 , 30 } , { 45 , 35 } } ,
{ { 10 , 15 } , { 25 , 30 } } ,
{ { 10 , 5 } , { 25 , 10 } }
} ) , TestSuite : : Compare : : Container ) ;
/* Querying glyph IDs in a batch way. Invalid IDs are set to 0. */
UnsignedInt glyphIds9 [ 5 ] ;
cache . glyphIdsInto ( font9 , { 5 , 6 , 3 , 6 , 1 } , glyphIds9 ) ;
CORRADE_COMPARE_AS ( Containers : : arrayView ( glyphIds9 ) , Containers : : arrayView ( {
0u , 1u , 3u , 1u , 0u
} ) , TestSuite : : Compare : : Container ) ;
UnsignedInt glyphIds3 [ 3 ] ;
cache . glyphIdsInto ( font3 , { 2 , 0 , 1 } , glyphIds3 ) ;
CORRADE_COMPARE_AS ( Containers : : arrayView ( glyphIds3 ) , Containers : : arrayView ( {
0u , 0u , 2u
} ) , TestSuite : : Compare : : Container ) ;
}
void AbstractGlyphCacheTest : : accessInvalid ( ) {
CORRADE_SKIP_IF_NO_DEBUG_ASSERT ( ) ;
/* Silly test name, but these all test debug asserts while
accessBatchInvalid ( ) tests non - debug asserts */
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
/* Default padding of 1 makes it impossible to add a glyph at zero offset
as it ' s out of range . Don ' t want to bother with that here so resetting
it to 0. */
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , { } } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
cache . addFont ( 9 ) ;
UnsignedInt fontId = cache . addFont ( 3 ) ;
cache . addGlyph ( 0 , 1 , { } , 2 , { } ) ;
cache . addGlyph ( fontId , 1 , { } , 2 , { } ) ;
cache . addGlyph ( fontId , 2 , { } , 2 , { } ) ;
UnsignedInt fontGlyphIds [ ] {
0 , 0 , cache . fontGlyphCount ( fontId ) , 0
} ;
UnsignedInt glyphIds [ 4 ] ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . glyph ( cache . fontCount ( ) , 0 ) ;
cache . glyphId ( cache . fontCount ( ) , 0 ) ;
cache . glyph ( fontId , cache . fontGlyphCount ( fontId ) ) ;
cache . glyphId ( fontId , cache . fontGlyphCount ( fontId ) ) ;
cache . glyphIdsInto ( fontId , fontGlyphIds , glyphIds ) ;
cache . glyph ( cache . glyphCount ( ) ) ;
CORRADE_COMPARE_AS ( out ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
" Text::AbstractGlyphCache::glyph(): index 2 out of range for 2 fonts \n "
" Text::AbstractGlyphCache::glyphId(): index 2 out of range for 2 fonts \n "
" Text::AbstractGlyphCache::glyph(): index 3 out of range for 3 glyphs in font 1 \n "
" Text::AbstractGlyphCache::glyphId(): index 3 out of range for 3 glyphs in font 1 \n "
" Text::AbstractGlyphCache::glyphIdsInto(): glyph 2 index 3 out of range for 3 glyphs in font 1 \n "
" Text::AbstractGlyphCache::glyph(): index 4 out of range for 4 glyphs \n " ,
TestSuite : : Compare : : String ) ;
}
void AbstractGlyphCacheTest : : accessBatchInvalid ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
Text: make glyph caches pad by one pixel by default to avoid artifacts.
Took me quite a while to realize what was going on, but in retrospect
it's obvious -- the rasterizer just *rounds* the sub-pixel-positioned
glyph quads to nearest pixels. Which then can cause the neighboring
glyph data to leak to it (because the texture is then sampled not
directly on the edge pixel, but slightly outside of it), or it can also
cut away the edge, when it gets rounded in the other direction.
This was a problem with the original -- laughably inefficient -- atlas
packer as well, but because that packer had excessive padding around
everything, only the second edge-cutting artifact manifested, and that
one is rather subtle unless you know what to look for.
This means the packing is now slightly worse than before and sizes that
previously worked may no longer fit anymore. But since the new atlas
packer is relatively new (well, from September, time sure flies
different here), and the improvement compared to the original packer is
still quite massive, I don't think this is a problem.
2 years ago
/* Default padding of 1 makes it impossible to add a glyph at zero offset
as it ' s out of range . Don ' t want to bother with that here so resetting
it to 0. */
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } , { } } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
cache . addFont ( 9 ) ;
UnsignedInt fontId = cache . addFont ( 3 ) ;
cache . addGlyph ( fontId , 1 , { } , 2 , { } ) ;
cache . addGlyph ( fontId , 2 , { } , 2 , { } ) ;
UnsignedInt fontGlyphIds [ 4 ] ;
UnsignedInt glyphIds [ 4 ] ;
UnsignedInt glyphIdsInvalid [ 3 ] ;
Containers : : String out ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
Error redirectError { & out } ;
cache . glyphIdsInto ( cache . fontCount ( ) , fontGlyphIds , glyphIds ) ;
cache . glyphIdsInto ( fontId , fontGlyphIds , glyphIdsInvalid ) ;
CORRADE_COMPARE ( out ,
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
" Text::AbstractGlyphCache::glyphIdsInto(): index 2 out of range for 2 fonts \n "
" Text::AbstractGlyphCache::glyphIdsInto(): expected fontGlyphIds and glyphIds views to have the same size but got 4 and 3 \n " ) ;
}
# ifdef MAGNUM_BUILD_DEPRECATED
void AbstractGlyphCacheTest : : accessDeprecated ( ) {
DummyGlyphCache cache { PixelFormat : : R8Unorm , { 100 , 200 } , { 2 , 3 } } ;
cache . setInvalidGlyph ( { 3 , 5 } , { { 10 , 10 } , { 23 , 45 } } ) ;
UnsignedInt fontId = cache . addFont ( 25 ) ;
cache . addGlyph ( fontId , 15 , { 3 , 4 } , { { 15 , 30 } , { 45 , 35 } } ) ;
CORRADE_IGNORE_DEPRECATED_PUSH
CORRADE_COMPARE ( cache [ 15 ] , std : : make_pair (
Vector2i { 1 , 1 } ,
Range2Di { { 13 , 27 } , { 47 , 38 } }
) ) ;
CORRADE_IGNORE_DEPRECATED_POP
/* ID 0 gets the invalid glyph */
CORRADE_IGNORE_DEPRECATED_PUSH
CORRADE_COMPARE ( cache [ 0 ] , std : : make_pair (
Vector2i { 1 , 2 } ,
Range2Di { { 8 , 7 } , { 25 , 48 } }
) ) ;
CORRADE_IGNORE_DEPRECATED_POP
/* Glyph IDs out of bounds get the invalid glyph too */
CORRADE_IGNORE_DEPRECATED_PUSH
CORRADE_COMPARE ( cache [ 45 ] , std : : make_pair (
Vector2i { 1 , 2 } ,
Range2Di { { 8 , 7 } , { 25 , 48 } }
) ) ;
CORRADE_IGNORE_DEPRECATED_POP
}
void AbstractGlyphCacheTest : : accessDeprecatedNot2D ( ) {
CORRADE_SKIP_IF_NO_ASSERT ( ) ;
DummyGlyphCache cache { PixelFormat : : R32F , { 1024 , 512 , 3 } } ;
Containers : : String out ;
Error redirectError { & out } ;
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
CORRADE_IGNORE_DEPRECATED_PUSH
cache [ 5 ] ;
CORRADE_IGNORE_DEPRECATED_POP
CORRADE_COMPARE ( out , " Text::AbstractGlyphCache::operator[](): can't be used on an array glyph cache \n " ) ;
}
Text: rework AbstractGlyphCache for better flexibility and efficiency.
The class now supports incremental filling, multiple fonts, texture
arrays, removes all reliance on STL containers and is finally properly
documented.
To avoid complete breakage of every use, as much as possible was kept as
deprecated APIs -- in particular the reserve() with the nasty
std::vectors, the insert() that assumes a 2D cache and a single font
and textureSize() that returns a 2D vector. Those behave the same as
before, but will assert if the cache is an array or contains more than
one font.
On the other hand, begin() / end() access with std::unordered_map iterators
(ew!) was removed as the internals simply aren't a hashmap anymore. The
image() that returned an Image2D is now used to fill the glyph cache
instead of querying its potentially processed contents, and returns a
MutableImageView3D. I considered keeping it and adding sourceImage()
instead, but such naming turned out to be too inconsistent. For querying
processed image data (such as with the distance field cache) there's a
new processedImage() query, guarded by new GlyphCacheFeature bits -- if
both ImageProcessing and ProcessedImageDownload is set, it can be used
to retrieve the processed image (so, similar as ImageDownload was
before), and if neither is set, the cache contents are queryable
directly through image(), without needing any special support from
the GPU API.
Existing code is updated only in the minimal way possible to ensure that
no serious breakage was introduced by reimplementing the deprecated APIs
on top of the new backend. Porting away from deprecated APIs will be
done in next commits. The GlyphCache and DistanceFieldGlyphCache have
their public API kept intact for now, as a similar rework will be needed
for them as well.
Additionally, the MagnumFont and MagnumFontConverter plugins aren't
compiling yet as they require substantial changes to deal with the new
glyph cache features. That is not the case with other plugins in the
magnum-plugins repository tho, for those the backwards compatibility
"just works". On the other hand, since layout of the AbstractGlyphChange
changed, I'm bumping the AbstractFont plugin interface version to
force-trigger a rebuild of dependent projects. Because I ran a stale
magnum-player binary, it worked without crashing or GL errors but just
didn't show ANY text whatsoever due to ABI differences, and I wasted
some precious minutes before realizing that a simple rebuild would fix
it.
3 years ago
# endif
} } } }
CORRADE_TEST_MAIN ( Magnum : : Text : : Test : : AbstractGlyphCacheTest )
