Will need those in the Ui library, in a form that's quite restricted
compared to what the generic shader provides (2D only, uniform
buffers always, ...). Plus it doesn't make sense to make the Ui library
depend on Shaders, so it would be just these two files copied verbatim
between the two.
Compared to Corrade, the improvement in compile time is about a minute
cumulative across all cores, or about 8 seconds on an 8-core system (~2
minutes before, ~1:52 after). Not bad at all. And this is with a
deprecated build, the non-deprecated build is 1:48 -> 1:41.
This caused MeshToolsCompileGLTest to fail in a strange way, and
PhongGLTest::renderLowLightAngle() as well. Which looked rare enough that
I first suspected some random driver bug, but apparently it was all
caused by these using the default infinity range instead of explicitly
calling setLightRanges() on the shader.
The test is now updated to explicitly verify the default value when a
setter isn't called, to catch this problem better in case it reappears in
a different form elsewhere.
Along with the bits in Text library this is one of the last things that
still assume OpenGL present by default.
As usual, the old name and header is now a deprecated typedef.
The output is now *much* closer to the Vector shader output (mean
difference went from ~6 to 1.79), and in the multi-draw case it's
clearly visible that it's no longer slightly weirdly uneven.
The test files are now also much smaller as the originals were created
before RLE was implemented in TgaImageConverter.
Not that C++ STL and exceptions would be anything to take inspiration
from, but there's std::out_of_range. Python IndexError is also specified
as "index out of range", not "bounds".
Partially needed to avoid build breakages because Corrade itself
switched as well, partially because a cleanup is always good. Done
except for (STL-heavy) code that's deprecated or SceneGraph-related APIs
that are still quite full of STL as well.
Those caused a warning to be printed to console during construction on
platforms without explicit uniform locations, and setting them with an
explicit location apparently causes a GL error on Qualcomm Adreno. They
*are* there in the input source, just DCE'd.
I won't argue whether that's a valid driver behavior or not. Just simply
not doing this anymore and silently skipping those uniforms instead.
There's still a TODO to actually support those properly, i.e. to be able
to change wireframe width or smoothness on a GS-less rendering. Then
they won't be DCE'd and this workaround wouldn't be needed anymore.
With the workarounds moved to the GL::Shader class itself, it's just a
complicated wrapper for adding the compatibility.glsl file and a rather
strange way to define a file-local helper for resource import on static
builds. Do that directly instead.
As this is now documented, it means 3rd party code can now directly make
use of these without having to reinvent the same logic, or worse,
rediscover the same driver bugs.
The compatibility.glsl file however stays private -- I don't expect
real-world projects needing *that much* diversity in their supported
GLSL versions, often the baseline is GLES 3.0 which makes a large part
of the file unnecessary, and the projects might choose to for example
always have implicitly queried uniform locations to not have to
maintain two code paths.
Interesting, didn't know this kind of feature support was possible. I
guess it's still better than no GLES3.2 at all. Also, it's a phone from
2017, so probably not all that important to care about anymore anyway.
Most of the testing scaffolding here is a preparation for the actually
complex formats like BC6/7 or ASTC. Also, it's great to be able to use
Magnum from Python to prepare data for testing the C++ Magnum APIs.
The perf cost is just too great for these to be enabled always. The only
place where the assertions are kept always is in the batch APIs -- there
it's assumed the function is called on large enough data to offset this
overhead, plus since it's often dealing with large blocks of data the
memory safety is more important than various FP drifts which were the
usual case why other assertions were firing.
The UBO definitons are not tied to GL in any way, so they should be
available always. MeshTools::generateLine() now is as well, and this
made the tests fail to link on ES2 builds due to the debug operator not
being present.
This is unfortunately a breaking change to compileLines(), which now
takes the output of generateLines() instead of a line mesh. There's a
new assertion that'll blow up if the code is used the previous way,
sorry for the breakage.
What's however very useful about this change is that now it's possible
to take those generated line meshes and concatenate() them together,
achieving what's otherwise not implemented yet, such as drawing several
disconnected line strips or loops together.
It's all still partially private (the custom mesh attribute names are)
and I'm marking both APIs as experimental now to hint that it's not in
the final form/functionality yet. In particular, the data layout
optimizations described in the shader docs aren't used by these tools
yet, and if/once the line-specific vertex attributes become builtin,
compileLines() will not need to exist anymore as compile() will handle
that directly.
Done this way only in the Phong shader, everywhere else it's just the
MAGNUM_ASSERT_GL_EXTENSION_SUPPORTED() macro. Some WIP code that I
forgot to clean up?
Vladimír Vondruš