As we now have 100 tests (yay!), ctest needs three more characters to
display the progress, thus we need to shorten the long test names a bit
more. Should suffice for a long time, I'm not planning to have >999
tests :-)
Caused linker problems mainly on Windows, where every symbol would need
to be exported manually, on Linux this was done implicitly (or with
`-fvisibility=default`). I now maintain glLoadGen fork at
https://github.com/mosra/glloadgen.git containing changes needed for
Magnum.
The `ogl_*()` functions are used only internally, various global
variables for extension queries are not used at all and thus they don't
need to be exported. I thus enabled `-fvisibility=hidden`.
Fixes#16.
Resolves some linker errors in Windows, removes code bloat from Platform
namespace (previously it was copied in _every_ *Application).
Originally this was in Platform namespace because I thought that it
would be possible to have one version of Magnum library for both desktop
and ES and the actual Application would determine which GL edition will
be used. Fun idea, but it would remove the static checks (e.g.
accidentaly using geometry shaders on ES), which isn't worth it.
It's counterintuitive when you have to enable `WITH_GL_TESTS` just to
install the header, moreover the dependent projects have no easy way to
detect the header presence, so they just fail on preprocessor error.
Removed all known GLEW workarounds, added one small workaround for
missing ARB_texture_compression_bptc. I didn't want to patch glLoadGen
for just four enum values, this way it's possible to use stock one
without any patching (except for missing OpenGL 2.1 support, as stated
in external/OpenGL/GL/README.md).
As one file now replaces both `glew.h` and `glcorearb.h` and it has the
same size as `glcorearb.h` alone , it saves approximately 18k LOC,
resulting in 15 second shorter compilation time (5:03 before, 4:48 now).
Not bad.
It contains a few (un)packing functions present in ARB_gpu_shader5, but
isn't mentioned anywhere except GLSL 4.20 changelog (and there it has
its old name, ARB_shading_language_pack2f). It might be supported on
machines where ARB_gpu_shader5 can't be supported (GL3 HW).
Now named MaterialType. It's easier to type this:
auto t = Trade::MaterialType::Phong;
Than this overlong name, which is moreover ambiguous due to subclassing:
auto t = Trade::AbstractMaterialData::Type::Phong;
// or should I use rather Trade::PhongMaterialData::Type::Phong?
This requires inclusion of heavyweight <memory> header (~25k LOC), but
data import isn't something the user would want to do in every file, so
it hopefully won't hurt compilation times too much.
Saves unnecessary heap allocation and saves user from the burden of
explicit deletion, otherwise the usage is the same. Polymorphic types
can't be done using `std::optional`, will fix that later.
We would need to duplicate all the functionality found in Vector, which
I don't think is needed at all. If anyone needs to do this, it is
possible to "linearize" the matrix into long vector and do the
operations on it.
Also updated subclass operator implementation, added tests for it, both
for proper returned value and proper result type.
Operators that are part of Vector are operating only with the same type
as Vector itself, operators for multiplying/dividing integral vectors
with floating-point numbers and vectors are now out-of-class and enabled
only for integer vectors. It allows better control (e.g. multiplying
integer and floating-point vector will _always_ result in floating-point
one). Thoroughly tested integer/FP operations and also reworked and
tested operator and funciton reimplementations in subclasses, both for
value correctness and result type correctness.
Found even more stuff broken -- copied contents of
Corrade::Utility::Directory::Flags into Magnum::Context::Flags (WTF!),
asterisks in code listings... IS IT SO HARD TO FIX ONE THING AND DON'T
BREAK THOUSAND OTHERS, DAMMIT??
It is possible to do everything with Mesh::addInterleavedVertexBuffer().
Moreover Mesh::addVertexBuffer() was dependent on vertex count, which
was counterintuitive and not always what the user wants. Also in many
times I mistakenly used Mesh::addVertexBuffer() instead of
Mesh::addInterleavedVertexBuffer() and then spent endless hours trying
to figure out what is wrong. This is now over. Thanks, brain!
On systems without VAOs and without gl_VertexID the buffer needs to be
bound before every draw call. This way the buffer reference would become
dangling one, because the buffer is moved out somewhere else. Now
creating it on heap and returning pointer instead.
This multiple code path implementation is really starting to pay out,
yay!
Vladimír Vondruš