ic
/
magnum
mirror of https://github.com/mosra/magnum.git
6
0
Fork 0
Code Issues Projects Releases Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
6801 Commits
21 Branches
19 Tags
70 MiB
Tree: b05c887375
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from 'b05c887375'
${ noResults }
magnum/src/Magnum/GL/Test/MeshGLTest.cpp

3130 lines
113 KiB
Raw Normal View History Unescape

Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
/*
This file is part of Magnum.
Updated copyright year.
7 years ago
Copyright © 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019
Happy new year too.
12 years ago
Vladimír Vondruš <mosra@centrum.cz>
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
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.
*/
GL: assert that addVertexBuffer() is called with a valid Buffer.
8 years ago
#include <sstream>
Adapted to Corrade changes. Mostly missing includes now.
7 years ago
#include <Corrade/Utility/DebugStl.h>
GL: assert that addVertexBuffer() is called with a valid Buffer.
8 years ago
Making includes absolute, part 5: absolute includes everywhere. The only places where they aren't absolute are: - when header is included from corresponding source file - when including headers which are not part of final installation (e.g. test-specific configuration, headers from Implementation/)
12 years ago
#include "Magnum/Image.h"
#include "Magnum/Mesh.h"
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
#include "Magnum/GL/AbstractShaderProgram.h"
#include "Magnum/GL/Buffer.h"
#include "Magnum/GL/Context.h"
#include "Magnum/GL/Extensions.h"
#include "Magnum/GL/Framebuffer.h"
#include "Magnum/GL/Mesh.h"
#include "Magnum/GL/MeshView.h"
#include "Magnum/GL/OpenGLTester.h"
#include "Magnum/GL/PixelFormat.h"
#include "Magnum/GL/Renderbuffer.h"
#include "Magnum/GL/RenderbufferFormat.h"
#include "Magnum/GL/Shader.h"
Moved Color to Math namespace. Apart from different include (<Magnum/Math/Color.h> instead of <Magnum/Color.h>) there shouldn't be any visible change to the user. The BasicColor3 and BasicColor4 classes are now Math::Color3 and Math::Color4. The Color3, Color4, Color3ub and Color4ub typedefs in Magnum namespace stayed the same. BasicColor3 and BasicColor4 is now an alias to Math::Color3 and Math::Color4, is marked as deprecated and will be removed in future release. The same goes for the <Magnum/Color.h> include, which now just includes the <Magnum/Math/Color.h> header.
11 years ago
#include "Magnum/Math/Color.h"
Test: we have halves now, so why not test them.
9 years ago
#include "Magnum/Math/Half.h"
Making includes absolute, part 5: absolute includes everywhere. The only places where they aren't absolute are: - when header is included from corresponding source file - when including headers which are not part of final installation (e.g. test-specific configuration, headers from Implementation/)
12 years ago
#include "Magnum/Math/Matrix.h"
#include "Magnum/Math/Vector4.h"
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
namespace Magnum { namespace GL { namespace Test { namespace {
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
/* Tests also the MeshView class. */
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Made Test/AbstractOpenGLTester.h a public and documented API. There will be numerous additions to this one so it made sense to make it a static library instead of a header-only library. That also allows CMake users to just link to Magnum::OpenGLTester instead of going through the pain of a huge branching in order to find a correct windowless application just to run their tests. It could have been done even without the static library using a INTERFACE target, but that wouldn't work on CMake < 3.0 (which, unfortunately, quite a few people are still stuck with). Unfortunately it's already heavily used elsewhere so I had to go through the pain of deprecating the old implementation. The old implementation was header-only so it can't be just typedef'd to the new one as there would be linker failures. So the old header is just kept as it was, with only the macros reduced.
9 years ago
struct MeshGLTest: OpenGLTester {
Making test declarations more compact. Yet again I experienced how insanely large this thing got over the years.
11 years ago
explicit MeshGLTest();
void construct();
void constructMove();
Ability to wrap and release existing OpenGL objects. Allows for better interaction with third-party libraries. I should probably write a new documentation page about OpenGL wrapping.
11 years ago
void wrap();
Making test declarations more compact. Yet again I experienced how insanely large this thing got over the years.
11 years ago
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
template<class T> void primitive();
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#ifndef MAGNUM_TARGET_WEBGL
Making test declarations more compact. Yet again I experienced how insanely large this thing got over the years.
11 years ago
void label();
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#endif
Making test declarations more compact. Yet again I experienced how insanely large this thing got over the years.
11 years ago
#ifndef MAGNUM_TARGET_GLES2
void addVertexBufferUnsignedInt();
void addVertexBufferInt();
#endif
void addVertexBufferFloat();
#ifndef MAGNUM_TARGET_GLES
void addVertexBufferDouble();
#endif
#ifndef MAGNUM_TARGET_GLES2
void addVertexBufferVectorNui();
void addVertexBufferVectorNi();
#endif
void addVertexBufferVectorN();
#ifndef MAGNUM_TARGET_GLES
void addVertexBufferVectorNd();
#endif
void addVertexBufferMatrixNxN();
#ifndef MAGNUM_TARGET_GLES
void addVertexBufferMatrixNxNd();
#endif
#ifndef MAGNUM_TARGET_GLES2
void addVertexBufferMatrixMxN();
#endif
#ifndef MAGNUM_TARGET_GLES
void addVertexBufferMatrixMxNd();
#endif
#ifndef MAGNUM_TARGET_GLES2
void addVertexBufferUnsignedIntWithUnsignedShort();
void addVertexBufferUnsignedIntWithShort();
void addVertexBufferIntWithUnsignedShort();
void addVertexBufferIntWithShort();
#endif
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#if !(defined(MAGNUM_TARGET_WEBGL) && defined(MAGNUM_TARGET_GLES2))
Making test declarations more compact. Yet again I experienced how insanely large this thing got over the years.
11 years ago
/* Other Float types omitted (covered by addVertexBufferNormalized()) */
void addVertexBufferFloatWithHalfFloat();
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#endif
Making test declarations more compact. Yet again I experienced how insanely large this thing got over the years.
11 years ago
#ifndef MAGNUM_TARGET_GLES
void addVertexBufferFloatWithDouble();
void addVertexBufferVector3WithUnsignedInt10f11f11fRev();
#endif
#ifndef MAGNUM_TARGET_GLES2
void addVertexBufferVector4WithUnsignedInt2101010Rev();
void addVertexBufferVector4WithInt2101010Rev();
#endif
void addVertexBufferLessVectorComponents();
void addVertexBufferNormalized();
#ifndef MAGNUM_TARGET_GLES
void addVertexBufferBGRA();
#endif
void addVertexBufferMultiple();
void addVertexBufferMultipleGaps();
GL: assert that addVertexBuffer() is called with a valid Buffer.
8 years ago
void addVertexBufferMovedOutInstance();
GL: make it possible for a Mesh to own the vertex/index buffers. I wanted to do this since 2015.
8 years ago
void addVertexBufferTransferOwnwership();
void addVertexBufferInstancedTransferOwnwership();
void addVertexBufferDynamicTransferOwnwership();
void addVertexBufferInstancedDynamicTransferOwnwership();
GL: assert that addVertexBuffer() is called with a valid Buffer.
8 years ago
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
template<class T> void setIndexBuffer();
template<class T> void setIndexBufferRange();
Making test declarations more compact. Yet again I experienced how insanely large this thing got over the years.
11 years ago
void setIndexBufferUnsignedInt();
GL: assert that addVertexBuffer() is called with a valid Buffer.
8 years ago
void setIndexBufferMovedOutInstance();
GL: make it possible for a Mesh to own the vertex/index buffers. I wanted to do this since 2015.
8 years ago
template<class T> void setIndexBufferTransferOwnership();
template<class T> void setIndexBufferRangeTransferOwnership();
GL: assert that addVertexBuffer() is called with a valid Buffer.
8 years ago
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
void unbindVAOWhenSettingIndexBufferData();
GL: test yet another interaction of VAOs and index buffer binding. Fails (well, why else would I be adding the test, right?).
7 years ago
void unbindIndexBufferWhenBindingVao();
GL: further harden VAO and index buffer binding state interaction tests. This one crashes. Turns out the 169031fb7bf30e2c98c81cbf865aa5a4662aa716 contained a random temporary test state instead of the real solution (and so the comment didn't even match the code, it should have been resetting that to 0). That also made some tests fail with DSA disabled, but none of the tests were actual Mesh tests, just accidentally hitting the problematic code path. I took the opportunity to look at this more closely and investigate *why* this failed -- turns out, in setIndexBuffer(), I was resetting the state tracker to a VAO state that was about to be set in the very next step, and then, when doing that next step the state tracker "optimized away" the state change because it thought it was already done (even though it wasn't). The new test in MeshGLTest covers this particular case.
7 years ago
void resetIndexBufferBindingWhenBindingVao();
Improve state tracker resetting functionality.
8 years ago
void unbindVAOBeforeEnteringExternalSection();
GL: make it possible to bind a scratch VAO for external GL code. Used in those sad and unfortunate cases where we need to call shitty external GL code that's not VAO-aware.
8 years ago
void bindScratchVaoWhenEnteringExternalSection();
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
Making test declarations more compact. Yet again I experienced how insanely large this thing got over the years.
11 years ago
#ifndef MAGNUM_TARGET_GLES
void setBaseVertex();
#endif
void setInstanceCount();
void setInstanceCountIndexed();
#ifndef MAGNUM_TARGET_GLES
void setInstanceCountBaseInstance();
void setInstanceCountBaseInstanceIndexed();
void setInstanceCountBaseVertex();
void setInstanceCountBaseVertexBaseInstance();
#endif
void addVertexBufferInstancedFloat();
#ifndef MAGNUM_TARGET_GLES2
void addVertexBufferInstancedInteger();
#endif
#ifndef MAGNUM_TARGET_GLES
void addVertexBufferInstancedDouble();
#endif
GL: test that vertex attrib divisor is properly reset when not using VAOs. Fails, because it's not. Fix in next commit.
8 years ago
void resetDivisorAfterInstancedDraw();
Making test declarations more compact. Yet again I experienced how insanely large this thing got over the years.
11 years ago
void multiDraw();
void multiDrawIndexed();
#ifndef MAGNUM_TARGET_GLES
void multiDrawBaseVertex();
#endif
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
};
MeshGLTest::MeshGLTest() {
Testing Mesh move constructor and assignment.
13 years ago
addTests({&MeshGLTest::construct,
&MeshGLTest::constructMove,
Ability to wrap and release existing OpenGL objects. Allows for better interaction with third-party libraries. I should probably write a new documentation page about OpenGL wrapping.
11 years ago
&MeshGLTest::wrap,
Testing Mesh move constructor and assignment.
13 years ago
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
&MeshGLTest::primitive<GL::MeshPrimitive>,
&MeshGLTest::primitive<Magnum::MeshPrimitive>,
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#ifndef MAGNUM_TARGET_WEBGL
&MeshGLTest::label
#endif
});
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
/* First instance is always using Attribute, second DynamicAttribute */
addInstancedTests({
#ifndef MAGNUM_TARGET_GLES2
&MeshGLTest::addVertexBufferUnsignedInt,
&MeshGLTest::addVertexBufferInt,
#endif
&MeshGLTest::addVertexBufferFloat,
#ifndef MAGNUM_TARGET_GLES
&MeshGLTest::addVertexBufferDouble,
#endif
#ifndef MAGNUM_TARGET_GLES2
&MeshGLTest::addVertexBufferVectorNui,
&MeshGLTest::addVertexBufferVectorNi,
#endif
&MeshGLTest::addVertexBufferVectorN,
#ifndef MAGNUM_TARGET_GLES
&MeshGLTest::addVertexBufferVectorNd,
#endif
&MeshGLTest::addVertexBufferMatrixNxN,
#ifndef MAGNUM_TARGET_GLES
&MeshGLTest::addVertexBufferMatrixNxNd,
#endif
#ifndef MAGNUM_TARGET_GLES2
&MeshGLTest::addVertexBufferMatrixMxN,
#endif
#ifndef MAGNUM_TARGET_GLES
&MeshGLTest::addVertexBufferMatrixMxNd,
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
#ifndef MAGNUM_TARGET_GLES2
&MeshGLTest::addVertexBufferUnsignedIntWithUnsignedShort,
&MeshGLTest::addVertexBufferUnsignedIntWithShort,
&MeshGLTest::addVertexBufferIntWithUnsignedShort,
&MeshGLTest::addVertexBufferIntWithShort,
#endif
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#if !(defined(MAGNUM_TARGET_WEBGL) && defined(MAGNUM_TARGET_GLES2))
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
&MeshGLTest::addVertexBufferFloatWithHalfFloat,
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#endif
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
#ifndef MAGNUM_TARGET_GLES
&MeshGLTest::addVertexBufferFloatWithDouble,
&MeshGLTest::addVertexBufferVector3WithUnsignedInt10f11f11fRev,
#endif
#ifndef MAGNUM_TARGET_GLES2
&MeshGLTest::addVertexBufferVector4WithUnsignedInt2101010Rev,
&MeshGLTest::addVertexBufferVector4WithInt2101010Rev,
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
&MeshGLTest::addVertexBufferLessVectorComponents,
&MeshGLTest::addVertexBufferNormalized,
#ifndef MAGNUM_TARGET_GLES
&MeshGLTest::addVertexBufferBGRA
#endif
}, 2);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
addTests({&MeshGLTest::addVertexBufferMultiple,
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
&MeshGLTest::addVertexBufferMultipleGaps,
GL: assert that addVertexBuffer() is called with a valid Buffer.
8 years ago
&MeshGLTest::addVertexBufferMovedOutInstance,
GL: make it possible for a Mesh to own the vertex/index buffers. I wanted to do this since 2015.
8 years ago
&MeshGLTest::addVertexBufferTransferOwnwership,
&MeshGLTest::addVertexBufferInstancedTransferOwnwership,
&MeshGLTest::addVertexBufferDynamicTransferOwnwership,
&MeshGLTest::addVertexBufferInstancedDynamicTransferOwnwership,
GL: assert that addVertexBuffer() is called with a valid Buffer.
8 years ago
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
&MeshGLTest::setIndexBuffer<GL::MeshIndexType>,
&MeshGLTest::setIndexBuffer<Magnum::MeshIndexType>,
&MeshGLTest::setIndexBufferRange<GL::MeshIndexType>,
&MeshGLTest::setIndexBufferRange<Magnum::MeshIndexType>,
Added test for mesh base vertex.
12 years ago
&MeshGLTest::setIndexBufferUnsignedInt,
GL: assert that addVertexBuffer() is called with a valid Buffer.
8 years ago
&MeshGLTest::setIndexBufferMovedOutInstance,
GL: make it possible for a Mesh to own the vertex/index buffers. I wanted to do this since 2015.
8 years ago
&MeshGLTest::setIndexBufferTransferOwnership<GL::MeshIndexType>,
&MeshGLTest::setIndexBufferTransferOwnership<Magnum::MeshIndexType>,
&MeshGLTest::setIndexBufferRangeTransferOwnership<GL::MeshIndexType>,
&MeshGLTest::setIndexBufferRangeTransferOwnership<Magnum::MeshIndexType>,
GL: assert that addVertexBuffer() is called with a valid Buffer.
8 years ago
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
&MeshGLTest::unbindVAOWhenSettingIndexBufferData,
GL: test yet another interaction of VAOs and index buffer binding. Fails (well, why else would I be adding the test, right?).
7 years ago
&MeshGLTest::unbindIndexBufferWhenBindingVao,
GL: further harden VAO and index buffer binding state interaction tests. This one crashes. Turns out the 169031fb7bf30e2c98c81cbf865aa5a4662aa716 contained a random temporary test state instead of the real solution (and so the comment didn't even match the code, it should have been resetting that to 0). That also made some tests fail with DSA disabled, but none of the tests were actual Mesh tests, just accidentally hitting the problematic code path. I took the opportunity to look at this more closely and investigate *why* this failed -- turns out, in setIndexBuffer(), I was resetting the state tracker to a VAO state that was about to be set in the very next step, and then, when doing that next step the state tracker "optimized away" the state change because it thought it was already done (even though it wasn't). The new test in MeshGLTest covers this particular case.
7 years ago
&MeshGLTest::resetIndexBufferBindingWhenBindingVao,
Improve state tracker resetting functionality.
8 years ago
&MeshGLTest::unbindVAOBeforeEnteringExternalSection,
GL: make it possible to bind a scratch VAO for external GL code. Used in those sad and unfortunate cases where we need to call shitty external GL code that's not VAO-aware.
8 years ago
&MeshGLTest::bindScratchVaoWhenEnteringExternalSection,
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
Added test for mesh base vertex.
12 years ago
#ifndef MAGNUM_TARGET_GLES
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
&MeshGLTest::setBaseVertex,
#endif
&MeshGLTest::setInstanceCount,
&MeshGLTest::setInstanceCountIndexed,
#ifndef MAGNUM_TARGET_GLES
&MeshGLTest::setInstanceCountBaseInstance,
&MeshGLTest::setInstanceCountBaseInstanceIndexed,
&MeshGLTest::setInstanceCountBaseVertex,
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
&MeshGLTest::setInstanceCountBaseVertexBaseInstance,
#endif
&MeshGLTest::addVertexBufferInstancedFloat,
#ifndef MAGNUM_TARGET_GLES2
&MeshGLTest::addVertexBufferInstancedInteger,
#endif
#ifndef MAGNUM_TARGET_GLES
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
&MeshGLTest::addVertexBufferInstancedDouble,
#endif
GL: test that vertex attrib divisor is properly reset when not using VAOs. Fails, because it's not. Fix in next commit.
8 years ago
&MeshGLTest::resetDivisorAfterInstancedDraw,
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
&MeshGLTest::multiDraw,
&MeshGLTest::multiDrawIndexed,
#ifndef MAGNUM_TARGET_GLES
&MeshGLTest::multiDrawBaseVertex
Added test for mesh base vertex.
12 years ago
#endif
});
Testing Mesh move constructor and assignment.
13 years ago
}
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
using namespace Math::Literals;
Testing Mesh move constructor and assignment.
13 years ago
void MeshGLTest::construct() {
{
const Mesh mesh;
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Mesh: set VAO id to 0 when the extension is not available. Also expect this case in test.
13 years ago
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(Context::current().isExtensionSupported<Extensions::ARB::vertex_array_object>())
Mesh: set VAO id to 0 when the extension is not available. Also expect this case in test.
13 years ago
#elif defined(MAGNUM_TARGET_GLES2)
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(Context::current().isExtensionSupported<Extensions::OES::vertex_array_object>())
Mesh: set VAO id to 0 when the extension is not available. Also expect this case in test.
13 years ago
#endif
{
CORRADE_VERIFY(mesh.id() > 0);
}
Testing Mesh move constructor and assignment.
13 years ago
}
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Testing Mesh move constructor and assignment.
13 years ago
}
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
struct FloatShader: AbstractShaderProgram {
explicit FloatShader(const std::string& type, const std::string& conversion);
};
GL: get rid of std::vector in Mesh header. This is actually a preparation to make buffer-owning meshes a possibility (where I would need an union of vectors otherwise), nevertheless it removes the dependency on a vector.
8 years ago
Testing Mesh move constructor and assignment.
13 years ago
void MeshGLTest::constructMove() {
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
const Float data = Math::unpack<Float, UnsignedByte>(
#ifndef MAGNUM_TARGET_GLES2
96
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
85
#endif
);
GL: get rid of std::vector in Mesh header. This is actually a preparation to make buffer-owning meshes a possibility (where I would need an union of vectors otherwise), nevertheless it removes the dependency on a vector.
8 years ago
Buffer buffer1, buffer2;
buffer1.setData({&data, 1}, BufferUsage::StaticDraw);
GL: fix various OOB / unaligned access issues in Mesh test.
8 years ago
buffer2.setData({nullptr, 8}, BufferUsage::StaticDraw);
GL: get rid of std::vector in Mesh header. This is actually a preparation to make buffer-owning meshes a possibility (where I would need an union of vectors otherwise), nevertheless it removes the dependency on a vector.
8 years ago
Testing Mesh move constructor and assignment.
13 years ago
Mesh a;
GL: get rid of std::vector in Mesh header. This is actually a preparation to make buffer-owning meshes a possibility (where I would need an union of vectors otherwise), nevertheless it removes the dependency on a vector.
8 years ago
a.addVertexBuffer(buffer1, 0, Attribute<0, Float>{});
Testing Mesh move constructor and assignment.
13 years ago
const Int id = a.id();
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Mesh: set VAO id to 0 when the extension is not available. Also expect this case in test.
13 years ago
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(Context::current().isExtensionSupported<Extensions::ARB::vertex_array_object>())
Mesh: set VAO id to 0 when the extension is not available. Also expect this case in test.
13 years ago
#elif defined(MAGNUM_TARGET_GLES2)
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(Context::current().isExtensionSupported<Extensions::OES::vertex_array_object>())
Mesh: set VAO id to 0 when the extension is not available. Also expect this case in test.
13 years ago
#endif
{
CORRADE_VERIFY(id > 0);
}
Testing Mesh move constructor and assignment.
13 years ago
GL: fix undefined behavior and potential leaks in Mesh. The change done in 680144f1c511997ee0530ea88bbb95aa20e058c8 was not properly handling these cases: * Mesh(NoCreateT) and wrap() were not constructing the internal vector, which blew up when move-assigning another instance. * ~Mesh() was not destructing the internal vector if the VAO ID was zero or non-owning wrap() was used. Strangely enough none of these were causing *any* problems for me on Linux (even ASan was totally quiet and due to the unfortunate combination of bugs even when I assigned totally random data to the storage vector). This however blew up on MSVC, assuming there the implementation is more checked. Because it's possible to construct Mesh with no GL context available, the move construction and destruction needs to avoid accessing Context unless really necessary (it would be also unclear which type of vector should be constructed if we have no context). Extended the tests to handle hopefully all the cases.
8 years ago
/* Move construct */
Testing Mesh move constructor and assignment.
13 years ago
Mesh b(std::move(a));
CORRADE_COMPARE(a.id(), 0);
CORRADE_COMPARE(b.id(), id);
GL: fix undefined behavior and potential leaks in Mesh. The change done in 680144f1c511997ee0530ea88bbb95aa20e058c8 was not properly handling these cases: * Mesh(NoCreateT) and wrap() were not constructing the internal vector, which blew up when move-assigning another instance. * ~Mesh() was not destructing the internal vector if the VAO ID was zero or non-owning wrap() was used. Strangely enough none of these were causing *any* problems for me on Linux (even ASan was totally quiet and due to the unfortunate combination of bugs even when I assigned totally random data to the storage vector). This however blew up on MSVC, assuming there the implementation is more checked. Because it's possible to construct Mesh with no GL context available, the move construction and destruction needs to avoid accessing Context unless really necessary (it would be also unclear which type of vector should be constructed if we have no context). Extended the tests to handle hopefully all the cases.
8 years ago
/* Move assign */
Testing Mesh move constructor and assignment.
13 years ago
Mesh c;
GL: fix various OOB / unaligned access issues in Mesh test.
8 years ago
c.addVertexBuffer(buffer2, 4, Attribute<1, Float>{});
Testing Mesh move constructor and assignment.
13 years ago
const Int cId = c.id();
c = std::move(b);
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Mesh: set VAO id to 0 when the extension is not available. Also expect this case in test.
13 years ago
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(Context::current().isExtensionSupported<Extensions::ARB::vertex_array_object>())
Mesh: set VAO id to 0 when the extension is not available. Also expect this case in test.
13 years ago
#elif defined(MAGNUM_TARGET_GLES2)
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(Context::current().isExtensionSupported<Extensions::OES::vertex_array_object>())
Mesh: set VAO id to 0 when the extension is not available. Also expect this case in test.
13 years ago
#endif
{
CORRADE_VERIFY(cId > 0);
}
Testing Mesh move constructor and assignment.
13 years ago
CORRADE_COMPARE(b.id(), cId);
CORRADE_COMPARE(c.id(), id);
GL: get rid of std::vector in Mesh header. This is actually a preparation to make buffer-owning meshes a possibility (where I would need an union of vectors otherwise), nevertheless it removes the dependency on a vector.
8 years ago
GL: fix undefined behavior and potential leaks in Mesh. The change done in 680144f1c511997ee0530ea88bbb95aa20e058c8 was not properly handling these cases: * Mesh(NoCreateT) and wrap() were not constructing the internal vector, which blew up when move-assigning another instance. * ~Mesh() was not destructing the internal vector if the VAO ID was zero or non-owning wrap() was used. Strangely enough none of these were causing *any* problems for me on Linux (even ASan was totally quiet and due to the unfortunate combination of bugs even when I assigned totally random data to the storage vector). This however blew up on MSVC, assuming there the implementation is more checked. Because it's possible to construct Mesh with no GL context available, the move construction and destruction needs to avoid accessing Context unless really necessary (it would be also unclear which type of vector should be constructed if we have no context). Extended the tests to handle hopefully all the cases.
8 years ago
/* Move assign to a NoCreate instance */
Mesh d{NoCreate};
d = std::move(c);
CORRADE_COMPARE(c.id(), 0);
CORRADE_COMPARE(d.id(), id);
/* Destroy */
b = Mesh{NoCreate};
GL: get rid of std::vector in Mesh header. This is actually a preparation to make buffer-owning meshes a possibility (where I would need an union of vectors otherwise), nevertheless it removes the dependency on a vector.
8 years ago
/* Test that drawing still works properly */
{
MAGNUM_VERIFY_NO_GL_ERROR();
Renderbuffer renderbuffer;
renderbuffer.setStorage(
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Vector2i(1));
Framebuffer framebuffer{{{}, Vector2i(1)}};
framebuffer.attachRenderbuffer(Framebuffer::ColorAttachment(0), renderbuffer)
.bind();
FloatShader shader{"float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"};
GL: fix undefined behavior and potential leaks in Mesh. The change done in 680144f1c511997ee0530ea88bbb95aa20e058c8 was not properly handling these cases: * Mesh(NoCreateT) and wrap() were not constructing the internal vector, which blew up when move-assigning another instance. * ~Mesh() was not destructing the internal vector if the VAO ID was zero or non-owning wrap() was used. Strangely enough none of these were causing *any* problems for me on Linux (even ASan was totally quiet and due to the unfortunate combination of bugs even when I assigned totally random data to the storage vector). This however blew up on MSVC, assuming there the implementation is more checked. Because it's possible to construct Mesh with no GL context available, the move construction and destruction needs to avoid accessing Context unless really necessary (it would be also unclear which type of vector should be constructed if we have no context). Extended the tests to handle hopefully all the cases.
8 years ago
d.setPrimitive(MeshPrimitive::Points)
GL: get rid of std::vector in Mesh header. This is actually a preparation to make buffer-owning meshes a possibility (where I would need an union of vectors otherwise), nevertheless it removes the dependency on a vector.
8 years ago
.setCount(1)
.draw(shader);
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: get rid of std::vector in Mesh header. This is actually a preparation to make buffer-owning meshes a possibility (where I would need an union of vectors otherwise), nevertheless it removes the dependency on a vector.
8 years ago
CORRADE_COMPARE(framebuffer.read({{}, Vector2i{1}}, {PixelFormat::RGBA, PixelType::UnsignedByte}).data<UnsignedByte>()[0], 96);
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else /* RGBA4, so less precision */
CORRADE_COMPARE(framebuffer.read({{}, Vector2i{1}}, {PixelFormat::RGBA, PixelType::UnsignedByte}).data<UnsignedByte>()[0], 85);
#endif
GL: get rid of std::vector in Mesh header. This is actually a preparation to make buffer-owning meshes a possibility (where I would need an union of vectors otherwise), nevertheless it removes the dependency on a vector.
8 years ago
}
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
}
Ability to wrap and release existing OpenGL objects. Allows for better interaction with third-party libraries. I should probably write a new documentation page about OpenGL wrapping.
11 years ago
void MeshGLTest::wrap() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_array_object>())
CORRADE_SKIP(Extensions::ARB::vertex_array_object::string() + std::string{" is not supported."});
Ability to wrap and release existing OpenGL objects. Allows for better interaction with third-party libraries. I should probably write a new documentation page about OpenGL wrapping.
11 years ago
#elif defined(MAGNUM_TARGET_GLES2)
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::OES::vertex_array_object>())
CORRADE_SKIP(Extensions::OES::vertex_array_object::string() + std::string{" is not supported."});
Ability to wrap and release existing OpenGL objects. Allows for better interaction with third-party libraries. I should probably write a new documentation page about OpenGL wrapping.
11 years ago
#endif
GLuint id;
Fix Mesh test build on ES2. Yeah, thanks, Jenkins, for being so thorough, but... eh.
11 years ago
#ifndef MAGNUM_TARGET_GLES2
Ability to wrap and release existing OpenGL objects. Allows for better interaction with third-party libraries. I should probably write a new documentation page about OpenGL wrapping.
11 years ago
glGenVertexArrays(1, &id);
Fix Mesh test build on ES2. Yeah, thanks, Jenkins, for being so thorough, but... eh.
11 years ago
#else
glGenVertexArraysOES(1, &id);
#endif
Ability to wrap and release existing OpenGL objects. Allows for better interaction with third-party libraries. I should probably write a new documentation page about OpenGL wrapping.
11 years ago
/* Releasing won't delete anything */
{
auto mesh = Mesh::wrap(id, ObjectFlag::DeleteOnDestruction);
CORRADE_COMPARE(mesh.release(), id);
}
/* ...so we can wrap it again */
Mesh::wrap(id);
Fix Mesh test build on ES2. Yeah, thanks, Jenkins, for being so thorough, but... eh.
11 years ago
#ifndef MAGNUM_TARGET_GLES2
Ability to wrap and release existing OpenGL objects. Allows for better interaction with third-party libraries. I should probably write a new documentation page about OpenGL wrapping.
11 years ago
glDeleteVertexArrays(1, &id);
Fix Mesh test build on ES2. Yeah, thanks, Jenkins, for being so thorough, but... eh.
11 years ago
#else
glDeleteVertexArraysOES(1, &id);
#endif
Ability to wrap and release existing OpenGL objects. Allows for better interaction with third-party libraries. I should probably write a new documentation page about OpenGL wrapping.
11 years ago
}
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
template<class T> void MeshGLTest::primitive() {
setTestCaseName(std::is_same<T, MeshPrimitive>::value ?
GL: properly name the test case.
8 years ago
"primitive<GL::MeshPrimitive>" :
"primitive<Magnum::MeshPrimitive>");
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
{
Mesh mesh{T::LineLoop};
CORRADE_COMPARE(mesh.primitive(), MeshPrimitive::LineLoop);
} {
Mesh mesh;
mesh.setPrimitive(T::TriangleFan);
CORRADE_COMPARE(mesh.primitive(), MeshPrimitive::TriangleFan);
}
}
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#ifndef MAGNUM_TARGET_WEBGL
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
void MeshGLTest::label() {
GL: don't try to test mesh labels if VAOs are disabled.
8 years ago
#ifndef MAGNUM_TARGET_GLES
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_array_object>())
CORRADE_SKIP(Extensions::ARB::vertex_array_object::string() + std::string(" is not available."));
#elif defined(MAGNUM_TARGET_GLES2)
if(!Context::current().isExtensionSupported<Extensions::OES::vertex_array_object>())
CORRADE_SKIP(Extensions::OES::vertex_array_object::string() + std::string(" is not available."));
#endif
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
/* No-Op version is tested in AbstractObjectGLTest */
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::KHR::debug>() &&
!Context::current().isExtensionSupported<Extensions::EXT::debug_label>())
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
CORRADE_SKIP("Required extension is not available");
Mesh mesh;
Use object at least once before testing its label. Otherwise my NVidia emits GL_INVALID_VALUE with message "that <object> is not valid object", which isn't in the specs. I hate this "deferred everything" approach in GL. Is this even legal optimization?
12 years ago
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
CORRADE_COMPARE(mesh.label(), "");
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
mesh.setLabel("MyMesh");
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Ensure that the mesh object is properly created before using it.
12 years ago
CORRADE_COMPARE(mesh.label(), "MyMesh");
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
}
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#endif
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
struct IntegerShader: AbstractShaderProgram {
explicit IntegerShader(const std::string& type);
};
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
#ifndef MAGNUM_TARGET_GLES
struct DoubleShader: AbstractShaderProgram {
explicit DoubleShader(const std::string& type, const std::string& outputType, const std::string& conversion);
};
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
struct Checker {
Checker(AbstractShaderProgram&& shader, RenderbufferFormat format, Mesh& mesh);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
template<class T> T get(PixelFormat format, PixelType type);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
Renderbuffer renderbuffer;
Framebuffer framebuffer;
};
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Various documentation fixes and updates. {} in Matrix constructor caused Doxygen to forget all functions after it.
13 years ago
#ifndef DOXYGEN_GENERATING_OUTPUT
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
FloatShader::FloatShader(const std::string& type, const std::string& conversion) {
/* We need special version for ES3, because GLSL in ES2 doesn't support
rectangle matrices */
#ifndef MAGNUM_TARGET_GLES
Don't use GLSL < 130 in GL tests on OSX. OSX doesn't support anything below what's in GL 3.1 when using core contexts.
10 years ago
Shader vert(
#ifndef CORRADE_TARGET_APPLE
Version::GL210
#else
Version::GL310
#endif
, Shader::Type::Vertex);
Shader frag(
#ifndef CORRADE_TARGET_APPLE
Version::GL210
#else
Version::GL310
#endif
, Shader::Type::Fragment);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#elif defined(MAGNUM_TARGET_GLES2)
Shader vert(Version::GLES200, Shader::Type::Vertex);
Make use of threaded shader compilation in MeshGLTest.
12 years ago
Shader frag(Version::GLES200, Shader::Type::Fragment);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#else
Shader vert(Version::GLES300, Shader::Type::Vertex);
Make use of threaded shader compilation in MeshGLTest.
12 years ago
Shader frag(Version::GLES300, Shader::Type::Fragment);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
vert.addSource(
Define-out precision qualifiers in test shaders for GLSL 1.20. Causes the errors to abort on AMD cards.
11 years ago
"#if !defined(GL_ES) && __VERSION__ == 120\n"
"#define mediump\n"
"#endif\n"
GL: don't use old GLSL names in ES3 Mesh tests.
8 years ago
"#if (defined(GL_ES) && __VERSION__ < 300) || __VERSION__ == 120\n"
Don't use GLSL < 130 in GL tests on OSX. OSX doesn't support anything below what's in GL 3.1 when using core contexts.
10 years ago
"#define in attribute\n"
"#define out varying\n"
"#endif\n"
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
"in mediump " + type + " value;\n"
"out mediump " + type + " valueInterpolated;\n"
"void main() {\n"
" valueInterpolated = value;\n"
" gl_Position = vec4(0.0, 0.0, 0.0, 1.0);\n"
"}\n");
Don't use GLSL < 130 in GL tests on OSX. OSX doesn't support anything below what's in GL 3.1 when using core contexts.
10 years ago
frag.addSource(
"#if !defined(GL_ES) && __VERSION__ == 120\n"
"#define mediump\n"
"#endif\n"
GL: don't use old GLSL names in ES3 Mesh tests.
8 years ago
"#if (defined(GL_ES) && __VERSION__ < 300) || __VERSION__ == 120\n"
Don't use GLSL < 130 in GL tests on OSX. OSX doesn't support anything below what's in GL 3.1 when using core contexts.
10 years ago
"#define in varying\n"
"#define result gl_FragColor\n"
"#endif\n"
"in mediump " + type + " valueInterpolated;\n"
GL: don't use old GLSL names in ES3 Mesh tests.
8 years ago
"#if (defined(GL_ES) && __VERSION__ >= 300) || (!defined(GL_ES) && __VERSION__ >= 130)\n"
Don't use GLSL < 130 in GL tests on OSX. OSX doesn't support anything below what's in GL 3.1 when using core contexts.
10 years ago
"out mediump vec4 result;\n"
"#endif\n"
"void main() { result = " + conversion + "; }\n");
Make use of threaded shader compilation in MeshGLTest.
12 years ago
CORRADE_INTERNAL_ASSERT_OUTPUT(Shader::compile({vert, frag}));
Use threaded shader compilation etc. also in tests.
12 years ago
attachShaders({vert, frag});
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
bindAttributeLocation(0, "value");
CORRADE_INTERNAL_ASSERT_OUTPUT(link());
}
#ifndef MAGNUM_TARGET_GLES2
IntegerShader::IntegerShader(const std::string& type) {
#ifndef MAGNUM_TARGET_GLES
Don't use GLSL < 130 in GL tests on OSX. OSX doesn't support anything below what's in GL 3.1 when using core contexts.
10 years ago
Shader vert(
#ifndef CORRADE_TARGET_APPLE
Version::GL300
#else
Version::GL310
#endif
, Shader::Type::Vertex);
Shader frag(
#ifndef CORRADE_TARGET_APPLE
Version::GL300
#else
Version::GL310
#endif
, Shader::Type::Fragment);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#else
Shader vert(Version::GLES300, Shader::Type::Vertex);
Make use of threaded shader compilation in MeshGLTest.
12 years ago
Shader frag(Version::GLES300, Shader::Type::Fragment);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
Make use of threaded shader compilation in MeshGLTest.
12 years ago
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
vert.addSource("in mediump " + type + " value;\n"
"flat out mediump " + type + " valueInterpolated;\n"
"void main() {\n"
" valueInterpolated = value;\n"
" gl_Position = vec4(0.0, 0.0, 0.0, 1.0);\n"
"}\n");
frag.addSource("flat in mediump " + type + " valueInterpolated;\n"
"out mediump " + type + " result;\n"
"void main() { result = valueInterpolated; }\n");
Make use of threaded shader compilation in MeshGLTest.
12 years ago
CORRADE_INTERNAL_ASSERT_OUTPUT(Shader::compile({vert, frag}));
Use threaded shader compilation etc. also in tests.
12 years ago
attachShaders({vert, frag});
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
bindAttributeLocation(0, "value");
CORRADE_INTERNAL_ASSERT_OUTPUT(link());
}
#endif
#ifndef MAGNUM_TARGET_GLES
DoubleShader::DoubleShader(const std::string& type, const std::string& outputType, const std::string& conversion) {
Test: don't require GLSL 4.10 if not strictly needed. Fixes test failures on soft Mesa drivers that support only GLSL 3.30.
9 years ago
constexpr const Version version =
#ifndef CORRADE_TARGET_APPLE
Version::GL300;
#else
Version::GL400;
#endif
Shader vert{version, Shader::Type::Vertex};
Shader frag(version, Shader::Type::Fragment);
Make use of threaded shader compilation in MeshGLTest.
12 years ago
Test: don't require GLSL 4.10 if not strictly needed. Fixes test failures on soft Mesa drivers that support only GLSL 3.30.
9 years ago
vert.addSource(
"#extension GL_ARB_vertex_attrib_64bit: require\n"
"#extension GL_ARB_gpu_shader_fp64: require\n"
"in " + type + " value;\n"
"out " + outputType + " valueInterpolated;\n"
"void main() {\n"
" valueInterpolated = " + conversion + ";\n"
" gl_Position = vec4(0.0, 0.0, 0.0, 1.0);\n"
"}\n");
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
frag.addSource("in " + outputType + " valueInterpolated;\n"
"out " + outputType + " result;\n"
"void main() { result = valueInterpolated; }\n");
Make use of threaded shader compilation in MeshGLTest.
12 years ago
CORRADE_INTERNAL_ASSERT_OUTPUT(Shader::compile({vert, frag}));
Use threaded shader compilation etc. also in tests.
12 years ago
attachShaders({vert, frag});
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
bindAttributeLocation(0, "value");
CORRADE_INTERNAL_ASSERT_OUTPUT(link());
}
#endif
Checker::Checker(AbstractShaderProgram&& shader, RenderbufferFormat format, Mesh& mesh): framebuffer({{}, Vector2i(1)}) {
renderbuffer.setStorage(format, Vector2i(1));
framebuffer.attachRenderbuffer(Framebuffer::ColorAttachment(0), renderbuffer);
Simplified *Framebuffer::bind() functionality. This was a leftover from some not-well-thought-out design decision. The function is now used exclusively for binding for draw, as all framebuffer reading functions (blit(), read()) are doing the read binding internally. Moreover it required the user to be extra careful on ES2, because in many cases there are no separate binding points for reading and drawing. The function is now parameter-less and always bind the framebuffer for drawing. The logic for internal binding was also simplified and on ES2 there are separate implementations for single/separate binding points. For *Framebuffer::checkStatus() the documentation was updated to explain the meaning of the parameter on ES2 implementation. Also removed the need for FramebufferTarget::ReadDraw binding, as it was rather confusing. Old *Framebuffer::bind(FramebufferTarget) is now just an alias to the parameter-less function, ignoring the parameter. Along with FramebufferTarget::ReadDraw it is marked as deprecated and will be removed in some future release.
11 years ago
framebuffer.bind();
Don't differentiate between vertex and index count in Mesh. Only one value from these two was used in the end, wasting precious bytes. Also these two values were used to differentiate between indexed and non-indexed mesh (instead of relying on actual index buffer being bound), which was very confusing. This approach looks more clean. The MeshView class is not yet updated, as the change would expose some features that aren't possible in current implementation (base vertex specification). Merged Mesh::setVertexCount() and Mesh::setIndexCount() into one Mesh::setCount(), the two original functions are now guarded aliases to the new one, are marked as deprecated and will be removed in future release, similarly for the getters. In particular, if the mesh is indexed, setVertexCount() does nothing and vertexCount() returns 0. The setIndexCount() and indexCount() do and return the same regardless of whether the mesh is indexed or not.
12 years ago
mesh.setPrimitive(MeshPrimitive::Points)
.setCount(2);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
/* Skip first vertex so we test also offsets */
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
MeshView view{mesh};
view.setCount(1)
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
.setBaseVertex(mesh.baseVertex())
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
.setInstanceCount(mesh.instanceCount())
#ifndef MAGNUM_TARGET_GLES
.setBaseInstance(mesh.baseInstance())
#endif
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
;
if(view.mesh().isIndexed()) view.setIndexRange(1);
view.draw(shader);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
}
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
template<class T> T Checker::get(PixelFormat format, PixelType type) {
Making use of the shorter *Framebuffer::read(). I like that!
11 years ago
return framebuffer.read({{}, Vector2i{1}}, {format, type}).data<T>()[0];
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
}
Various documentation fixes and updates. {} in Matrix constructor caused Doxygen to forget all functions after it.
13 years ago
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#ifndef MAGNUM_TARGET_GLES2
void MeshGLTest::addVertexBufferUnsignedInt() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::EXT::gpu_shader4>())
CORRADE_SKIP(Extensions::EXT::gpu_shader4::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
constexpr UnsignedInt data[] = { 0, 157, 35681 };
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 4, Attribute<0, UnsignedInt>{});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4, 4, DynamicAttribute{
DynamicAttribute::Kind::Integral, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::UnsignedInt});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(IntegerShader("uint"), RenderbufferFormat::R32UI, mesh)
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
.get<UnsignedInt>(PixelFormat::RedInteger, PixelType::UnsignedInt);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, 35681);
}
void MeshGLTest::addVertexBufferInt() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::EXT::gpu_shader4>())
CORRADE_SKIP(Extensions::EXT::gpu_shader4::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
constexpr Int data[] = { 0, 457931, 27530 };
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 4, Attribute<0, Int>{});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4, 4, DynamicAttribute{
DynamicAttribute::Kind::Integral, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::Int});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(IntegerShader("int"), RenderbufferFormat::R32I, mesh)
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
.get<Int>(PixelFormat::RedInteger, PixelType::Int);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, 27530);
}
#endif
void MeshGLTest::addVertexBufferFloat() {
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
const Float data[] = { 0.0f, -0.7f, Math::unpack<Float, UnsignedByte>(
#ifndef MAGNUM_TARGET_GLES2
96
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
85
#endif
) };
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 4, Attribute<0, Float>{});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4, 4, DynamicAttribute{
DynamicAttribute::Kind::Generic, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::Float});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(FloatShader("float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"),
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<UnsignedByte>(PixelFormat::RGBA, PixelType::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, 96);
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 85);
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
}
#ifndef MAGNUM_TARGET_GLES
void MeshGLTest::addVertexBufferDouble() {
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_attrib_64bit>())
CORRADE_SKIP(Extensions::ARB::vertex_attrib_64bit::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
const Double data[] = { 0.0, -0.7, Math::unpack<Double, UnsignedShort>(45828) };
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 8, Attribute<0, Double>{});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 8, 8, DynamicAttribute{
DynamicAttribute::Kind::Long, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::Double});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Fix type conversion errors in Mesh test shader code. Reported by latest NVidia drivers (352.09), apparently worked until now. Huh.
11 years ago
const auto value = Checker(DoubleShader("double", "float", "float(value)"),
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
RenderbufferFormat::R16, mesh).get<UnsignedShort>(PixelFormat::Red, PixelType::UnsignedShort);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, 45828);
}
#endif
#ifndef MAGNUM_TARGET_GLES2
void MeshGLTest::addVertexBufferVectorNui() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::EXT::gpu_shader4>())
CORRADE_SKIP(Extensions::EXT::gpu_shader4::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
constexpr Vector3ui data[] = { {}, {37448, 547686, 156}, {27592, 157, 25} };
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 3*4, Attribute<0, Vector3ui>{});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 3*4, 3*4, DynamicAttribute{
DynamicAttribute::Kind::Integral, 0,
DynamicAttribute::Components::Three,
DynamicAttribute::DataType::UnsignedInt});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(IntegerShader("uvec3"), RenderbufferFormat::RGBA32UI, mesh)
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
.get<Vector3ui>(PixelFormat::RGBAInteger, PixelType::UnsignedInt);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, Vector3ui(27592, 157, 25));
}
void MeshGLTest::addVertexBufferVectorNi() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::EXT::gpu_shader4>())
CORRADE_SKIP(Extensions::EXT::gpu_shader4::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
constexpr Vector2i data[] = { {}, {-37448, 547686}, {27592, -157} };
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 2*4, Attribute<0, Vector2i>{});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 2*4, 2*4, DynamicAttribute{
DynamicAttribute::Kind::Integral, 0,
DynamicAttribute::Components::Two,
DynamicAttribute::DataType::Int});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(IntegerShader("ivec2"), RenderbufferFormat::RG32I, mesh)
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
.get<Vector2i>(PixelFormat::RGInteger, PixelType::Int);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, Vector2i(27592, -157));
}
#endif
void MeshGLTest::addVertexBufferVectorN() {
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
const Vector3 data[] = { {}, {0.0f, -0.9f, 1.0f},
#ifndef MAGNUM_TARGET_GLES2
0x60189c_rgbf
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
0x551199_rgbf
#endif
};
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 3*4, Attribute<0, Vector3>{});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 3*4, 3*4, DynamicAttribute{
DynamicAttribute::Kind::Generic, 0,
DynamicAttribute::Components::Three,
DynamicAttribute::DataType::Float});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(FloatShader("vec3", "vec4(valueInterpolated, 0.0)"),
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color3ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
CORRADE_COMPARE(value, 0x60189c_rgb);
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 0x551199_rgb);
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
}
#ifndef MAGNUM_TARGET_GLES
void MeshGLTest::addVertexBufferVectorNd() {
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_attrib_64bit>())
CORRADE_SKIP(Extensions::ARB::vertex_attrib_64bit::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const Vector4d data[] = {
{}, {0.0, -0.9, 1.0, 1.25},
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Math::unpack<Vector4d>(Math::Vector4<UnsignedShort>(315, 65201, 2576, 12))
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
};
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 4*8, Attribute<0, Vector4d>{});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4*8, 4*8, DynamicAttribute{
DynamicAttribute::Kind::Long, 0,
DynamicAttribute::Components::Four,
DynamicAttribute::DataType::Double});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(DoubleShader("dvec4", "vec4", "vec4(value)"),
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
RenderbufferFormat::RGBA16, mesh).get<Math::Vector4<UnsignedShort>>(PixelFormat::RGBA, PixelType::UnsignedShort);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, Math::Vector4<UnsignedShort>(315, 65201, 2576, 12));
}
#endif
void MeshGLTest::addVertexBufferMatrixNxN() {
const Matrix3x3 data[] = {
{},
Matrix3x3::fromDiagonal({0.0f, -0.9f, 1.0f}),
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
Matrix3x3::fromDiagonal(
#ifndef MAGNUM_TARGET_GLES2
0x60189c_rgbf
#else
/* Using only RGBA4, supply less precision. This has to be one on
the input because SwiftShader stores RGBA4 as RGBA8 internally,
thus preserving the full precision of the input. */
0x551199_rgbf
#endif
)
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
};
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 3*3*4, Attribute<0, Matrix3x3>{});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 3*3*4, 3*3*4, DynamicAttribute{
DynamicAttribute::Kind::Generic, 0,
DynamicAttribute::Components::Three,
DynamicAttribute::DataType::Float})
.addVertexBuffer(buffer, 3*3*4 + 3*4, 3*3*4, DynamicAttribute{
DynamicAttribute::Kind::Generic, 1,
DynamicAttribute::Components::Three,
DynamicAttribute::DataType::Float})
.addVertexBuffer(buffer, 3*3*4 + 6*4, 3*3*4, DynamicAttribute{
DynamicAttribute::Kind::Generic, 2,
DynamicAttribute::Components::Three,
DynamicAttribute::DataType::Float});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(FloatShader("mat3",
"vec4(valueInterpolated[0][0], valueInterpolated[1][1], valueInterpolated[2][2], 0.0)"),
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
mesh).get<Color3ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
CORRADE_COMPARE(value, 0x60189c_rgb);
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 0x551199_rgb);
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
}
#ifndef MAGNUM_TARGET_GLES
void MeshGLTest::addVertexBufferMatrixNxNd() {
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_attrib_64bit>())
CORRADE_SKIP(Extensions::ARB::vertex_attrib_64bit::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const Matrix3x3d data[] = {
{},
Matrix3x3d::fromDiagonal({0.0, -0.9, 1.0}),
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Matrix3x3d::fromDiagonal(Math::unpack<Vector3d>(Math::Vector3<UnsignedShort>(315, 65201, 2576)))
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
};
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 3*3*8, Attribute<0, Matrix3x3d>{});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 3*3*8, 3*3*8, DynamicAttribute{
DynamicAttribute::Kind::Long, 0,
DynamicAttribute::Components::Three,
DynamicAttribute::DataType::Double})
.addVertexBuffer(buffer, 3*3*8 + 3*8, 3*3*8, DynamicAttribute{
DynamicAttribute::Kind::Long, 1,
DynamicAttribute::Components::Three,
DynamicAttribute::DataType::Double})
Test: wow I am stupid!
8 years ago
.addVertexBuffer(buffer, 3*3*8 + 6*8, 3*3*8, DynamicAttribute{
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
DynamicAttribute::Kind::Long, 2,
DynamicAttribute::Components::Three,
DynamicAttribute::DataType::Double});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(DoubleShader("dmat3", "vec4",
"vec4(value[0][0], value[1][1], value[2][2], 0.0)"),
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
RenderbufferFormat::RGBA16, mesh).get<Math::Vector3<UnsignedShort>>(PixelFormat::RGB, PixelType::UnsignedShort);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Expect failure in MeshGLTest. After updating my NVidia drivers the double matrix tests are extracting only the first two values and I don't know why.
12 years ago
{
GL: add XFAIL to some well-known failures with Intel Windows drivers.
7 years ago
auto drivers = Context::DetectedDriver::Amd|Context::DetectedDriver::NVidia;
#ifdef CORRADE_TARGET_WINDOWS
drivers |= Context::DetectedDriver::IntelWindows;
#endif
CORRADE_EXPECT_FAIL_IF(Context::current().detectedDriver() & drivers, "Somehow only first two values are extracted on AMD, NVidia and Intel Windows drivers.");
Expect failure in MeshGLTest. After updating my NVidia drivers the double matrix tests are extracting only the first two values and I don't know why.
12 years ago
CORRADE_COMPARE(value, Math::Vector3<UnsignedShort>(315, 65201, 2576));
}
More vendor-specific XFAIL cases in Mesh GL test.
11 years ago
Revert "More vendor-specific XFAIL cases in Mesh GL test." This is behaving properly on AMD 15.300.1025.0. This reverts commit 8dd4dfcf2b1daf6d7f27868e6831c4dbc23ff89e.
11 years ago
/* This is wrong, but check if it's still the right wrong. Fails on AMD
15.201.1151 but seems to be fixed in 15.300.1025.0 */
More consistent naming in Context::DetectedDriver. DetectedDriver::AMD is now DetectedDriver::Amd, DetectedDriver::ProbablyAngle is now just DetectedDriver::Angle. The old names are still present, but deprecated and will be removed in the future.
9 years ago
if(Context::current().detectedDriver() & (Context::DetectedDriver::Amd|Context::DetectedDriver::NVidia))
Make the XFAIL in Mesh GL test more restrictive. Surprisingly OSX drivers are the only ones that are doing it right?!
10 years ago
CORRADE_COMPARE(value, Math::Vector3<UnsignedShort>(315, 65201, 0));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
}
#endif
#ifndef MAGNUM_TARGET_GLES2
void MeshGLTest::addVertexBufferMatrixMxN() {
const Matrix3x4 data[] = {
{},
Matrix3x4::fromDiagonal({0.0f, -0.9f, 1.0f}),
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Matrix3x4::fromDiagonal(Math::unpack<Vector3>(Color3ub(96, 24, 156)))
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
};
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 3*4*4, Attribute<0, Matrix3x4>{});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 3*4*4, 3*4*4, DynamicAttribute{
DynamicAttribute::Kind::Generic, 0,
DynamicAttribute::Components::Four,
DynamicAttribute::DataType::Float})
.addVertexBuffer(buffer, 3*4*4 + 4*4, 3*4*4, DynamicAttribute{
DynamicAttribute::Kind::Generic, 1,
DynamicAttribute::Components::Four,
DynamicAttribute::DataType::Float})
.addVertexBuffer(buffer, 3*4*4 + 8*4, 3*4*4, DynamicAttribute{
DynamicAttribute::Kind::Generic, 2,
DynamicAttribute::Components::Four,
DynamicAttribute::DataType::Float});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(FloatShader("mat3x4",
"vec4(valueInterpolated[0][0], valueInterpolated[1][1], valueInterpolated[2][2], 0.0)"),
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
RenderbufferFormat::RGBA8, mesh).get<Color3ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, Color3ub(96, 24, 156));
}
#endif
#ifndef MAGNUM_TARGET_GLES
void MeshGLTest::addVertexBufferMatrixMxNd() {
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_attrib_64bit>())
CORRADE_SKIP(Extensions::ARB::vertex_attrib_64bit::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const Matrix3x4d data[] = {
{},
GL: use proper literals in a test.
7 years ago
Matrix3x4d::fromDiagonal({0.0, -0.9, 1.0}),
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Matrix3x4d::fromDiagonal(Math::unpack<Vector3d>(Math::Vector3<UnsignedShort>(315, 65201, 2576)))
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
};
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 3*4*8, Attribute<0, Matrix3x4d>{});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 3*4*8, 3*4*8, DynamicAttribute{
DynamicAttribute::Kind::Long, 0,
DynamicAttribute::Components::Four,
DynamicAttribute::DataType::Double})
.addVertexBuffer(buffer, 3*4*8 + 4*8, 3*4*8, DynamicAttribute{
DynamicAttribute::Kind::Long, 1,
DynamicAttribute::Components::Four,
DynamicAttribute::DataType::Double})
.addVertexBuffer(buffer, 3*4*8 + 8*8, 3*4*8, DynamicAttribute{
DynamicAttribute::Kind::Long, 2,
DynamicAttribute::Components::Four,
DynamicAttribute::DataType::Double});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(DoubleShader("dmat3x4", "vec4",
"vec4(value[0][0], value[1][1], value[2][2], 0.0)"),
Revert "More vendor-specific XFAIL cases in Mesh GL test." This is behaving properly on AMD 15.300.1025.0. This reverts commit 8dd4dfcf2b1daf6d7f27868e6831c4dbc23ff89e.
11 years ago
RenderbufferFormat::RGBA16, mesh).get<Math::Vector3<UnsignedShort>>(PixelFormat::RGB, PixelType::UnsignedShort);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Expect failure in MeshGLTest. After updating my NVidia drivers the double matrix tests are extracting only the first two values and I don't know why.
12 years ago
{
GL: add XFAIL to some well-known failures with Intel Windows drivers.
7 years ago
auto drivers = Context::DetectedDriver::Amd|Context::DetectedDriver::NVidia;
#ifdef CORRADE_TARGET_WINDOWS
drivers |= Context::DetectedDriver::IntelWindows;
#endif
CORRADE_EXPECT_FAIL_IF(Context::current().detectedDriver() & drivers, "Somehow only first two values are extracted on AMD, NVidia and Intel Windows drivers.");
Expect failure in MeshGLTest. After updating my NVidia drivers the double matrix tests are extracting only the first two values and I don't know why.
12 years ago
CORRADE_COMPARE(value, Math::Vector3<UnsignedShort>(315, 65201, 2576));
}
More vendor-specific XFAIL cases in Mesh GL test.
11 years ago
Revert "More vendor-specific XFAIL cases in Mesh GL test." This is behaving properly on AMD 15.300.1025.0. This reverts commit 8dd4dfcf2b1daf6d7f27868e6831c4dbc23ff89e.
11 years ago
/* This is wrong, but check if it's still the right wrong. Fails on AMD
15.201.1151 but seems to be fixed in 15.300.1025.0 */
More consistent naming in Context::DetectedDriver. DetectedDriver::AMD is now DetectedDriver::Amd, DetectedDriver::ProbablyAngle is now just DetectedDriver::Angle. The old names are still present, but deprecated and will be removed in the future.
9 years ago
if(Context::current().detectedDriver() & (Context::DetectedDriver::Amd|Context::DetectedDriver::NVidia))
Make the XFAIL in Mesh GL test more restrictive. Surprisingly OSX drivers are the only ones that are doing it right?!
10 years ago
CORRADE_COMPARE(value, Math::Vector3<UnsignedShort>(315, 65201, 0));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
}
#endif
#ifndef MAGNUM_TARGET_GLES2
void MeshGLTest::addVertexBufferUnsignedIntWithUnsignedShort() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::EXT::gpu_shader4>())
CORRADE_SKIP(Extensions::EXT::gpu_shader4::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
Fix AMD complaints about misaligned data in Mesh GL test.
11 years ago
constexpr UnsignedShort data[] = { 0, 49563, 2128, 3821, 16583 };
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 2, 2, Attribute<0, UnsignedInt>{Attribute<0, UnsignedInt>::DataType::UnsignedShort});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4, 4, DynamicAttribute{
DynamicAttribute::Kind::Integral, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::UnsignedShort});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(IntegerShader("uint"), RenderbufferFormat::R16UI, mesh)
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
.get<UnsignedShort>(PixelFormat::RedInteger, PixelType::UnsignedShort);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
GL: XFAIL on SwiftShader with integer data type mismatches. Accepted everywhere else (or so is my experience).
7 years ago
#if defined(MAGNUM_TARGET_GLES) && !defined(MAGNUM_TARGET_WEBGL)
CORRADE_EXPECT_FAIL_IF(Context::current().detectedDriver() & Context::DetectedDriver::SwiftShader,
"SwiftShader doesn't like integer buffers with anything else than (Unsigned)Int");
#endif
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, 16583);
}
void MeshGLTest::addVertexBufferUnsignedIntWithShort() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::EXT::gpu_shader4>())
CORRADE_SKIP(Extensions::EXT::gpu_shader4::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
Fix AMD complaints about misaligned data in Mesh GL test.
11 years ago
constexpr Short data[] = { 0, 24563, 2128, 3821, 16583 };
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 2, 2, Attribute<0, UnsignedInt>{Attribute<0, UnsignedInt>::DataType::Short});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4, 4, DynamicAttribute{
DynamicAttribute::Kind::Integral, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::Short});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(IntegerShader("uint"), RenderbufferFormat::R16I, mesh)
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
.get<Short>(PixelFormat::RedInteger, PixelType::Short);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
GL: XFAIL on SwiftShader with integer data type mismatches. Accepted everywhere else (or so is my experience).
7 years ago
#if defined(MAGNUM_TARGET_GLES) && !defined(MAGNUM_TARGET_WEBGL)
CORRADE_EXPECT_FAIL_IF(Context::current().detectedDriver() & Context::DetectedDriver::SwiftShader,
"SwiftShader doesn't like integer buffers with anything else than (Unsigned)Int");
#endif
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, 16583);
}
void MeshGLTest::addVertexBufferIntWithUnsignedShort() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::EXT::gpu_shader4>())
CORRADE_SKIP(Extensions::EXT::gpu_shader4::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
Fix AMD complaints about misaligned data in Mesh GL test.
11 years ago
constexpr UnsignedShort data[] = { 0, 49563, 2128, 3821, 16583 };
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 2, 2, Attribute<0, Int>{Attribute<0, Int>::DataType::UnsignedShort});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4, 4, DynamicAttribute{
DynamicAttribute::Kind::Integral, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::UnsignedShort});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(IntegerShader("int"), RenderbufferFormat::R16UI, mesh)
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
.get<UnsignedShort>(PixelFormat::RedInteger, PixelType::UnsignedShort);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
GL: XFAIL on SwiftShader with integer data type mismatches. Accepted everywhere else (or so is my experience).
7 years ago
#if defined(MAGNUM_TARGET_GLES) && !defined(MAGNUM_TARGET_WEBGL)
CORRADE_EXPECT_FAIL_IF(Context::current().detectedDriver() & Context::DetectedDriver::SwiftShader,
"SwiftShader doesn't like integer buffers with anything else than (Unsigned)Int");
#endif
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, 16583);
}
void MeshGLTest::addVertexBufferIntWithShort() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::EXT::gpu_shader4>())
CORRADE_SKIP(Extensions::EXT::gpu_shader4::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
Fix AMD complaints about misaligned data in Mesh GL test.
11 years ago
constexpr Short data[] = { 0, 24563, 2128, 3821, -16583 };
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 2, 2, Attribute<0, Int>{Attribute<0, Int>::DataType::Short});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4, 4, DynamicAttribute{
DynamicAttribute::Kind::Integral, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::Short});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(IntegerShader("int"), RenderbufferFormat::R16I, mesh)
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
.get<Short>(PixelFormat::RedInteger, PixelType::Short);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
GL: XFAIL on SwiftShader with integer data type mismatches. Accepted everywhere else (or so is my experience).
7 years ago
#if defined(MAGNUM_TARGET_GLES) && !defined(MAGNUM_TARGET_WEBGL)
CORRADE_EXPECT_FAIL_IF(Context::current().detectedDriver() & Context::DetectedDriver::SwiftShader,
"SwiftShader doesn't like integer buffers with anything else than (Unsigned)Int");
#endif
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, -16583);
}
#endif
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#if !(defined(MAGNUM_TARGET_WEBGL) && defined(MAGNUM_TARGET_GLES2))
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
void MeshGLTest::addVertexBufferFloatWithHalfFloat() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::half_float_vertex>())
CORRADE_SKIP(Extensions::ARB::half_float_vertex::string() + std::string(" is not supported."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#elif defined(MAGNUM_TARGET_GLES2)
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::OES::vertex_half_float>())
CORRADE_SKIP(Extensions::OES::vertex_half_float::string() + std::string(" is not supported."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
Test: we have halves now, so why not test them.
9 years ago
const Half data[] = { 0.0_h, -0.7_h, Half(Math::unpack<Float, UnsignedByte>(186)) };
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Buffer buffer;
Test: we have halves now, so why not test them.
9 years ago
buffer.setData(data, BufferUsage::StaticDraw);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 2, Attribute<0, Float>{Attribute<0, Float>::DataType::HalfFloat});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 2, 2, DynamicAttribute{
DynamicAttribute::Kind::Generic, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::HalfFloat});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Test: we have halves now, so why not test them.
9 years ago
const auto value = Checker(FloatShader("float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"),
Test: what am I doing here?! Fixes test failure on ES3.
8 years ago
RenderbufferFormat::RGBA8, mesh).get<UnsignedByte>(PixelFormat::RGBA, PixelType::UnsignedByte);
Test: we have halves now, so why not test them.
9 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Test: we have halves now, so why not test them.
9 years ago
CORRADE_COMPARE(value, 186);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
}
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#ifndef MAGNUM_TARGET_GLES
void MeshGLTest::addVertexBufferFloatWithDouble() {
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
const Double data[] = { 0.0, -0.7, Math::unpack<Double, UnsignedByte>(186) };
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
mesh.setBaseVertex(1)
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
.addVertexBuffer(buffer, 8, Attribute<0, Float>(Attribute<0, Float>::DataType::Double));
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 8, Attribute<0, Float>{Attribute<0, Float>::DataType::Double});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 8, 8, DynamicAttribute{
DynamicAttribute::Kind::Generic, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::Double});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(FloatShader("float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"),
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
RenderbufferFormat::RGBA8, mesh).get<UnsignedByte>(PixelFormat::RGBA, PixelType::UnsignedShort);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, 186);
}
Implemented ARB_vertex_type_10f_11f_11f_rev (GL 4.4). Please don't look into AbstractShaderProgram::Attribute implementation. Your life will never be the same afterwards.
12 years ago
void MeshGLTest::addVertexBufferVector3WithUnsignedInt10f11f11fRev() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_type_10f_11f_11f_rev>())
CORRADE_SKIP(Extensions::ARB::vertex_type_10f_11f_11f_rev::string() + std::string(" is not available."));
Implemented ARB_vertex_type_10f_11f_11f_rev (GL 4.4). Please don't look into AbstractShaderProgram::Attribute implementation. Your life will never be the same afterwards.
12 years ago
#endif
Buffer buffer;
buffer.setData({nullptr, 12}, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 4, Attribute<0, Vector3>{Attribute<0, Vector3>::DataType::UnsignedInt10f11f11fRev});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4, 4, DynamicAttribute{
DynamicAttribute::Kind::Generic, 0,
DynamicAttribute::Components::Three,
DynamicAttribute::DataType::UnsignedInt10f11f11fRev});
} else CORRADE_ASSERT_UNREACHABLE();
Implemented ARB_vertex_type_10f_11f_11f_rev (GL 4.4). Please don't look into AbstractShaderProgram::Attribute implementation. Your life will never be the same afterwards.
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Implemented ARB_vertex_type_10f_11f_11f_rev (GL 4.4). Please don't look into AbstractShaderProgram::Attribute implementation. Your life will never be the same afterwards.
12 years ago
/* Won't test the actual values */
}
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
#ifndef MAGNUM_TARGET_GLES2
void MeshGLTest::addVertexBufferVector4WithUnsignedInt2101010Rev() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_type_2_10_10_10_rev>())
CORRADE_SKIP(Extensions::ARB::vertex_type_2_10_10_10_rev::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
Buffer buffer;
buffer.setData({nullptr, 12}, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 4, Attribute<0, Vector4>{Attribute<0, Vector4>::DataType::UnsignedInt2101010Rev});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4, 4, DynamicAttribute{
DynamicAttribute::Kind::Generic, 0,
DynamicAttribute::Components::Four,
DynamicAttribute::DataType::UnsignedInt2101010Rev});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
/* Won't test the actual values */
}
void MeshGLTest::addVertexBufferVector4WithInt2101010Rev() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_type_2_10_10_10_rev>())
CORRADE_SKIP(Extensions::ARB::vertex_type_2_10_10_10_rev::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
Buffer buffer;
buffer.setData({nullptr, 12}, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 4, Attribute<0, Vector4>{Attribute<0, Vector4>::DataType::Int2101010Rev});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4, 4, DynamicAttribute{
DynamicAttribute::Kind::Generic, 0,
DynamicAttribute::Components::Four,
DynamicAttribute::DataType::Int2101010Rev});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
/* Won't test the actual values */
}
#endif
void MeshGLTest::addVertexBufferLessVectorComponents() {
const Vector3 data[] = {
{}, {0.0f, -0.9f, 1.0f},
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
0x60189c_rgbf
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
0x551199_rgbf
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
};
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
mesh.addVertexBuffer(buffer, 3*4, Attribute<0, Vector4>{Attribute<0, Vector4>::Components::Three});
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 3*4, 3*4, DynamicAttribute{
DynamicAttribute::Kind::Generic, 0,
DynamicAttribute::Components::Three,
DynamicAttribute::DataType::Float});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(FloatShader("vec4", "valueInterpolated"),
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
CORRADE_COMPARE(value, 0x60189cff_rgba);
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 0x551199ff_rgba);
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
}
void MeshGLTest::addVertexBufferNormalized() {
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
constexpr Color4ub data[] = { {}, {0, 128, 64},
#ifndef MAGNUM_TARGET_GLES2
0x209ce4_rgb
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
0x1199dd_rgb
#endif
};
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
typedef Attribute<0, Vector3> Attribute;
mesh.addVertexBuffer(buffer, 4, Attribute(Attribute::DataType::UnsignedByte, Attribute::DataOption::Normalized), 1);
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4, 4, DynamicAttribute{
DynamicAttribute::Kind::GenericNormalized, 0,
DynamicAttribute::Components::Three,
DynamicAttribute::DataType::UnsignedByte});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(FloatShader("vec3", "vec4(valueInterpolated, 0.0)"),
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color3ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
CORRADE_COMPARE(value, 0x209ce4_rgb);
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 0x1199dd_rgb);
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
}
#ifndef MAGNUM_TARGET_GLES
void MeshGLTest::addVertexBufferBGRA() {
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_array_bgra>())
CORRADE_SKIP(Extensions::ARB::vertex_array_bgra::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
constexpr Color4ub data[] = { {}, {0, 128, 64, 161}, {96, 24, 156, 225} };
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Ability to specify shader attribute properties fully at runtime. There's a new DynamicAttribute class that is very similar to Attribute, but it has the location and base type as runtime properties instead of them being a part of template. This allows for more flexibility, but OTOH also more typing and more responsibility on the user. See MeshGLTest for details and usage comparison to the Attribute API.
9 years ago
mesh.setBaseVertex(1);
if(testCaseInstanceId() == 0) {
setTestCaseDescription("Attribute");
typedef Attribute<0, Vector4> Attribute;
mesh.addVertexBuffer(buffer, 4, Attribute(Attribute::Components::BGRA, Attribute::DataType::UnsignedByte, Attribute::DataOption::Normalized));
} else if(testCaseInstanceId() == 1) {
setTestCaseDescription("DynamicAttribute");
mesh.addVertexBuffer(buffer, 4, 4, DynamicAttribute{
DynamicAttribute::Kind::GenericNormalized, 0,
DynamicAttribute::Components::BGRA,
DynamicAttribute::DataType::UnsignedByte});
} else CORRADE_ASSERT_UNREACHABLE();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(FloatShader("vec4", "valueInterpolated"),
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
RenderbufferFormat::RGBA8, mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, Color4ub(156, 24, 96, 225));
}
#endif
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
struct MultipleShader: AbstractShaderProgram {
typedef Attribute<0, Vector3> Position;
typedef Attribute<1, Vector3> Normal;
typedef Attribute<2, Vector2> TextureCoordinates;
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
explicit MultipleShader();
};
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Various documentation fixes and updates. {} in Matrix constructor caused Doxygen to forget all functions after it.
13 years ago
#ifndef DOXYGEN_GENERATING_OUTPUT
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
MultipleShader::MultipleShader() {
#ifndef MAGNUM_TARGET_GLES
Don't use GLSL < 130 in GL tests on OSX. OSX doesn't support anything below what's in GL 3.1 when using core contexts.
10 years ago
Shader vert(
#ifndef CORRADE_TARGET_APPLE
Version::GL210
#else
Version::GL310
#endif
, Shader::Type::Vertex);
Shader frag(
#ifndef CORRADE_TARGET_APPLE
Version::GL210
#else
Version::GL310
#endif
, Shader::Type::Fragment);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#else
Shader vert(Version::GLES200, Shader::Type::Vertex);
Make use of threaded shader compilation in MeshGLTest.
12 years ago
Shader frag(Version::GLES200, Shader::Type::Fragment);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
Don't use GLSL < 130 in GL tests on OSX. OSX doesn't support anything below what's in GL 3.1 when using core contexts.
10 years ago
vert.addSource(
"#if !defined(GL_ES) && __VERSION__ == 120\n"
"#define mediump\n"
"#endif\n"
"#if defined(GL_ES) || __VERSION__ == 120\n"
"#define in attribute\n"
"#define out varying\n"
"#endif\n"
"in mediump vec4 position;\n"
"in mediump vec3 normal;\n"
"in mediump vec2 textureCoordinates;\n"
"out mediump vec4 valueInterpolated;\n"
"void main() {\n"
" valueInterpolated = position + vec4(normal, 0.0) + vec4(textureCoordinates, 0.0, 0.0);\n"
" gl_Position = vec4(0.0, 0.0, 0.0, 1.0);\n"
"}\n");
frag.addSource(
"#if !defined(GL_ES) && __VERSION__ == 120\n"
"#define mediump\n"
"#endif\n"
"#if defined(GL_ES) || __VERSION__ == 120\n"
"#define in varying\n"
"#define result gl_FragColor\n"
"#endif\n"
"in mediump vec4 valueInterpolated;\n"
"#if !defined(GL_ES) && __VERSION__ >= 130\n"
"out mediump vec4 result;\n"
"#endif\n"
"void main() { result = valueInterpolated; }\n");
Make use of threaded shader compilation in MeshGLTest.
12 years ago
CORRADE_INTERNAL_ASSERT_OUTPUT(Shader::compile({vert, frag}));
Use threaded shader compilation etc. also in tests.
12 years ago
attachShaders({vert, frag});
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
bindAttributeLocation(Position::Location, "position");
bindAttributeLocation(Normal::Location, "normal");
bindAttributeLocation(TextureCoordinates::Location, "textureCoordinates");
CORRADE_INTERNAL_ASSERT_OUTPUT(link());
}
Various documentation fixes and updates. {} in Matrix constructor caused Doxygen to forget all functions after it.
13 years ago
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
void MeshGLTest::addVertexBufferMultiple() {
const Float data[] = {
0.0f, /* Offset */
/* First attribute */
0.3f, 0.1f, 0.5f,
0.4f, 0.0f, -0.9f,
1.0f, -0.5f,
/* Second attribute */
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Math::unpack<Float, UnsignedByte>(64),
Math::unpack<Float, UnsignedByte>(17),
Math::unpack<Float, UnsignedByte>(56),
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Math::unpack<Float, UnsignedByte>(15),
Math::unpack<Float, UnsignedByte>(164),
Math::unpack<Float, UnsignedByte>(17),
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
Math::unpack<Float, UnsignedByte>(9),
Math::unpack<Float, UnsignedByte>(159),
Math::unpack<Float, UnsignedByte>(12),
#endif
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Math::unpack<Float, UnsignedByte>(97),
Math::unpack<Float, UnsignedByte>(28)
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
};
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
mesh.setBaseVertex(1)
.addVertexBuffer(buffer, 1*4, MultipleShader::Position(),
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
MultipleShader::Normal(), MultipleShader::TextureCoordinates());
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(MultipleShader(),
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, Color4ub(64 + 15 + 97, 17 + 164 + 28, 56 + 17, 255));
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 0xaacc44ff_rgba);
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
}
void MeshGLTest::addVertexBufferMultipleGaps() {
const Float data[] = {
0.0f, 0.0f, 0.0f, 0.0f, /* Offset */
/* First attribute */
0.3f, 0.1f, 0.5f, 0.0f,
0.4f, 0.0f, -0.9f, 0.0f,
1.0f, -0.5f, 0.0f, 0.0f,
/* Second attribute */
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Math::unpack<Float, UnsignedByte>(64),
Math::unpack<Float, UnsignedByte>(17),
Math::unpack<Float, UnsignedByte>(56), 0.0f,
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Math::unpack<Float, UnsignedByte>(15),
Math::unpack<Float, UnsignedByte>(164),
Math::unpack<Float, UnsignedByte>(17), 0.0f,
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
Math::unpack<Float, UnsignedByte>(9),
Math::unpack<Float, UnsignedByte>(159),
Math::unpack<Float, UnsignedByte>(12), 0.0f,
#endif
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Math::unpack<Float, UnsignedByte>(97),
Math::unpack<Float, UnsignedByte>(28), 0.0f, 0.0f
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
};
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
mesh.setBaseVertex(1)
.addVertexBuffer(buffer, 4*4,
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
MultipleShader::Position(), 1*4,
MultipleShader::Normal(), 1*4,
MultipleShader::TextureCoordinates(), 2*4);
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
const auto value = Checker(MultipleShader(),
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, Color4ub(64 + 15 + 97, 17 + 164 + 28, 56 + 17, 255));
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 0xaacc44ff_rgba);
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
}
GL: assert that addVertexBuffer() is called with a valid Buffer.
8 years ago
void MeshGLTest::addVertexBufferMovedOutInstance() {
Buffer buffer{NoCreate};
Mesh mesh;
std::ostringstream out;
Error redirectError{&out};
mesh.addVertexBuffer(buffer, 0, Attribute<0, Float>{});
CORRADE_COMPARE(out.str(), "GL::Mesh::addVertexBuffer(): empty or moved-out Buffer instance was passed\n");
}
GL: make it possible for a Mesh to own the vertex/index buffers. I wanted to do this since 2015.
8 years ago
void MeshGLTest::addVertexBufferTransferOwnwership() {
const Float data = 1.0f;
Buffer buffer;
buffer.setData({&data, 1}, BufferUsage::StaticDraw);
const GLuint id = buffer.id();
CORRADE_VERIFY(glIsBuffer(id));
{
Mesh mesh;
mesh.addVertexBuffer(buffer, 0, Attribute<0, Float>{});
CORRADE_VERIFY(buffer.id());
CORRADE_VERIFY(glIsBuffer(id));
}
CORRADE_VERIFY(glIsBuffer(id));
{
Mesh mesh;
mesh.addVertexBuffer(std::move(buffer), 0, Attribute<0, Float>{});
CORRADE_VERIFY(!buffer.id());
CORRADE_VERIFY(glIsBuffer(id));
}
CORRADE_VERIFY(!glIsBuffer(id));
}
void MeshGLTest::addVertexBufferInstancedTransferOwnwership() {
GL: disable indexed Mesh tests on drivers w/o indexed extensions.
7 years ago
#ifndef MAGNUM_TARGET_GLES
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_instanced>())
CORRADE_SKIP(Extensions::ARB::draw_instanced::string() + std::string(" is not available."));
#elif defined(MAGNUM_TARGET_GLES2)
#ifndef MAGNUM_TARGET_WEBGL
if(!Context::current().isExtensionSupported<Extensions::ANGLE::instanced_arrays>() &&
!Context::current().isExtensionSupported<Extensions::EXT::instanced_arrays>() &&
!Context::current().isExtensionSupported<Extensions::NV::instanced_arrays>())
CORRADE_SKIP("Required extension is not available.");
#else
if(!Context::current().isExtensionSupported<Extensions::ANGLE::instanced_arrays>())
CORRADE_SKIP(Extensions::ANGLE::instanced_arrays::string() + std::string(" is not available."));
#endif
#endif
GL: make it possible for a Mesh to own the vertex/index buffers. I wanted to do this since 2015.
8 years ago
const Float data = 1.0f;
Buffer buffer;
buffer.setData({&data, 1}, BufferUsage::StaticDraw);
const GLuint id = buffer.id();
CORRADE_VERIFY(glIsBuffer(id));
{
Mesh mesh;
mesh.addVertexBufferInstanced(buffer, 1, 0, Attribute<0, Float>{});
CORRADE_VERIFY(buffer.id());
CORRADE_VERIFY(glIsBuffer(id));
}
CORRADE_VERIFY(glIsBuffer(id));
{
Mesh mesh;
mesh.addVertexBufferInstanced(std::move(buffer), 1, 0, Attribute<0, Float>{});
CORRADE_VERIFY(!buffer.id());
CORRADE_VERIFY(glIsBuffer(id));
}
CORRADE_VERIFY(!glIsBuffer(id));
}
void MeshGLTest::addVertexBufferDynamicTransferOwnwership() {
const Float data = 1.0f;
Buffer buffer;
buffer.setData({&data, 1}, BufferUsage::StaticDraw);
const GLuint id = buffer.id();
CORRADE_VERIFY(glIsBuffer(id));
{
Mesh mesh;
mesh.addVertexBuffer(buffer, 0, 4, DynamicAttribute{
DynamicAttribute::Kind::GenericNormalized, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::Float});
CORRADE_VERIFY(buffer.id());
CORRADE_VERIFY(glIsBuffer(id));
}
CORRADE_VERIFY(glIsBuffer(id));
{
Mesh mesh;
mesh.addVertexBuffer(std::move(buffer), 0, 4, DynamicAttribute{
DynamicAttribute::Kind::GenericNormalized, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::Float});
CORRADE_VERIFY(!buffer.id());
CORRADE_VERIFY(glIsBuffer(id));
}
CORRADE_VERIFY(!glIsBuffer(id));
}
void MeshGLTest::addVertexBufferInstancedDynamicTransferOwnwership() {
GL: disable indexed Mesh tests on drivers w/o indexed extensions.
7 years ago
#ifndef MAGNUM_TARGET_GLES
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_instanced>())
CORRADE_SKIP(Extensions::ARB::draw_instanced::string() + std::string(" is not available."));
#elif defined(MAGNUM_TARGET_GLES2)
#ifndef MAGNUM_TARGET_WEBGL
if(!Context::current().isExtensionSupported<Extensions::ANGLE::instanced_arrays>() &&
!Context::current().isExtensionSupported<Extensions::EXT::instanced_arrays>() &&
!Context::current().isExtensionSupported<Extensions::NV::instanced_arrays>())
CORRADE_SKIP("Required extension is not available.");
#else
if(!Context::current().isExtensionSupported<Extensions::ANGLE::instanced_arrays>())
CORRADE_SKIP(Extensions::ANGLE::instanced_arrays::string() + std::string(" is not available."));
#endif
#endif
GL: make it possible for a Mesh to own the vertex/index buffers. I wanted to do this since 2015.
8 years ago
const Float data = 1.0f;
Buffer buffer;
buffer.setData({&data, 1}, BufferUsage::StaticDraw);
const GLuint id = buffer.id();
CORRADE_VERIFY(glIsBuffer(id));
{
Mesh mesh;
mesh.addVertexBufferInstanced(buffer, 1, 0, 4, DynamicAttribute{
DynamicAttribute::Kind::GenericNormalized, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::Float});
CORRADE_VERIFY(buffer.id());
CORRADE_VERIFY(glIsBuffer(id));
}
CORRADE_VERIFY(glIsBuffer(id));
{
Mesh mesh;
mesh.addVertexBufferInstanced(std::move(buffer), 1, 0, 4, DynamicAttribute{
DynamicAttribute::Kind::GenericNormalized, 0,
DynamicAttribute::Components::One,
DynamicAttribute::DataType::Float});
CORRADE_VERIFY(!buffer.id());
CORRADE_VERIFY(glIsBuffer(id));
}
CORRADE_VERIFY(!glIsBuffer(id));
}
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
const Float indexedVertexData[] = {
0.0f, /* Offset */
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
/* First vertex */
Math::unpack<Float, UnsignedByte>(64),
Math::unpack<Float, UnsignedByte>(17),
Math::unpack<Float, UnsignedByte>(56),
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
Math::unpack<Float, UnsignedByte>(15),
Math::unpack<Float, UnsignedByte>(164),
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Math::unpack<Float, UnsignedByte>(17),
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else
/* Using only RGBA4, supply less precision. This has to be one on the input
because SwiftShader stores RGBA4 as RGBA8 internally, thus preserving
the full precision of the input. */
Math::unpack<Float, UnsignedByte>(9),
Math::unpack<Float, UnsignedByte>(159),
Math::unpack<Float, UnsignedByte>(12),
#endif
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
Math::unpack<Float, UnsignedByte>(97),
Math::unpack<Float, UnsignedByte>(28),
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
/* Second vertex */
0.3f, 0.1f, 0.5f,
0.4f, 0.0f, -0.9f,
1.0f, -0.5f
};
Cleanup and code reorganization in MeshGLTest.
12 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
#ifndef MAGNUM_TARGET_GLES
const Float indexedVertexDataBaseVertex[] = {
0.0f, 0.0f, /* Offset */
Cleanup and code reorganization in MeshGLTest.
12 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
/* First vertex */
0.0f, 0.0f, 0.0f,
Cleanup and code reorganization in MeshGLTest.
12 years ago
0.0f, 0.0f, 0.0f,
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
0.0f, 0.0f,
Cleanup and code reorganization in MeshGLTest.
12 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
/* Second vertex */
0.0f, 0.0f, 0.0f,
Cleanup and code reorganization in MeshGLTest.
12 years ago
0.0f, 0.0f, 0.0f,
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
0.0f, 0.0f,
/* Third vertex */
Math::unpack<Float, UnsignedByte>(64),
Math::unpack<Float, UnsignedByte>(17),
Math::unpack<Float, UnsignedByte>(56),
Math::unpack<Float, UnsignedByte>(15),
Math::unpack<Float, UnsignedByte>(164),
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Math::unpack<Float, UnsignedByte>(17),
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
Math::unpack<Float, UnsignedByte>(97),
Math::unpack<Float, UnsignedByte>(28),
Cleanup and code reorganization in MeshGLTest.
12 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
/* Fourth vertex */
0.3f, 0.1f, 0.5f,
0.4f, 0.0f, -0.9f,
1.0f, -0.5f
};
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
constexpr Color4ub indexedResult{64 + 15 + 97, 17 + 164 + 28, 56 + 17, 255};
#else /* RGBA4, so less precision */
constexpr Color4ub indexedResult{0xaa, 0xcc, 0x44, 0xff};
#endif
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
template<class T> void MeshGLTest::setIndexBuffer() {
setTestCaseName(std::is_same<T, MeshIndexType>::value ?
"setIndexBuffer<GL::MeshIndexType>" :
"setIndexBuffer<Magnum::MeshIndexType>");
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Buffer vertices;
vertices.setData(indexedVertexData, BufferUsage::StaticDraw);
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
constexpr UnsignedByte indexData[] = { 2, 1, 0 };
Implemented indexed buffer binding. The indexed binding is allowed for only some types (atomic counters, uniforms, shader storage and transform feedback), thus we need separate enum for that. Because the bind() function will be used far more often than setTargetHint(), the original Target enum is now renamed to TargetHint and the new Target enum contains (in non-deprecated build) only three values. For backwards compatibility, though, we need to have all original Target values, thus the new Target enum contains also all other values from TargetHint, but they are marked as deprecated and (at least) run-time checked in bind() so they aren't accidentaly used for indexed binding. Similarly there are also deprecated Target overloads of Buffer() and setTargetHint(). It's ugly, but hopefully will suffice for now. This mess will be removed as soon as possible in some upcoming version.
12 years ago
Buffer indices{Buffer::TargetHint::ElementArray};
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
indices.setData(indexData, BufferUsage::StaticDraw);
Mesh mesh;
mesh.addVertexBuffer(vertices, 1*4, MultipleShader::Position(),
MultipleShader::Normal(), MultipleShader::TextureCoordinates())
GL: fix various OOB / unaligned access issues in Mesh test.
8 years ago
.setIndexBuffer(indices, 1, T::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
CORRADE_COMPARE(mesh.indexType(), MeshIndexType::UnsignedByte);
CORRADE_COMPARE(mesh.indexTypeSize(), 1);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
const auto value = Checker(MultipleShader{},
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, indexedResult);
}
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
template<class T> void MeshGLTest::setIndexBufferRange() {
setTestCaseName(std::is_same<T, MeshIndexType>::value ?
"setIndexBufferRange<GL::MeshIndexType>" :
"setIndexBufferRange<Magnum::MeshIndexType>");
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Buffer vertices;
vertices.setData(indexedVertexData, BufferUsage::StaticDraw);
constexpr UnsignedShort indexData[] = { 2, 1, 0 };
Implemented indexed buffer binding. The indexed binding is allowed for only some types (atomic counters, uniforms, shader storage and transform feedback), thus we need separate enum for that. Because the bind() function will be used far more often than setTargetHint(), the original Target enum is now renamed to TargetHint and the new Target enum contains (in non-deprecated build) only three values. For backwards compatibility, though, we need to have all original Target values, thus the new Target enum contains also all other values from TargetHint, but they are marked as deprecated and (at least) run-time checked in bind() so they aren't accidentaly used for indexed binding. Similarly there are also deprecated Target overloads of Buffer() and setTargetHint(). It's ugly, but hopefully will suffice for now. This mess will be removed as soon as possible in some upcoming version.
12 years ago
Buffer indices{Buffer::TargetHint::ElementArray};
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
indices.setData(indexData, BufferUsage::StaticDraw);
Mesh mesh;
mesh.addVertexBuffer(vertices, 1*4, MultipleShader::Position(),
MultipleShader::Normal(), MultipleShader::TextureCoordinates())
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
.setIndexBuffer(indices, 2, T::UnsignedShort, 0, 1);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
CORRADE_COMPARE(mesh.indexType(), GL::MeshIndexType::UnsignedShort);
CORRADE_COMPARE(mesh.indexTypeSize(), 2);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
const auto value = Checker(MultipleShader{},
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, indexedResult);
}
void MeshGLTest::setIndexBufferUnsignedInt() {
#ifdef MAGNUM_TARGET_GLES2
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::OES::element_index_uint>())
CORRADE_SKIP(Extensions::OES::element_index_uint::string() + std::string(" is not available."));
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
#endif
Buffer vertices;
vertices.setData(indexedVertexData, BufferUsage::StaticDraw);
constexpr UnsignedInt indexData[] = { 2, 1, 0 };
Implemented indexed buffer binding. The indexed binding is allowed for only some types (atomic counters, uniforms, shader storage and transform feedback), thus we need separate enum for that. Because the bind() function will be used far more often than setTargetHint(), the original Target enum is now renamed to TargetHint and the new Target enum contains (in non-deprecated build) only three values. For backwards compatibility, though, we need to have all original Target values, thus the new Target enum contains also all other values from TargetHint, but they are marked as deprecated and (at least) run-time checked in bind() so they aren't accidentaly used for indexed binding. Similarly there are also deprecated Target overloads of Buffer() and setTargetHint(). It's ugly, but hopefully will suffice for now. This mess will be removed as soon as possible in some upcoming version.
12 years ago
Buffer indices{Buffer::TargetHint::ElementArray};
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
indices.setData(indexData, BufferUsage::StaticDraw);
Mesh mesh;
mesh.addVertexBuffer(vertices, 1*4, MultipleShader::Position(),
MultipleShader::Normal(), MultipleShader::TextureCoordinates())
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
.setIndexBuffer(indices, 4, MeshIndexType::UnsignedInt);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
CORRADE_COMPARE(mesh.indexType(), GL::MeshIndexType::UnsignedInt);
CORRADE_COMPARE(mesh.indexTypeSize(), 4);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
const auto value = Checker(MultipleShader{},
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Fully test mesh functionality. This was painful. The test currently fails for a few reasons, see following commits.
13 years ago
CORRADE_COMPARE(value, indexedResult);
}
GL: assert that addVertexBuffer() is called with a valid Buffer.
8 years ago
void MeshGLTest::setIndexBufferMovedOutInstance() {
Buffer buffer{NoCreate};
Mesh mesh;
std::ostringstream out;
Error redirectError{&out};
mesh.setIndexBuffer(buffer, 0, MeshIndexType::UnsignedByte);
CORRADE_COMPARE(out.str(), "GL::Mesh::setIndexBuffer(): empty or moved-out Buffer instance was passed\n");
}
GL: make it possible for a Mesh to own the vertex/index buffers. I wanted to do this since 2015.
8 years ago
template<class T> void MeshGLTest::setIndexBufferTransferOwnership() {
setTestCaseName(std::is_same<T, MeshIndexType>::value ?
"setIndexBufferTransferOwnership<GL::MeshIndexType>" :
"setIndexBufferTransferOwnership<Magnum::MeshIndexType>");
const UnsignedShort data = 0;
GL: properly specify Buffer target hints for tests on WebGL.
8 years ago
Buffer buffer{Buffer::TargetHint::ElementArray};
GL: make it possible for a Mesh to own the vertex/index buffers. I wanted to do this since 2015.
8 years ago
buffer.setData({&data, 1}, BufferUsage::StaticDraw);
const GLuint id = buffer.id();
CORRADE_VERIFY(glIsBuffer(id));
{
Mesh mesh;
mesh.setIndexBuffer(buffer, 0, T::UnsignedShort);
CORRADE_VERIFY(buffer.id());
CORRADE_VERIFY(glIsBuffer(id));
}
CORRADE_VERIFY(glIsBuffer(id));
{
Mesh mesh;
mesh.setIndexBuffer(std::move(buffer), 0, T::UnsignedShort);
CORRADE_VERIFY(!buffer.id());
CORRADE_VERIFY(glIsBuffer(id));
}
CORRADE_VERIFY(!glIsBuffer(id));
}
template<class T> void MeshGLTest::setIndexBufferRangeTransferOwnership() {
setTestCaseName(std::is_same<T, MeshIndexType>::value ?
"setIndexBufferRangeTransferOwnership<GL::MeshIndexType>" :
"setIndexBufferRangeTransferOwnership<Magnum::MeshIndexType>");
const UnsignedShort data = 0;
GL: properly specify Buffer target hints for tests on WebGL.
8 years ago
Buffer buffer{Buffer::TargetHint::ElementArray};
GL: make it possible for a Mesh to own the vertex/index buffers. I wanted to do this since 2015.
8 years ago
buffer.setData({&data, 1}, BufferUsage::StaticDraw);
const GLuint id = buffer.id();
CORRADE_VERIFY(glIsBuffer(id));
{
Mesh mesh;
mesh.setIndexBuffer(buffer, 0, T::UnsignedShort, 0, 1);
CORRADE_VERIFY(buffer.id());
CORRADE_VERIFY(glIsBuffer(id));
}
CORRADE_VERIFY(glIsBuffer(id));
{
Mesh mesh;
mesh.setIndexBuffer(std::move(buffer), 0, T::UnsignedShort, 0, 1);
CORRADE_VERIFY(!buffer.id());
CORRADE_VERIFY(glIsBuffer(id));
}
CORRADE_VERIFY(!glIsBuffer(id));
}
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
void MeshGLTest::unbindVAOWhenSettingIndexBufferData() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_array_object>())
CORRADE_SKIP(Extensions::ARB::vertex_array_object::string() + std::string(" is not available."));
if(Context::current().isExtensionSupported<Extensions::ARB::direct_state_access>())
CORRADE_SKIP(Extensions::ARB::direct_state_access::string() + std::string(" is active with circumvents the issue tested here."));
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
#elif defined(MAGNUM_TARGET_GLES2)
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::OES::vertex_array_object>())
CORRADE_SKIP(Extensions::OES::vertex_array_object::string() + std::string(" is not available."));
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
#endif
typedef Attribute<0, Float> Attribute;
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
const Float data[] = { -0.7f, Math::unpack<Float, UnsignedByte>(
#ifndef MAGNUM_TARGET_GLES2
92
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
85
#endif
), Math::unpack<Float, UnsignedByte>(32) };
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
Buffer buffer{Buffer::TargetHint::Array};
buffer.setData(data, BufferUsage::StaticDraw);
Buffer indices{Buffer::TargetHint::ElementArray};
indices.setData(std::vector<UnsignedByte>{5, 0}, BufferUsage::StaticDraw);
Mesh mesh;
mesh.addVertexBuffer(buffer, 4, Attribute{})
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
.setIndexBuffer(indices, 0, MeshIndexType::UnsignedByte);
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
/* This buffer should have no effect on the mesh above */
Buffer otherIndices{Buffer::TargetHint::ElementArray};
otherIndices.setData(std::vector<UnsignedByte>{100, 1}, BufferUsage::StaticDraw);
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
const auto value = Checker(FloatShader("float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"),
#ifndef MAGNUM_TARGET_GLES2
Improve state tracker resetting functionality.
8 years ago
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
mesh).get<UnsignedByte>(PixelFormat::RGBA, PixelType::UnsignedByte);
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
Improve state tracker resetting functionality.
8 years ago
CORRADE_COMPARE(value, 92);
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 85);
#endif
Improve state tracker resetting functionality.
8 years ago
}
GL: test yet another interaction of VAOs and index buffer binding. Fails (well, why else would I be adding the test, right?).
7 years ago
void MeshGLTest::unbindIndexBufferWhenBindingVao() {
#ifndef MAGNUM_TARGET_GLES
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_array_object>())
CORRADE_SKIP(Extensions::ARB::vertex_array_object::string() + std::string(" is not available."));
if(Context::current().isExtensionSupported<Extensions::ARB::direct_state_access>())
CORRADE_SKIP(Extensions::ARB::direct_state_access::string() + std::string(" is active with circumvents the issue tested here."));
#elif defined(MAGNUM_TARGET_GLES2)
if(!Context::current().isExtensionSupported<Extensions::OES::vertex_array_object>())
CORRADE_SKIP(Extensions::OES::vertex_array_object::string() + std::string(" is not available."));
#endif
typedef Attribute<0, Float> Attribute;
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
const Float data[] = { -0.7f, Math::unpack<Float, UnsignedByte>(
#ifndef MAGNUM_TARGET_GLES2
92
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
85
#endif
), Math::unpack<Float, UnsignedByte>(32) };
GL: test yet another interaction of VAOs and index buffer binding. Fails (well, why else would I be adding the test, right?).
7 years ago
Buffer vertices{Buffer::TargetHint::Array};
vertices.setData(data, BufferUsage::StaticDraw);
Buffer indices{Buffer::TargetHint::ElementArray};
/* Just reserve the memory first */
indices.setData({nullptr, 2}, BufferUsage::StaticDraw);
/* Create an indexed mesh first */
Mesh indexed;
indexed.addVertexBuffer(vertices, 0, Attribute{})
.setIndexBuffer(indices, 0, MeshIndexType::UnsignedByte);
/* Now bind an nonindexed mesh */
Mesh nonindexed;
nonindexed.addVertexBuffer(vertices, 0, Attribute{});
/* Fill index buffer for the indexed mesh */
indices.setData(std::vector<UnsignedByte>{5, 1}, BufferUsage::StaticDraw);
MAGNUM_VERIFY_NO_GL_ERROR();
/* Draw the indexed mesh. The index buffer should be correctly updated,
picking the second vertex with value of 92. */
const auto value = Checker(FloatShader("float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"),
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
indexed).get<UnsignedByte>(PixelFormat::RGBA, PixelType::UnsignedByte);
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: test yet another interaction of VAOs and index buffer binding. Fails (well, why else would I be adding the test, right?).
7 years ago
CORRADE_COMPARE(value, 92);
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 85);
#endif
GL: test yet another interaction of VAOs and index buffer binding. Fails (well, why else would I be adding the test, right?).
7 years ago
}
GL: further harden VAO and index buffer binding state interaction tests. This one crashes. Turns out the 169031fb7bf30e2c98c81cbf865aa5a4662aa716 contained a random temporary test state instead of the real solution (and so the comment didn't even match the code, it should have been resetting that to 0). That also made some tests fail with DSA disabled, but none of the tests were actual Mesh tests, just accidentally hitting the problematic code path. I took the opportunity to look at this more closely and investigate *why* this failed -- turns out, in setIndexBuffer(), I was resetting the state tracker to a VAO state that was about to be set in the very next step, and then, when doing that next step the state tracker "optimized away" the state change because it thought it was already done (even though it wasn't). The new test in MeshGLTest covers this particular case.
7 years ago
void MeshGLTest::resetIndexBufferBindingWhenBindingVao() {
#ifndef MAGNUM_TARGET_GLES
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_array_object>())
CORRADE_SKIP(Extensions::ARB::vertex_array_object::string() + std::string(" is not available."));
if(Context::current().isExtensionSupported<Extensions::ARB::direct_state_access>())
CORRADE_SKIP(Extensions::ARB::direct_state_access::string() + std::string(" is active with circumvents the issue tested here."));
#elif defined(MAGNUM_TARGET_GLES2)
if(!Context::current().isExtensionSupported<Extensions::OES::vertex_array_object>())
CORRADE_SKIP(Extensions::OES::vertex_array_object::string() + std::string(" is not available."));
#endif
typedef Attribute<0, Float> Attribute;
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
const Float data[] = { -0.7f, Math::unpack<Float, UnsignedByte>(
#ifndef MAGNUM_TARGET_GLES2
92
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
85
#endif
), Math::unpack<Float, UnsignedByte>(32) };
GL: further harden VAO and index buffer binding state interaction tests. This one crashes. Turns out the 169031fb7bf30e2c98c81cbf865aa5a4662aa716 contained a random temporary test state instead of the real solution (and so the comment didn't even match the code, it should have been resetting that to 0). That also made some tests fail with DSA disabled, but none of the tests were actual Mesh tests, just accidentally hitting the problematic code path. I took the opportunity to look at this more closely and investigate *why* this failed -- turns out, in setIndexBuffer(), I was resetting the state tracker to a VAO state that was about to be set in the very next step, and then, when doing that next step the state tracker "optimized away" the state change because it thought it was already done (even though it wasn't). The new test in MeshGLTest covers this particular case.
7 years ago
Buffer vertices{Buffer::TargetHint::Array};
vertices.setData(data);
/* Create an indexed mesh */
Mesh indexed;
indexed.addVertexBuffer(vertices, 0, Attribute{});
/* Create an index buffer and fill it (the VAO is bound now, so it'll get
unbound to avoid messing with its state). */
Buffer indices{Buffer::TargetHint::ElementArray};
indices.setData(std::vector<UnsignedByte>{5, 1});
/* Add the index buffer. The VAO is unbound, so it gets bound. That resets
the element array buffer binding and then the buffer gets bound to the
VAO. */
indexed.setIndexBuffer(indices, 0, MeshIndexType::UnsignedByte);
MAGNUM_VERIFY_NO_GL_ERROR();
/* Draw the indexed mesh. The index buffer should be correctly bound,
picking the second vertex with value of 92. */
const auto value = Checker(FloatShader("float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"),
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
indexed).get<UnsignedByte>(PixelFormat::RGBA, PixelType::UnsignedByte);
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: further harden VAO and index buffer binding state interaction tests. This one crashes. Turns out the 169031fb7bf30e2c98c81cbf865aa5a4662aa716 contained a random temporary test state instead of the real solution (and so the comment didn't even match the code, it should have been resetting that to 0). That also made some tests fail with DSA disabled, but none of the tests were actual Mesh tests, just accidentally hitting the problematic code path. I took the opportunity to look at this more closely and investigate *why* this failed -- turns out, in setIndexBuffer(), I was resetting the state tracker to a VAO state that was about to be set in the very next step, and then, when doing that next step the state tracker "optimized away" the state change because it thought it was already done (even though it wasn't). The new test in MeshGLTest covers this particular case.
7 years ago
CORRADE_COMPARE(value, 92);
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 85);
#endif
GL: further harden VAO and index buffer binding state interaction tests. This one crashes. Turns out the 169031fb7bf30e2c98c81cbf865aa5a4662aa716 contained a random temporary test state instead of the real solution (and so the comment didn't even match the code, it should have been resetting that to 0). That also made some tests fail with DSA disabled, but none of the tests were actual Mesh tests, just accidentally hitting the problematic code path. I took the opportunity to look at this more closely and investigate *why* this failed -- turns out, in setIndexBuffer(), I was resetting the state tracker to a VAO state that was about to be set in the very next step, and then, when doing that next step the state tracker "optimized away" the state change because it thought it was already done (even though it wasn't). The new test in MeshGLTest covers this particular case.
7 years ago
}
Improve state tracker resetting functionality.
8 years ago
void MeshGLTest::unbindVAOBeforeEnteringExternalSection() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_array_object>())
CORRADE_SKIP(Extensions::ARB::vertex_array_object::string() + std::string(" is not available."));
Improve state tracker resetting functionality.
8 years ago
#elif defined(MAGNUM_TARGET_GLES2)
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::OES::vertex_array_object>())
CORRADE_SKIP(Extensions::OES::vertex_array_object::string() + std::string(" is not available."));
Improve state tracker resetting functionality.
8 years ago
#endif
typedef Attribute<0, Float> Attribute;
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
const Float data[] = { -0.7f, Math::unpack<Float, UnsignedByte>(
#ifndef MAGNUM_TARGET_GLES2
92
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
85
#endif
), Math::unpack<Float, UnsignedByte>(32) };
Improve state tracker resetting functionality.
8 years ago
Buffer buffer{Buffer::TargetHint::Array};
buffer.setData(data, BufferUsage::StaticDraw);
Buffer indices{Buffer::TargetHint::ElementArray};
indices.setData(std::vector<UnsignedByte>{5, 0}, BufferUsage::StaticDraw);
Mesh mesh;
mesh.addVertexBuffer(buffer, 4, Attribute{})
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
.setIndexBuffer(indices, 0, MeshIndexType::UnsignedByte);
Improve state tracker resetting functionality.
8 years ago
{
/* Comment this out to watch the world burn */
Context::current().resetState(Context::State::MeshVao);
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
/* Be nice to the other tests */
Context::current().resetState(Context::State::ExitExternal);
}
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Improve state tracker resetting functionality.
8 years ago
const auto value = Checker(FloatShader("float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"),
#ifndef MAGNUM_TARGET_GLES2
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
mesh).get<UnsignedByte>(PixelFormat::RGBA, PixelType::UnsignedByte);
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
CORRADE_COMPARE(value, 92);
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 85);
#endif
Test that binding index buffer internally doesn't affect bound VAOs. Fails. Problem spotted on WebGL 2 and is an unfortunate consequence of these events: 1. Neither EXT_DSA nor ARB_DSA is available, but VAOs are available, which means virtually all ES 3 and WebGL 2 implementations and also ES 2 / WebGL 1 with OES_VAO available. 2. Index buffer gets created with TargetHint::ElementArray, bound to GL_ELEMENT_ARRAY_BUFFER and filled with data. 3. Mesh object with VAO inside is created and the index buffer from above gets bound to GL_ELEMENT_ARRAY_BUFFER again to attach it to the VAO. 4. Another index buffer, possibly for another mesh, gets created with TargetHint::ElementArray and bound to GL_ELEMENT_ARRAY_BUFFER in order to be filled with data. But because the VAO from above is still bound, the index buffer attachment is then stomped on. 5. Rendering such mesh will use a different index buffer, most probably causing some out-of-range GL error and nothing rendered.
9 years ago
}
GL: make it possible to bind a scratch VAO for external GL code. Used in those sad and unfortunate cases where we need to call shitty external GL code that's not VAO-aware.
8 years ago
void MeshGLTest::bindScratchVaoWhenEnteringExternalSection() {
typedef Attribute<0, Float> Attribute;
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
const Float data[] = { -0.7f, Math::unpack<Float, UnsignedByte>(
#ifndef MAGNUM_TARGET_GLES2
92
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
85
#endif
), Math::unpack<Float, UnsignedByte>(32) };
GL: make it possible to bind a scratch VAO for external GL code. Used in those sad and unfortunate cases where we need to call shitty external GL code that's not VAO-aware.
8 years ago
Buffer buffer{Buffer::TargetHint::Array};
buffer.setData(data, BufferUsage::StaticDraw);
Buffer indices{Buffer::TargetHint::ElementArray};
indices.setData(std::vector<UnsignedByte>{5, 0}, BufferUsage::StaticDraw);
Mesh mesh;
mesh.addVertexBuffer(buffer, 4, Attribute{})
.setIndexBuffer(indices, 0, MeshIndexType::UnsignedByte);
{
GL: fix MeshGLTest::bindScratchVaoWhenEnteringExternalSection(). Would fail on an incomplete framebuffer error on framebuffer-less windowless contexts. Didn't happen before, I think it's because of a new check in a new Mesa.
7 years ago
/* Bind a scratch framebuffer so glDrawArrays() doesn't complain about
an incomplete framebuffer in case we're on a framebuffer-less
context */
Renderbuffer renderbuffer;
renderbuffer.setStorage(
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Vector2i{1});
Framebuffer framebuffer{{{}, Vector2i{1}}};
framebuffer.attachRenderbuffer(Framebuffer::ColorAttachment{0}, renderbuffer)
.bind();
GL: make it possible to bind a scratch VAO for external GL code. Used in those sad and unfortunate cases where we need to call shitty external GL code that's not VAO-aware.
8 years ago
/* Should bind a scratch VAO only on desktop with core profile and be
a no-op everywhere else */
Context::current().resetState(Context::State::EnterExternal
|Context::State::BindScratchVao /* Comment this out to watch the world burn */
);
/* Should throw no GL error if scratch VAO is bound */
glDrawArrays(GL_POINTS, 0, 0);
/* Be nice to the other tests */
Context::current().resetState(Context::State::ExitExternal);
}
MAGNUM_VERIFY_NO_GL_ERROR();
const auto value = Checker(FloatShader("float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"),
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
mesh).get<UnsignedByte>(PixelFormat::RGBA, PixelType::UnsignedByte);
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make it possible to bind a scratch VAO for external GL code. Used in those sad and unfortunate cases where we need to call shitty external GL code that's not VAO-aware.
8 years ago
CORRADE_COMPARE(value, 92);
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 85);
#endif
GL: make it possible to bind a scratch VAO for external GL code. Used in those sad and unfortunate cases where we need to call shitty external GL code that's not VAO-aware.
8 years ago
}
Added test for mesh base vertex.
12 years ago
#ifndef MAGNUM_TARGET_GLES
void MeshGLTest::setBaseVertex() {
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_elements_base_vertex>())
CORRADE_SKIP(Extensions::ARB::draw_elements_base_vertex::string() + std::string(" is not available."));
glDrawElementsBaseVertex() needs ARB_draw_elements_base_vertex.
12 years ago
Added test for mesh base vertex.
12 years ago
Buffer vertices;
Cleanup and code reorganization in MeshGLTest.
12 years ago
vertices.setData(indexedVertexDataBaseVertex, BufferUsage::StaticDraw);
Added test for mesh base vertex.
12 years ago
constexpr UnsignedShort indexData[] = { 2, 1, 0 };
Implemented indexed buffer binding. The indexed binding is allowed for only some types (atomic counters, uniforms, shader storage and transform feedback), thus we need separate enum for that. Because the bind() function will be used far more often than setTargetHint(), the original Target enum is now renamed to TargetHint and the new Target enum contains (in non-deprecated build) only three values. For backwards compatibility, though, we need to have all original Target values, thus the new Target enum contains also all other values from TargetHint, but they are marked as deprecated and (at least) run-time checked in bind() so they aren't accidentaly used for indexed binding. Similarly there are also deprecated Target overloads of Buffer() and setTargetHint(). It's ugly, but hopefully will suffice for now. This mess will be removed as soon as possible in some upcoming version.
12 years ago
Buffer indices{Buffer::TargetHint::ElementArray};
Added test for mesh base vertex.
12 years ago
indices.setData(indexData, BufferUsage::StaticDraw);
Mesh mesh;
Cleanup and code reorganization in MeshGLTest.
12 years ago
mesh.setBaseVertex(2)
.addVertexBuffer(vertices, 2*4, MultipleShader::Position(),
Added test for mesh base vertex.
12 years ago
MultipleShader::Normal(), MultipleShader::TextureCoordinates())
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
.setIndexBuffer(indices, 2, MeshIndexType::UnsignedShort);
Added test for mesh base vertex.
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added test for mesh base vertex.
12 years ago
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
const auto value = Checker(MultipleShader{}, RenderbufferFormat::RGBA8,
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Added test for mesh base vertex.
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added test for mesh base vertex.
12 years ago
CORRADE_COMPARE(value, indexedResult);
}
#endif
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
void MeshGLTest::setInstanceCount() {
/* Verbatim copy of addVertexBufferFloat() with added extension check and
setInstanceCount() call. It would just render three times the same
value. I'm too lazy to invent proper test case, so I'll just check that
it didn't generate any error and rendered something */
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_instanced>())
CORRADE_SKIP(Extensions::ARB::draw_instanced::string() + std::string(" is not available."));
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
#elif defined(MAGNUM_TARGET_GLES2)
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#ifndef MAGNUM_TARGET_WEBGL
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ANGLE::instanced_arrays>() &&
GL: wait, there's not just three ES instanced extensions, there's FIVE. Like, why can't they just depend on each other or something?
7 years ago
!Context::current().isExtensionSupported<Extensions::EXT::instanced_arrays>() &&
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
!Context::current().isExtensionSupported<Extensions::EXT::draw_instanced>() &&
GL: wait, there's not just three ES instanced extensions, there's FIVE. Like, why can't they just depend on each other or something?
7 years ago
!Context::current().isExtensionSupported<Extensions::NV::instanced_arrays>() &&
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
!Context::current().isExtensionSupported<Extensions::NV::draw_instanced>())
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
CORRADE_SKIP("Required extension is not available.");
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#else
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ANGLE::instanced_arrays>())
CORRADE_SKIP(Extensions::ANGLE::instanced_arrays::string() + std::string(" is not available."));
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#endif
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
#endif
Moved Attribute class out of AbstractShaderProgram enum. No backward compatibility issues should exist, as the class is in most (if not all) cases used with unscoped name: class MyShader: public AbstractShaderProgram { public: typedef Attribute<0, Vector3> Position; // ... };
12 years ago
typedef Attribute<0, Float> Attribute;
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
const Float data[] = { 0.0f, -0.7f, Math::unpack<Float, UnsignedByte>(96) };
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
mesh.setBaseVertex(1)
.setInstanceCount(3)
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
.addVertexBuffer(buffer, 4, Attribute());
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
const auto value = Checker(FloatShader("float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"),
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<UnsignedByte>(PixelFormat::RGBA, PixelType::UnsignedByte);
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
CORRADE_COMPARE(value, 96);
}
void MeshGLTest::setInstanceCountIndexed() {
/* Verbatim copy of setIndexBuffer() with added extension check and
setInstanceCount() call. It would just render three times the same
value. I'm too lazy to invent proper test case, so I'll just check that
it didn't generate any error and rendered something */
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_instanced>())
CORRADE_SKIP(Extensions::ARB::draw_instanced::string() + std::string(" is not available."));
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
#elif defined(MAGNUM_TARGET_GLES2)
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#ifndef MAGNUM_TARGET_WEBGL
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ANGLE::instanced_arrays>() &&
GL: wait, there's not just three ES instanced extensions, there's FIVE. Like, why can't they just depend on each other or something?
7 years ago
!Context::current().isExtensionSupported<Extensions::EXT::instanced_arrays>() &&
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
!Context::current().isExtensionSupported<Extensions::EXT::draw_instanced>() &&
GL: wait, there's not just three ES instanced extensions, there's FIVE. Like, why can't they just depend on each other or something?
7 years ago
!Context::current().isExtensionSupported<Extensions::NV::instanced_arrays>() &&
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
!Context::current().isExtensionSupported<Extensions::NV::draw_instanced>())
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
CORRADE_SKIP("Required extension is not available.");
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#else
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ANGLE::instanced_arrays>())
CORRADE_SKIP(Extensions::ANGLE::instanced_arrays::string() + std::string(" is not available."));
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#endif
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
#endif
Buffer vertices;
vertices.setData(indexedVertexData, BufferUsage::StaticDraw);
constexpr UnsignedShort indexData[] = { 2, 1, 0 };
Implemented indexed buffer binding. The indexed binding is allowed for only some types (atomic counters, uniforms, shader storage and transform feedback), thus we need separate enum for that. Because the bind() function will be used far more often than setTargetHint(), the original Target enum is now renamed to TargetHint and the new Target enum contains (in non-deprecated build) only three values. For backwards compatibility, though, we need to have all original Target values, thus the new Target enum contains also all other values from TargetHint, but they are marked as deprecated and (at least) run-time checked in bind() so they aren't accidentaly used for indexed binding. Similarly there are also deprecated Target overloads of Buffer() and setTargetHint(). It's ugly, but hopefully will suffice for now. This mess will be removed as soon as possible in some upcoming version.
12 years ago
Buffer indices{Buffer::TargetHint::ElementArray};
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
indices.setData(indexData, BufferUsage::StaticDraw);
Mesh mesh;
mesh.setInstanceCount(3)
.addVertexBuffer(vertices, 1*4, MultipleShader::Position(),
MultipleShader::Normal(), MultipleShader::TextureCoordinates())
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
.setIndexBuffer(indices, 2, MeshIndexType::UnsignedShort);
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
const auto value = Checker(MultipleShader{},
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
CORRADE_COMPARE(value, indexedResult);
}
#ifndef MAGNUM_TARGET_GLES
void MeshGLTest::setInstanceCountBaseInstance() {
/* Verbatim copy of setInstanceCount() with additional extension check and
setBaseInstance() call. It would just render three times the same
value. I'm too lazy to invent proper test case, so I'll just check that
it didn't generate any error and rendered something */
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_instanced>())
CORRADE_SKIP(Extensions::ARB::draw_instanced::string() + std::string(" is not available."));
if(!Context::current().isExtensionSupported<Extensions::ARB::base_instance>())
CORRADE_SKIP(Extensions::ARB::base_instance::string() + std::string(" is not available."));
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Moved Attribute class out of AbstractShaderProgram enum. No backward compatibility issues should exist, as the class is in most (if not all) cases used with unscoped name: class MyShader: public AbstractShaderProgram { public: typedef Attribute<0, Vector3> Position; // ... };
12 years ago
typedef Attribute<0, Float> Attribute;
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
const Float data[] = { 0.0f, -0.7f, Math::unpack<Float, UnsignedByte>(96) };
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
Removed redundant preprocessor branches from MeshGLTest. No need to have ES2 workarounds in code that is not compiled for ES2.
12 years ago
mesh.setBaseVertex(1)
.setInstanceCount(3)
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
.setBaseInstance(72)
.addVertexBuffer(buffer, 4, Attribute());
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
const auto value = Checker(FloatShader("float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"),
RenderbufferFormat::RGBA8,
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<UnsignedByte>(PixelFormat::RGBA, PixelType::UnsignedByte);
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
CORRADE_COMPARE(value, 96);
}
void MeshGLTest::setInstanceCountBaseInstanceIndexed() {
/* Verbatim copy of setInstanceCountIndexed() with additional extension
check and setBaseInstance() call. It would just render three times the
same value. I'm too lazy to invent proper test case, so I'll just check
that it didn't generate any error and rendered something */
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_instanced>())
CORRADE_SKIP(Extensions::ARB::draw_instanced::string() + std::string(" is not available."));
if(!Context::current().isExtensionSupported<Extensions::ARB::base_instance>())
CORRADE_SKIP(Extensions::ARB::base_instance::string() + std::string(" is not available."));
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Buffer vertices;
vertices.setData(indexedVertexData, BufferUsage::StaticDraw);
constexpr UnsignedShort indexData[] = { 2, 1, 0 };
Implemented indexed buffer binding. The indexed binding is allowed for only some types (atomic counters, uniforms, shader storage and transform feedback), thus we need separate enum for that. Because the bind() function will be used far more often than setTargetHint(), the original Target enum is now renamed to TargetHint and the new Target enum contains (in non-deprecated build) only three values. For backwards compatibility, though, we need to have all original Target values, thus the new Target enum contains also all other values from TargetHint, but they are marked as deprecated and (at least) run-time checked in bind() so they aren't accidentaly used for indexed binding. Similarly there are also deprecated Target overloads of Buffer() and setTargetHint(). It's ugly, but hopefully will suffice for now. This mess will be removed as soon as possible in some upcoming version.
12 years ago
Buffer indices{Buffer::TargetHint::ElementArray};
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
indices.setData(indexData, BufferUsage::StaticDraw);
Mesh mesh;
mesh.setInstanceCount(3)
.setBaseInstance(72)
.addVertexBuffer(vertices, 1*4, MultipleShader::Position(),
MultipleShader::Normal(), MultipleShader::TextureCoordinates())
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
.setIndexBuffer(indices, 2, MeshIndexType::UnsignedShort);
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
const auto value = Checker(MultipleShader{}, RenderbufferFormat::RGBA8,
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
CORRADE_COMPARE(value, indexedResult);
}
void MeshGLTest::setInstanceCountBaseVertex() {
/* Verbatim copy of setBaseVertex() with additional extension check and
setInstanceCount() call. It would just render three times the same
value. I'm too lazy to invent proper test case, so I'll just check
that it didn't generate any error and rendered something */
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_instanced>())
CORRADE_SKIP(Extensions::ARB::draw_instanced::string() + std::string(" is not available."));
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_elements_base_vertex>())
CORRADE_SKIP(Extensions::ARB::draw_elements_base_vertex::string() + std::string(" is not available."));
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Buffer vertices;
vertices.setData(indexedVertexDataBaseVertex, BufferUsage::StaticDraw);
constexpr UnsignedShort indexData[] = { 2, 1, 0 };
Implemented indexed buffer binding. The indexed binding is allowed for only some types (atomic counters, uniforms, shader storage and transform feedback), thus we need separate enum for that. Because the bind() function will be used far more often than setTargetHint(), the original Target enum is now renamed to TargetHint and the new Target enum contains (in non-deprecated build) only three values. For backwards compatibility, though, we need to have all original Target values, thus the new Target enum contains also all other values from TargetHint, but they are marked as deprecated and (at least) run-time checked in bind() so they aren't accidentaly used for indexed binding. Similarly there are also deprecated Target overloads of Buffer() and setTargetHint(). It's ugly, but hopefully will suffice for now. This mess will be removed as soon as possible in some upcoming version.
12 years ago
Buffer indices{Buffer::TargetHint::ElementArray};
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
indices.setData(indexData, BufferUsage::StaticDraw);
Mesh mesh;
mesh.setBaseVertex(2)
.setInstanceCount(3)
.addVertexBuffer(vertices, 2*4, MultipleShader::Position(),
MultipleShader::Normal(), MultipleShader::TextureCoordinates())
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
.setIndexBuffer(indices, 2, MeshIndexType::UnsignedShort);
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
const auto value = Checker(MultipleShader{}, RenderbufferFormat::RGBA8,
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
CORRADE_COMPARE(value, indexedResult);
}
void MeshGLTest::setInstanceCountBaseVertexBaseInstance() {
/* Verbatim copy of setInstanceCountBaseVertex() with added extension check
and setBaseInstance() call. It would just render three times the same
value. I'm too lazy to invent proper test case, so I'll just check
that it didn't generate any error and rendered something */
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_instanced>())
CORRADE_SKIP(Extensions::ARB::draw_instanced::string() + std::string(" is not available."));
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_elements_base_vertex>())
CORRADE_SKIP(Extensions::ARB::draw_elements_base_vertex::string() + std::string(" is not available."));
if(!Context::current().isExtensionSupported<Extensions::ARB::base_instance>())
CORRADE_SKIP(Extensions::ARB::base_instance::string() + std::string(" is not available."));
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Buffer vertices;
vertices.setData(indexedVertexDataBaseVertex, BufferUsage::StaticDraw);
constexpr UnsignedShort indexData[] = { 2, 1, 0 };
Implemented indexed buffer binding. The indexed binding is allowed for only some types (atomic counters, uniforms, shader storage and transform feedback), thus we need separate enum for that. Because the bind() function will be used far more often than setTargetHint(), the original Target enum is now renamed to TargetHint and the new Target enum contains (in non-deprecated build) only three values. For backwards compatibility, though, we need to have all original Target values, thus the new Target enum contains also all other values from TargetHint, but they are marked as deprecated and (at least) run-time checked in bind() so they aren't accidentaly used for indexed binding. Similarly there are also deprecated Target overloads of Buffer() and setTargetHint(). It's ugly, but hopefully will suffice for now. This mess will be removed as soon as possible in some upcoming version.
12 years ago
Buffer indices{Buffer::TargetHint::ElementArray};
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
indices.setData(indexData, BufferUsage::StaticDraw);
Mesh mesh;
mesh.setBaseVertex(2)
.setInstanceCount(3)
.setBaseInstance(72)
.addVertexBuffer(vertices, 2*4, MultipleShader::Position(),
MultipleShader::Normal(), MultipleShader::TextureCoordinates())
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
.setIndexBuffer(indices, 2, MeshIndexType::UnsignedShort);
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Removed duplicate code from MeshGLTest. As there is now no distinction between vertex and index count we don't need separate checker utility for non-indexed and indexed meshes.
12 years ago
const auto value = Checker(MultipleShader{}, RenderbufferFormat::RGBA8,
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Support for instanced mesh drawing and base instance. On desktop GL this is provided by ARB_draw_instanced (GL 3.1). Base instance is available only on desktop GL (4.2, ARB_base_instance). In ES2 the instanced functionality is provided by three (!) different extensions (ANGLE_instanced_arrays, EXT_draw_instanced, NV_draw_instanced), the proper implementation is chosen on context creation based on what extension is available. Though we don't have extension loader for ES yet, thus all these extensions are disabled and the implementation has assertion in it. Added blind test which tests only that something has been drawn and no errors were emitted, but not whether the right command is used. I'll probably need to check this later, because the Mesh::draw() behemoth is going slightly out of hand :)
12 years ago
CORRADE_COMPARE(value, indexedResult);
}
#endif
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
void MeshGLTest::addVertexBufferInstancedFloat() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_instanced>())
CORRADE_SKIP(Extensions::ARB::draw_instanced::string() + std::string(" is not available."));
if(!Context::current().isExtensionSupported<Extensions::ARB::instanced_arrays>())
CORRADE_SKIP(Extensions::ARB::instanced_arrays::string() + std::string(" is not available."));
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
#elif defined(MAGNUM_TARGET_GLES2)
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#ifndef MAGNUM_TARGET_WEBGL
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ANGLE::instanced_arrays>() &&
!Context::current().isExtensionSupported<Extensions::EXT::instanced_arrays>() &&
!Context::current().isExtensionSupported<Extensions::NV::instanced_arrays>())
GL: wait, there's not just three ES instanced extensions, there's FIVE. Like, why can't they just depend on each other or something?
7 years ago
CORRADE_SKIP("Required extension is not available.");
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#else
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ANGLE::instanced_arrays>())
CORRADE_SKIP(Extensions::ANGLE::instanced_arrays::string() + std::string(" is not available."));
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#endif
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
#endif
Moved Attribute class out of AbstractShaderProgram enum. No backward compatibility issues should exist, as the class is in most (if not all) cases used with unscoped name: class MyShader: public AbstractShaderProgram { public: typedef Attribute<0, Vector3> Position; // ... };
12 years ago
typedef Attribute<0, Float> Attribute;
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
const Float data[] = {
0.0f, /* Offset */
/* Base vertex is ignored for instanced arrays */
-0.7f, /* First instance */
0.3f, /* Second instance */
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Math::unpack<Float, UnsignedByte>(96) /* Third instance */
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
};
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
mesh.setInstanceCount(3)
.addVertexBufferInstanced(buffer, 1, 4, Attribute{});
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
const auto value = Checker(FloatShader("float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"),
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<UnsignedByte>(PixelFormat::RGBA, PixelType::UnsignedByte);
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
CORRADE_COMPARE(value, 96);
}
#ifndef MAGNUM_TARGET_GLES2
void MeshGLTest::addVertexBufferInstancedInteger() {
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_instanced>())
CORRADE_SKIP(Extensions::ARB::draw_instanced::string() + std::string(" is not available."));
if(!Context::current().isExtensionSupported<Extensions::ARB::instanced_arrays>())
CORRADE_SKIP(Extensions::ARB::instanced_arrays::string() + std::string(" is not available."));
if(!Context::current().isExtensionSupported<Extensions::EXT::gpu_shader4>())
CORRADE_SKIP(Extensions::EXT::gpu_shader4::string() + std::string(" is not available."));
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
#endif
Moved Attribute class out of AbstractShaderProgram enum. No backward compatibility issues should exist, as the class is in most (if not all) cases used with unscoped name: class MyShader: public AbstractShaderProgram { public: typedef Attribute<0, Vector3> Position; // ... };
12 years ago
typedef Attribute<0, UnsignedInt> Attribute;
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
constexpr UnsignedInt data[] = {
0, /* Offset */
/* Base vertex is ignored for instanced arrays */
157, /* First instance */
25, /* Second instance */
35681 /* Third instance */
};
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
mesh.setInstanceCount(3)
.addVertexBufferInstanced(buffer, 1, 4, Attribute{});
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
const auto value = Checker(IntegerShader("uint"), RenderbufferFormat::R32UI, mesh)
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
.get<UnsignedInt>(PixelFormat::RedInteger, PixelType::UnsignedInt);
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
CORRADE_COMPARE(value, 35681);
}
#endif
#ifndef MAGNUM_TARGET_GLES
void MeshGLTest::addVertexBufferInstancedDouble() {
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_instanced>())
CORRADE_SKIP(Extensions::ARB::draw_instanced::string() + std::string(" is not available."));
if(!Context::current().isExtensionSupported<Extensions::ARB::instanced_arrays>())
CORRADE_SKIP(Extensions::ARB::instanced_arrays::string() + std::string(" is not available."));
if(!Context::current().isExtensionSupported<Extensions::ARB::vertex_attrib_64bit>())
CORRADE_SKIP(Extensions::ARB::vertex_attrib_64bit::string() + std::string(" is not available."));
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
Moved Attribute class out of AbstractShaderProgram enum. No backward compatibility issues should exist, as the class is in most (if not all) cases used with unscoped name: class MyShader: public AbstractShaderProgram { public: typedef Attribute<0, Vector3> Position; // ... };
12 years ago
typedef Attribute<0, Double> Attribute;
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
const Double data[] = {
0.0, /* Offset */
/* Base vertex is ignored for instanced arrays */
-0.7, /* First instance */
0.3, /* Second instance */
Math: renamed {de,}normalize() to {,un}pack() in a new Packing.h header. To be more consistent with GLSL naming. Also, the original naming was quite misleading, as normalize() is used in GLSL for something completely different. If building with deprecated APIs, the Functions.h header includes the new Packing.h header and the {de,}normalize() functions are defined as deprecated aliases to the new functions. This will be removed at some point in the future.
9 years ago
Math::unpack<Double, UnsignedShort>(45828) /* Third instance */
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
};
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
mesh.setInstanceCount(3)
.addVertexBufferInstanced(buffer, 1, 8, Attribute{});
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
Fix type conversion errors in Mesh test shader code. Reported by latest NVidia drivers (352.09), apparently worked until now. Huh.
11 years ago
const auto value = Checker(DoubleShader("double", "float", "float(value)"),
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
RenderbufferFormat::R16, mesh).get<UnsignedShort>(PixelFormat::Red, PixelType::UnsignedShort);
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added Mesh::addVertexBufferInstanced(). Functionality provided by GL 3.3 and ARB_instanced_arrays, on ES2 this is again implemented in three different extensions -- {ANGLE,EXT,NV}_instanced_arrays. They are disabled until Magnum has proper extension loading on ES.
12 years ago
CORRADE_COMPARE(value, 45828);
}
#endif
GL: test that vertex attrib divisor is properly reset when not using VAOs. Fails, because it's not. Fix in next commit.
8 years ago
void MeshGLTest::resetDivisorAfterInstancedDraw() {
#ifndef MAGNUM_TARGET_GLES
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_instanced>())
CORRADE_SKIP(Extensions::ARB::draw_instanced::string() + std::string(" is not available."));
if(!Context::current().isExtensionSupported<Extensions::ARB::instanced_arrays>())
CORRADE_SKIP(Extensions::ARB::instanced_arrays::string() + std::string(" is not available."));
#elif defined(MAGNUM_TARGET_GLES2)
#ifndef MAGNUM_TARGET_WEBGL
if(!Context::current().isExtensionSupported<Extensions::ANGLE::instanced_arrays>() &&
!Context::current().isExtensionSupported<Extensions::EXT::instanced_arrays>() &&
!Context::current().isExtensionSupported<Extensions::NV::instanced_arrays>())
GL: wait, there's not just three ES instanced extensions, there's FIVE. Like, why can't they just depend on each other or something?
7 years ago
CORRADE_SKIP("Required extension is not available.");
GL: test that vertex attrib divisor is properly reset when not using VAOs. Fails, because it's not. Fix in next commit.
8 years ago
#else
if(!Context::current().isExtensionSupported<Extensions::ANGLE::instanced_arrays>())
CORRADE_SKIP(Extensions::ANGLE::instanced_arrays::string() + std::string(" is not available."));
#endif
#endif
/* This doesn't affect VAOs, because they encapsulate the state */
#ifndef MAGNUM_TARGET_GLES
if(Context::current().isExtensionSupported<Extensions::ARB::vertex_array_object>())
CORRADE_SKIP(Extensions::ARB::vertex_array_object::string() + std::string(" is enabled, can't test."));
#elif defined(MAGNUM_TARGET_GLES2)
if(Context::current().isExtensionSupported<Extensions::OES::vertex_array_object>())
CORRADE_SKIP(Extensions::OES::vertex_array_object::string() + std::string(" is enabled, can't test."));
#endif
typedef Attribute<0, Float> Attribute;
const Float data[]{
0,
Math::unpack<Float, UnsignedByte>(96),
Math::unpack<Float, UnsignedByte>(48),
};
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Renderbuffer renderbuffer;
renderbuffer.setStorage(
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Vector2i(1));
Framebuffer framebuffer{{{}, Vector2i(1)}};
framebuffer.attachRenderbuffer(Framebuffer::ColorAttachment(0), renderbuffer)
.bind();
FloatShader shader{"float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"};
MAGNUM_VERIFY_NO_GL_ERROR();
/* Draw instanced first. Two single-vertex instances of an attribute with
divisor 1, first draws 0, second draws 96 */
{
Mesh mesh;
mesh.setInstanceCount(2)
.addVertexBufferInstanced(buffer, 1, 0, Attribute{})
.setPrimitive(MeshPrimitive::Points)
.setCount(1)
.draw(shader);
MAGNUM_VERIFY_NO_GL_ERROR();
CORRADE_COMPARE(framebuffer.read({{}, Vector2i{1}}, {PixelFormat::RGBA, PixelType::UnsignedByte}).data<UnsignedByte>()[0], 96);
}
/* Draw normal after. One two-vertex instance of an attribute with divisor
0, first draws 96, second 48. In case divisor is not properly reset,
I'll get 96 on both. */
{
Mesh mesh;
mesh.setInstanceCount(1)
.addVertexBuffer(buffer, 4, Attribute{})
.setPrimitive(MeshPrimitive::Points)
.setCount(2)
.draw(shader);
MAGNUM_VERIFY_NO_GL_ERROR();
CORRADE_COMPARE(framebuffer.read({{}, Vector2i{1}}, {PixelFormat::RGBA, PixelType::UnsignedByte}).data<UnsignedByte>()[0], 48);
}
}
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
struct MultiChecker {
MultiChecker(AbstractShaderProgram&& shader, Mesh& mesh);
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
template<class T> T get(PixelFormat format, PixelType type);
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
Renderbuffer renderbuffer;
Framebuffer framebuffer;
};
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
#ifndef DOXYGEN_GENERATING_OUTPUT
MultiChecker::MultiChecker(AbstractShaderProgram&& shader, Mesh& mesh): framebuffer({{}, Vector2i(1)}) {
renderbuffer.setStorage(
#ifndef MAGNUM_TARGET_GLES2
RenderbufferFormat::RGBA8,
#else
RenderbufferFormat::RGBA4,
#endif
Vector2i(1));
framebuffer.attachRenderbuffer(Framebuffer::ColorAttachment(0), renderbuffer);
Simplified *Framebuffer::bind() functionality. This was a leftover from some not-well-thought-out design decision. The function is now used exclusively for binding for draw, as all framebuffer reading functions (blit(), read()) are doing the read binding internally. Moreover it required the user to be extra careful on ES2, because in many cases there are no separate binding points for reading and drawing. The function is now parameter-less and always bind the framebuffer for drawing. The logic for internal binding was also simplified and on ES2 there are separate implementations for single/separate binding points. For *Framebuffer::checkStatus() the documentation was updated to explain the meaning of the parameter on ES2 implementation. Also removed the need for FramebufferTarget::ReadDraw binding, as it was rather confusing. Old *Framebuffer::bind(FramebufferTarget) is now just an alias to the parameter-less function, ignoring the parameter. Along with FramebufferTarget::ReadDraw it is marked as deprecated and will be removed in some future release.
11 years ago
framebuffer.bind();
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
mesh.setPrimitive(MeshPrimitive::Points)
.setCount(2);
Check for mesh emptiness even before binding the shader. Prevents unnecessary state touching.
10 years ago
/* Set zero count so we test mesh skipping */
MeshView a{mesh};
a.setCount(0);
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
/* Skip first vertex so we test also offsets */
Check for mesh emptiness even before binding the shader. Prevents unnecessary state touching.
10 years ago
MeshView b{mesh};
b.setCount(1)
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
.setBaseVertex(mesh.baseVertex());
Check for mesh emptiness even before binding the shader. Prevents unnecessary state touching.
10 years ago
MeshView c{mesh};
c.setCount(1);
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
if(mesh.isIndexed()) {
Check for mesh emptiness even before binding the shader. Prevents unnecessary state touching.
10 years ago
c.setBaseVertex(mesh.baseVertex())
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
.setIndexRange(1);
Check for mesh emptiness even before binding the shader. Prevents unnecessary state touching.
10 years ago
} else c.setBaseVertex(1);
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
Check for mesh emptiness even before binding the shader. Prevents unnecessary state touching.
10 years ago
MeshView::draw(shader, {a, b, c});
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
}
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
template<class T> T MultiChecker::get(PixelFormat format, PixelType type) {
Making use of the shorter *Framebuffer::read(). I like that!
11 years ago
return framebuffer.read({{}, Vector2i{1}}, {format, type}).data<T>()[0];
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
}
#endif
void MeshGLTest::multiDraw() {
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#if defined(MAGNUM_TARGET_GLES) && !defined(MAGNUM_TARGET_WEBGL)
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::EXT::multi_draw_arrays>())
Debug() << Extensions::EXT::multi_draw_arrays::string() << "not supported, using fallback implementation";
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
#endif
Moved Attribute class out of AbstractShaderProgram enum. No backward compatibility issues should exist, as the class is in most (if not all) cases used with unscoped name: class MyShader: public AbstractShaderProgram { public: typedef Attribute<0, Vector3> Position; // ... };
12 years ago
typedef Attribute<0, Float> Attribute;
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
const Float data[] = { 0.0f, -0.7f, Math::unpack<Float, UnsignedByte>(
#ifndef MAGNUM_TARGET_GLES2
96
#else
/* Using only RGBA4, supply less precision. This has to be one on the
input because SwiftShader stores RGBA4 as RGBA8 internally, thus
preserving the full precision of the input. */
85
#endif
) };
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
Buffer buffer;
buffer.setData(data, BufferUsage::StaticDraw);
Mesh mesh;
mesh.addVertexBuffer(buffer, 4, Attribute());
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
const auto value = MultiChecker(FloatShader("float", "vec4(valueInterpolated, 0.0, 0.0, 0.0)"),
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
mesh).get<UnsignedByte>(PixelFormat::RGBA, PixelType::UnsignedByte);
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
CORRADE_COMPARE(value, 96);
GL: expect less precision on ES2 tests that use RGBA4 renderbuffers.
7 years ago
#else /* RGBA4, so less precision */
CORRADE_COMPARE(value, 85);
#endif
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
}
void MeshGLTest::multiDrawIndexed() {
Test: make the GL tests compile on WebGL 1. Not WebGL 2 yet, though.
8 years ago
#if defined(MAGNUM_TARGET_GLES) && !defined(MAGNUM_TARGET_WEBGL)
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::EXT::multi_draw_arrays>())
Debug() << Extensions::EXT::multi_draw_arrays::string() << "not supported, using fallback implementation";
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
#endif
Buffer vertices;
vertices.setData(indexedVertexData, BufferUsage::StaticDraw);
constexpr UnsignedShort indexData[] = { 2, 1, 0 };
Implemented indexed buffer binding. The indexed binding is allowed for only some types (atomic counters, uniforms, shader storage and transform feedback), thus we need separate enum for that. Because the bind() function will be used far more often than setTargetHint(), the original Target enum is now renamed to TargetHint and the new Target enum contains (in non-deprecated build) only three values. For backwards compatibility, though, we need to have all original Target values, thus the new Target enum contains also all other values from TargetHint, but they are marked as deprecated and (at least) run-time checked in bind() so they aren't accidentaly used for indexed binding. Similarly there are also deprecated Target overloads of Buffer() and setTargetHint(). It's ugly, but hopefully will suffice for now. This mess will be removed as soon as possible in some upcoming version.
12 years ago
Buffer indices{Buffer::TargetHint::ElementArray};
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
indices.setData(indexData, BufferUsage::StaticDraw);
Mesh mesh;
mesh.addVertexBuffer(vertices, 1*4, MultipleShader::Position(),
MultipleShader::Normal(), MultipleShader::TextureCoordinates())
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
.setIndexBuffer(indices, 2, MeshIndexType::UnsignedShort);
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
const auto value = MultiChecker(MultipleShader{}, mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
CORRADE_COMPARE(value, indexedResult);
}
#ifndef MAGNUM_TARGET_GLES
void MeshGLTest::multiDrawBaseVertex() {
Split the OpenGL layer out, pt 16: {Extensions::GL => GL::Extensions}. Was Magnum::GL::Extensions::GL before and the redundancy was completely unnecessary. Potential future extensions coming from GLX, EGL or whatnot will most probably be in the Platform namespace in a completely separate file, so this is not a problem. All code internal to the GL library is affected, not much the outside, as that is handled by the compatibility alias.
8 years ago
if(!Context::current().isExtensionSupported<Extensions::ARB::draw_elements_base_vertex>())
CORRADE_SKIP(Extensions::ARB::draw_elements_base_vertex::string() + std::string(" is not available."));
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
Buffer vertices;
vertices.setData(indexedVertexDataBaseVertex, BufferUsage::StaticDraw);
constexpr UnsignedShort indexData[] = { 2, 1, 0 };
Implemented indexed buffer binding. The indexed binding is allowed for only some types (atomic counters, uniforms, shader storage and transform feedback), thus we need separate enum for that. Because the bind() function will be used far more often than setTargetHint(), the original Target enum is now renamed to TargetHint and the new Target enum contains (in non-deprecated build) only three values. For backwards compatibility, though, we need to have all original Target values, thus the new Target enum contains also all other values from TargetHint, but they are marked as deprecated and (at least) run-time checked in bind() so they aren't accidentaly used for indexed binding. Similarly there are also deprecated Target overloads of Buffer() and setTargetHint(). It's ugly, but hopefully will suffice for now. This mess will be removed as soon as possible in some upcoming version.
12 years ago
Buffer indices{Buffer::TargetHint::ElementArray};
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
indices.setData(indexData, BufferUsage::StaticDraw);
Mesh mesh;
mesh.setBaseVertex(2)
.addVertexBuffer(vertices, 2*4, MultipleShader::Position(),
MultipleShader::Normal(), MultipleShader::TextureCoordinates())
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
.setIndexBuffer(indices, 2, MeshIndexType::UnsignedShort);
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
Pixel storage support, part 2: renamed ColorFormat to PixelFormat etc. Yeah, sorry, I know, the enums are renamed for second or third time in a row, first they were Image::Format, then ImageFormat, then ColorFormat and now PixelFormat. But this time it's final and last time they are renamed and now everything is finally consistent: * ColorFormat::DepthComponent -- depth is not a color, thus PixelFormat::DepthComponent makes a lot more sense. * There will be PixelStorage classes, which will be stored in images alonside PixelFormat/PixelType enums, making everything nicely aligned. * The GL documentation about glTexImage2D() etc. denotes the <format> and <type> parameters as format and type of *pixel* data, so now we are _finally_ consistent with the official naming. I wonder why did I not choose PixelFormat originally. Anyway, the old <Magnum/ColorFormat.h> header, ColorFormat, ColorType and CompressedColorFormat types are now aliases to the new ones, are marked as deprecated and will be removed in some future release (as always, I'm waiting at least six months before removing the deprecated functionality).
11 years ago
const auto value = MultiChecker(MultipleShader{}, mesh).get<Color4ub>(PixelFormat::RGBA, PixelType::UnsignedByte);
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
Split the OpenGL layer out, pt 17: adapted GL tests. Now the library compiles together with tests w/o deprecation warnings ... except for documentation snippets.
8 years ago
MAGNUM_VERIFY_NO_GL_ERROR();
Added support for glMultiDrawArrays()/glMultiDrawElements(). In OpenGL ES this is implemented in EXT_multi_draw_arrays extension, if it is not available, the functionality is emulated using sequence of normal draw() calls.
12 years ago
CORRADE_COMPARE(value, indexedResult);
}
#endif
Wrap all tests in unnamed namespaces. This makes the compiler warn about test cases that aren't ever used, which is a good thing.
7 years ago
}}}}
Support for object labels from KHR_debug and EXT_debug_label. Can't test EXT_debug_label, as that is apparently OSX 10.9-only. Added GL tests for all implemented objects. KHR_debug is selected first, if that is not available, fall back to EXT_debug_label. If neither is available, the functions are no-op. I hope EXT_debug_label gets replaced by KHR_debug later, thus it is now only "emulated" through KHR_debug enums.
13 years ago
Split the OpenGL layer out, pt 11: generic MeshPrimitive and MeshIndexType. Similarly to pixel formats, there is now generic Magnum::MeshPrimitive and Magnum::MeshIndexType, which is convertible to GL::MeshPrimitive and GL::MeshIndexType using GL::meshPrimitive() and GL::meshIndexType(). In addition, the following is done: * The original GL::Mesh::IndexType is now GL::MeshIndexType, original name is now just a typedef. * GL::Mesh::indexSize() is deprecated in favor of Magnum::meshIndexTypeSize() and GL::Mesh::indexTypeSize(). * New GL::Mesh::indexType() and GL::MeshView::mesh() getters (not sure why they were omitted) * GL::Mesh::indexType(), GL::Mesh::indexTypeSize(), GL::MeshView::setIndexRange() now expect that the mesh is indexed (useful property in my opinion, also avoids getting random results). * The extra MeshPrimitive::LinesAdjacency etc. are still present for backwards compatibility, but marked as deprecated. Use GL::MeshPrimitive values instead.
8 years ago
CORRADE_TEST_MAIN(Magnum::GL::Test::MeshGLTest)