The documentation of ARB_invalidate_subdata mentions that all the
functions are really just a hint for the implementation to make some
performance optimizations and they are not affecting behavior at all. So
it's perfectly fine to do nothing if the extension is not supported.
I didn't do this originally as I mistakenly thought that invalidating
depth buffer would somehow behave the same as clearing it, but that's
not the case.
In most cases the label is set directly from code, e.g.:
texture.setLabel("diffuse-duck");
Avoiding conversion to std::string and passing char(&)[size] directly
will avoid one allocation and deallocation. Better solution would be to
use std::string_view everywhere, but we're not in C++17 yet.
The previous way was half-working at best, as it handled array textures
improperly. Now there is overload for each texture type. The old way
with attachTexture*D() is marked as deprecated and will be removed in
future release.
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/)
Everything what was in src/ is now in src/Corrade, everything from
src/Plugins is now in src/MagnumPlugins, everything from external/ is in
src/MagnumExternal. Added new CMakeLists.txt file and updated the other
ones for the moves, no other change was made. If MAGNUM_BUILD_DEPRECATED
is set, everything compiles and installs like previously except for the
plugins, which are now in MagnumPlugins and not in Magnum/Plugins.
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.
Encourages vectorization and generic usage even more. Some functions
were rewritten to make use of the new features, resulting in shorter and
more readable code. This also fixes the annoying naming collision with
WINAPI Rectangle() function.
The old Rectangle is now subclass of Range2D, is marked as deprecated
and will be removed in future release.
Passing pointer as function parameter will now mean that it is possible
to pass `nullptr`. Some code examples now look like the parameter is
copied instead of referenced, which is misleading. Updated the
documentation to reflect that more clearly.
Makes some cases less consistent (and some convenience shortcuts
impossible), but goes well with the attitude "don't use pointer when it
can't be null".
Separated EXT_framebuffer_object, EXT_framebuffer_blit,
EXT_framebuffer_multisample and EXT_packed_depth_stencil don't have the
same functionality as ARB_framebuffer_object (e.g. missing
GL_FRAMEBUFFER_UNDEFINED in glCheckFramebufferStatus()) and separated
read/draw binding is only in EXT_framebuffer_blit, which complicates the
internals.
Checked with Mesa 8/9 and OpenGL 2.1, current one has
ARB_framebuffer_object and also all these four, Mesa 7.7 didn't have
EXT_framebuffer_multisample, but that's a long time ago, so not
supporting these separate extensions shouldn't be an issue.
The problem with unavailable separate binding points remains on OpenGL
ES 2.0, there are three different extensions bringing that
functionality, thus the code managing the available binding points
remains there.
Instead of having pair of functions for depth/stencil and color
attachments this is now solved using specialized class. It is now more
descriptive, as specifying color attachment is now done explicitly using
Framebuffer::ColorAttachment(2) instead of just `2`. Specifying "no
attachment" is now also done explicitly using
Framebuffer::DrawAttachment::None instead of `-1`.
Also updated related documentation.
* Calling *Mesh::draw() with parameter to start on arbitrary vertex
index might give users more freedom than they want to have (e.g.
lines rendered where gaps should be, broken triangle strips...).
* Single-precision floats have meaningful precision of ~6 decimal
places, everything after that would be random garbage anyway, so we
don't need anything "more precise" for icosphere.
* Texture1D can have only one target and it can be used as framebuffer
target.
Framebuffer::attach*() doesn't need the target at all (meaning the
attachment will be used for both reading and drawing), another
misunderstanding on my side.
Now the extension is used on all places where it can be used (except for
unimplemented features).
It prevents unwanted implicit conversions from e.g. nullptr to Camera,
Vector2 to Physics::Point etc. By making all the constructors explicit
it is easier to routinely add the keyword to all new classes instead of
thinking about cases when to add and when not to.
Viewport position and size is managed separately for each framebuffer
and glViewport() is called in bind() (and also from setViewport(), if
the framebuffer is currently bound) if the viewport differs from
current state. If used only one framebuffer size through whole
application lifetime, glViewport() doesn't need to be called at all.
Default framebuffer is now accessible throught defaultFramebuffer global
variable, named framebuffers are handled the same way as before. All
operations (clear, setViewport, blit, read) now are member functions
so they cannot be mistakenly used when unwanted framebuffer is bound.
Further rework (DSA, state tracking...) is on the way.
Vladimír Vondruš