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.
8847 Commits
21 Branches
19 Tags
70 MiB
Tree: 6958f4a413
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from '6958f4a413'
${ noResults }
magnum/src/Magnum/GL/Context.cpp

1330 lines
52 KiB
Raw Normal View History Unescape

Classes for handling version and extension information.
14 years ago
/*
This file is part of Magnum.
Updated copyright year.
6 years ago
Copyright © 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019,
Updated copyright year.
5 years ago
2020, 2021 Vladimír Vondruš <mosra@centrum.cz>
Relicensing to MIT/Expat license, part 3: source files. Added license header also to shader sources. Hopefully it won't harm compilation times too much.
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.
Classes for handling version and extension information.
14 years ago
*/
#include "Context.h"
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#include <algorithm>
Debug operator for Context::Flags.
9 years ago
#include <Corrade/Containers/EnumSet.hpp>
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
#include <Corrade/Containers/GrowableArray.h>
Initial support for passing arguments directly to the engine. The parameterless Platform::Context::Context() constructor and Platform::Context::tryCreate() function are deprecated in favor functions that take argc/argv pair. The Context class now accepts arguments starting with --magnum-* prefix. Currently there are none except for the implicit --magnum-help, but that will change with the following commits.
11 years ago
#include <Corrade/Utility/Arguments.h>
Making includes absolute, part 4: using absolute includes for Corrade. Also adapted to some Corrade renames.
12 years ago
#include <Corrade/Utility/Debug.h>
Classes for handling version and extension information.
14 years ago
Split the OpenGL layer out, pt 1: moving files to a new location. Minimal updates (just the include guards) so Git is hopefully able to detect the rename and track the history properly. Everything except Magnum::GL doesn't compile now.
8 years ago
#include "Magnum/GL/AbstractFramebuffer.h"
#include "Magnum/GL/AbstractShaderProgram.h"
#include "Magnum/GL/AbstractTexture.h"
#include "Magnum/GL/Buffer.h"
Fixed OpenGL ES 2.0 and 3.0 build.
12 years ago
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 1: moving files to a new location. Minimal updates (just the include guards) so Git is hopefully able to detect the rename and track the history properly. Everything except Magnum::GL doesn't compile now.
8 years ago
#include "Magnum/GL/BufferTexture.h"
Fixed OpenGL ES 2.0 and 3.0 build.
12 years ago
#endif
GL: new --magnum-gpu-validation command-line option. Enables KHR_debug, if available. This should have been here since ages.
7 years ago
#ifndef MAGNUM_TARGET_WEBGL
#include "Magnum/GL/DebugOutput.h"
#endif
Split the OpenGL layer out, pt 1: moving files to a new location. Minimal updates (just the include guards) so Git is hopefully able to detect the rename and track the history properly. Everything except Magnum::GL doesn't compile now.
8 years ago
#include "Magnum/GL/DefaultFramebuffer.h"
#include "Magnum/GL/Extensions.h"
#include "Magnum/GL/Framebuffer.h"
#include "Magnum/GL/Mesh.h"
#include "Magnum/GL/Renderbuffer.h"
#include "Magnum/GL/Renderer.h"
#include "Magnum/GL/Implementation/State.h"
#include "Magnum/GL/Implementation/ContextState.h"
#include "Magnum/GL/Implementation/BufferState.h"
#include "Magnum/GL/Implementation/FramebufferState.h"
#include "Magnum/GL/Implementation/MeshState.h"
#include "Magnum/GL/Implementation/RendererState.h"
#include "Magnum/GL/Implementation/ShaderProgramState.h"
#include "Magnum/GL/Implementation/TextureState.h"
Print error if transform feedback header is used in OpenGL ES 2.0 build.
11 years ago
#ifndef MAGNUM_TARGET_GLES2
Split the OpenGL layer out, pt 1: moving files to a new location. Minimal updates (just the include guards) so Git is hopefully able to detect the rename and track the history properly. Everything except Magnum::GL doesn't compile now.
8 years ago
#include "Magnum/GL/Implementation/TransformFeedbackState.h"
Print error if transform feedback header is used in OpenGL ES 2.0 build.
11 years ago
#endif
Classes for handling version and extension information.
14 years ago
Introduce a BUILD_STATIC_UNIQUE_GLOBALS option. Same as in Corrade. Because BUILD_STATIC is independent between Corrade and Magnum this option is also independent -- the corner cases and bad interactions would be otherwise too complex to handle (e.g., in case of a dynamic Corrade and static Magnum it would be impossible to enable this option for Magnum etc etc).
6 years ago
#if defined(CORRADE_TARGET_WINDOWS) && defined(MAGNUM_BUILD_STATIC_UNIQUE_GLOBALS) && !defined(CORRADE_TARGET_WINDOWS_RT)
Implement globals shared across DLLs on Windows.
7 years ago
#include "Magnum/Implementation/WindowsWeakSymbol.h"
#endif
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
namespace Magnum { namespace GL {
Classes for handling version and extension information.
14 years ago
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
using namespace Containers::Literals;
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
/* When adding a new list, Extension::extensions() and Context::Context() needs
to be adapted. Binary search is performed on the extensions, thus they have
to be sorted alphabetically. */
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
namespace {
#ifndef MAGNUM_TARGET_GLES
constexpr Extension ExtensionList[]{
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::AMD::sample_positions{},
Extensions::AMD::shader_explicit_vertex_parameter{},
Extensions::AMD::shader_trinary_minmax{},
Extensions::AMD::transform_feedback3_lines_triangles{},
Extensions::AMD::vertex_shader_layer{},
Extensions::ARB::ES3_2_compatibility{},
Extensions::ARB::bindless_texture{},
Extensions::ARB::compute_variable_group_size{},
Extensions::ARB::robustness{},
Extensions::ARB::robustness_application_isolation{},
Extensions::ARB::robustness_isolation{},
Extensions::ARB::robustness_share_group_isolation{},
Extensions::ARB::sample_locations{},
Extensions::ARB::seamless_cubemap_per_texture{},
Extensions::ARB::sparse_buffer{},
Extensions::ARB::sparse_texture{},
Extensions::ATI::texture_mirror_once{},
Extensions::EXT::debug_label{},
Extensions::EXT::debug_marker{},
Extensions::EXT::shader_integer_mix{},
Extensions::EXT::texture_compression_dxt1{},
Extensions::EXT::texture_compression_s3tc{},
Extensions::EXT::texture_filter_anisotropic{},
Extensions::EXT::texture_mirror_clamp{},
Extensions::EXT::texture_sRGB_R8{},
Extensions::EXT::texture_sRGB_decode{},
Extensions::GREMEDY::string_marker{},
Extensions::KHR::blend_equation_advanced{},
Extensions::KHR::blend_equation_advanced_coherent{},
Extensions::KHR::texture_compression_astc_hdr{},
Extensions::KHR::texture_compression_astc_ldr{},
Extensions::KHR::texture_compression_astc_sliced_3d{},
Extensions::NV::fragment_shader_barycentric{},
Extensions::NV::sample_locations{},
Extensions::OVR::multiview{},
Extensions::OVR::multiview2{}
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionList300[]{
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ARB::color_buffer_float{},
Extensions::ARB::depth_buffer_float{},
Extensions::ARB::framebuffer_object{},
Extensions::ARB::framebuffer_sRGB{},
Extensions::ARB::half_float_pixel{},
Extensions::ARB::half_float_vertex{},
Extensions::ARB::map_buffer_range{},
Extensions::ARB::texture_float{},
Extensions::ARB::texture_rg{},
Extensions::ARB::vertex_array_object{},
Extensions::EXT::draw_buffers2{},
Extensions::EXT::gpu_shader4{},
Extensions::EXT::packed_float{},
Extensions::EXT::texture_array{},
Extensions::EXT::texture_compression_rgtc{},
Extensions::EXT::texture_integer{},
Extensions::EXT::texture_shared_exponent{},
Extensions::EXT::transform_feedback{},
Extensions::MAGNUM::shader_vertex_id{},
Extensions::NV::conditional_render{},
Extensions::NV::depth_buffer_float{}
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionList310[]{
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ARB::copy_buffer{},
Extensions::ARB::draw_instanced{},
Extensions::ARB::texture_buffer_object{},
Extensions::ARB::texture_rectangle{},
Extensions::ARB::uniform_buffer_object{},
Extensions::EXT::texture_snorm{},
Extensions::NV::primitive_restart{}
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionList320[]{
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ARB::depth_clamp{},
Extensions::ARB::draw_elements_base_vertex{},
Extensions::ARB::fragment_coord_conventions{},
Extensions::ARB::geometry_shader4{},
Extensions::ARB::provoking_vertex{},
Extensions::ARB::seamless_cube_map{},
Extensions::ARB::sync{},
Extensions::ARB::texture_multisample{},
Extensions::ARB::vertex_array_bgra{}
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionList330[]{
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ARB::blend_func_extended{},
Extensions::ARB::explicit_attrib_location{},
Extensions::ARB::instanced_arrays{},
Extensions::ARB::occlusion_query2{},
Extensions::ARB::sampler_objects{},
Extensions::ARB::shader_bit_encoding{},
Extensions::ARB::texture_rgb10_a2ui{},
Extensions::ARB::texture_swizzle{},
Extensions::ARB::timer_query{},
Extensions::ARB::vertex_type_2_10_10_10_rev{}
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionList400[]{
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ARB::draw_buffers_blend{},
Extensions::ARB::draw_indirect{},
Extensions::ARB::gpu_shader5{},
Extensions::ARB::gpu_shader_fp64{},
Extensions::ARB::sample_shading{},
Extensions::ARB::shader_subroutine{},
Extensions::ARB::tessellation_shader{},
Extensions::ARB::texture_buffer_object_rgb32{},
Extensions::ARB::texture_cube_map_array{},
Extensions::ARB::texture_gather{},
Extensions::ARB::texture_query_lod{},
Extensions::ARB::transform_feedback2{},
Extensions::ARB::transform_feedback3{}
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionList410[]{
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ARB::ES2_compatibility{},
Extensions::ARB::get_program_binary{},
Extensions::ARB::separate_shader_objects{},
Extensions::ARB::shader_precision{},
Extensions::ARB::vertex_attrib_64bit{},
Extensions::ARB::viewport_array{}
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionList420[]{
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ARB::base_instance{},
Extensions::ARB::compressed_texture_pixel_storage{},
Extensions::ARB::conservative_depth{},
Extensions::ARB::internalformat_query{},
Extensions::ARB::map_buffer_alignment{},
Extensions::ARB::shader_atomic_counters{},
Extensions::ARB::shader_image_load_store{},
Extensions::ARB::shading_language_420pack{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
/* Mentioned in GLSL 4.20 specs as newly added */
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ARB::shading_language_packing{},
Extensions::ARB::texture_compression_bptc{},
Extensions::ARB::texture_storage{},
Extensions::ARB::transform_feedback_instanced{}
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionList430[]{
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ARB::ES3_compatibility{},
Extensions::ARB::arrays_of_arrays{},
Extensions::ARB::clear_buffer_object{},
Extensions::ARB::compute_shader{},
Extensions::ARB::copy_image{},
Extensions::ARB::explicit_uniform_location{},
Extensions::ARB::fragment_layer_viewport{},
Extensions::ARB::framebuffer_no_attachments{},
Extensions::ARB::internalformat_query2{},
Extensions::ARB::invalidate_subdata{},
Extensions::ARB::multi_draw_indirect{},
Extensions::ARB::program_interface_query{},
Extensions::ARB::robust_buffer_access_behavior{},
Extensions::ARB::shader_image_size{},
Extensions::ARB::shader_storage_buffer_object{},
Extensions::ARB::stencil_texturing{},
Extensions::ARB::texture_buffer_range{},
Extensions::ARB::texture_query_levels{},
Extensions::ARB::texture_storage_multisample{},
Extensions::ARB::texture_view{},
Extensions::ARB::vertex_attrib_binding{},
Extensions::KHR::debug{}
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionList440[]{
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ARB::buffer_storage{},
Extensions::ARB::clear_texture{},
Extensions::ARB::enhanced_layouts{},
Extensions::ARB::multi_bind{},
Extensions::ARB::query_buffer_object{},
Extensions::ARB::texture_mirror_clamp_to_edge{},
Extensions::ARB::texture_stencil8{},
Extensions::ARB::vertex_type_10f_11f_11f_rev{}
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionList450[]{
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ARB::ES3_1_compatibility{},
Extensions::ARB::clip_control{},
Extensions::ARB::conditional_render_inverted{},
Extensions::ARB::cull_distance{},
Extensions::ARB::derivative_control{},
Extensions::ARB::direct_state_access{},
Extensions::ARB::get_texture_sub_image{},
Extensions::ARB::shader_texture_image_samples{},
Extensions::ARB::texture_barrier{},
Extensions::KHR::context_flush_control{},
Extensions::KHR::robustness{}
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionList460[]{
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ARB::gl_spirv{},
Extensions::ARB::indirect_parameters{},
Extensions::ARB::pipeline_statistics_query{},
Extensions::ARB::polygon_offset_clamp{},
Extensions::ARB::shader_atomic_counter_ops{},
Extensions::ARB::shader_draw_parameters{},
Extensions::ARB::shader_group_vote{},
Extensions::ARB::spirv_extensions{},
Extensions::ARB::texture_filter_anisotropic{},
Extensions::ARB::transform_feedback_overflow_query{},
Extensions::KHR::no_error{}
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#elif defined(MAGNUM_TARGET_WEBGL)
constexpr Extension ExtensionList[]{
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::clip_cull_distance{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::color_buffer_float{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
GL: implement time queries for WebGL 2, and workarounds, and ... There's a new firefox-fake-disjoint-timer-query-webgl2 workaround and a half-page of text listing various caveats and issues you might run into. Also exposing them in the OpenGLTester (although quite shitty at this point).
7 years ago
#ifdef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::disjoint_timer_query{},
GL: implement time queries for WebGL 2, and workarounds, and ... There's a new firefox-fake-disjoint-timer-query-webgl2 workaround and a half-page of text listing various caveats and issues you might run into. Also exposing them in the OpenGLTester (although quite shitty at this point).
7 years ago
#endif
GL: fix and add a comment so I don't mess this up again in the future.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::disjoint_timer_query_webgl2{},
GL: fix and add a comment so I don't mess this up again in the future.
7 years ago
#endif
GL: implement {EXT,ARB}_draw_buffers{2,_blend,_indexed}. I also figured out a new, faster & less verbose way to handle multiple code paths in some cases -- why didn't I think of that earlier?
6 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::draw_buffers_indexed{},
GL: implement {EXT,ARB}_draw_buffers{2,_blend,_indexed}. I also figured out a new, faster & less verbose way to handle multiple code paths in some cases -- why didn't I think of that earlier?
6 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::float_blend{},
Extensions::EXT::texture_compression_bptc{},
Extensions::EXT::texture_compression_rgtc{},
Extensions::EXT::texture_filter_anisotropic{},
GL: EXT_texture_norm16 is not on WebGL 1 anymore. It got changed shortly after the original implementation in 083ecf72b8a3de8b92cd2df2a597722a636c19da to be WebGL 2 only.
5 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::texture_norm16{},
GL: EXT_texture_norm16 is not on WebGL 1 anymore. It got changed shortly after the original implementation in 083ecf72b8a3de8b92cd2df2a597722a636c19da to be WebGL 2 only.
5 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OES::texture_float_linear{},
external: expose the OVR_multiview, OVR_multiview2 extension entrypoints.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OVR::multiview2{},
external: expose the OVR_multiview, OVR_multiview2 extension entrypoints.
7 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::WEBGL::blend_equation_advanced_coherent{},
Extensions::WEBGL::compressed_texture_astc{},
Extensions::WEBGL::compressed_texture_pvrtc{},
Extensions::WEBGL::compressed_texture_s3tc{},
Extensions::WEBGL::compressed_texture_s3tc_srgb{},
Extensions::WEBGL::debug_renderer_info{},
Extensions::WEBGL::debug_shaders{},
GL: recognize all remaining WebGL extensions.
6 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::WEBGL::draw_instanced_base_vertex_base_instance{},
GL: recognize all remaining WebGL extensions.
6 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::WEBGL::lose_context{},
Extensions::WEBGL::multi_draw{},
GL: recognize all remaining WebGL extensions.
6 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::WEBGL::multi_draw_instanced_base_vertex_base_instance{},
GL: recognize all remaining WebGL extensions.
6 years ago
#endif
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionListES300[]{
#ifdef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ANGLE::instanced_arrays{},
Extensions::EXT::blend_minmax{},
Extensions::EXT::color_buffer_half_float{},
Extensions::EXT::frag_depth{},
Extensions::EXT::sRGB{},
Extensions::EXT::shader_texture_lod{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#endif
GL: don't include MAGNUM_shader_vertex_id on ES2 builds. It makes one think such feature *could* be supported there, while it couldn't.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::MAGNUM::shader_vertex_id{},
GL: don't include MAGNUM_shader_vertex_id on ES2 builds. It makes one think such feature *could* be supported there, while it couldn't.
7 years ago
#endif
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#ifdef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OES::element_index_uint{},
Extensions::OES::fbo_render_mipmap{},
Extensions::OES::standard_derivatives{},
Extensions::OES::texture_float{},
Extensions::OES::texture_half_float{},
Extensions::OES::texture_half_float_linear{},
Extensions::OES::vertex_array_object{},
Extensions::WEBGL::color_buffer_float{},
Extensions::WEBGL::depth_texture{},
Extensions::WEBGL::draw_buffers{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#endif
};
#else
constexpr Extension ExtensionList[]{
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ANDROID::extension_pack_es31a{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#endif
GL: recognize and load ANGLE multi-draw extensions. Needed a change in flextGL to allow merging in 3rd party gl.xml additions because Chrome is apparently a center of the universe and thus doesn't need to bother upstreaming its extensions, ffs.
5 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ANGLE::base_vertex_base_instance{},
GL: recognize and load ANGLE multi-draw extensions. Needed a change in flextGL to allow merging in 3rd party gl.xml additions because Chrome is apparently a center of the universe and thus doesn't need to bother upstreaming its extensions, ffs.
5 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ANGLE::multi_draw{},
Extensions::ANGLE::texture_compression_dxt1{},
Extensions::ANGLE::texture_compression_dxt3{},
Extensions::ANGLE::texture_compression_dxt5{},
Extensions::APPLE::clip_distance{},
Extensions::APPLE::texture_format_BGRA8888{},
Extensions::ARM::shader_framebuffer_fetch{},
Extensions::ARM::shader_framebuffer_fetch_depth_stencil{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::clip_cull_distance{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::debug_label{},
Extensions::EXT::debug_marker{},
Extensions::EXT::disjoint_timer_query{},
Extensions::EXT::draw_elements_base_vertex{},
Extensions::EXT::multi_draw_arrays{},
Extensions::EXT::multisampled_render_to_texture{},
Extensions::EXT::polygon_offset_clamp{},
Extensions::EXT::pvrtc_sRGB{},
Extensions::EXT::read_format_bgra{},
Extensions::EXT::robustness{},
Extensions::EXT::sRGB_write_control{},
Extensions::EXT::separate_shader_objects{},
Extensions::EXT::shader_framebuffer_fetch{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::shader_integer_mix{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#endif
GL: forgot to list those GLES extensions. They were listed on WebGL tho. Sigh.
6 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::texture_compression_bptc{},
GL: forgot to list those GLES extensions. They were listed on WebGL tho. Sigh.
6 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::texture_compression_dxt1{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::texture_compression_rgtc{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::texture_compression_s3tc{},
Extensions::EXT::texture_compression_s3tc_srgb{},
Extensions::EXT::texture_filter_anisotropic{},
Extensions::EXT::texture_format_BGRA8888{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::texture_norm16{},
Extensions::EXT::texture_sRGB_R8{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::texture_sRGB_RG8{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::texture_sRGB_decode{},
Extensions::IMG::texture_compression_pvrtc{},
Extensions::KHR::blend_equation_advanced_coherent{},
Extensions::KHR::context_flush_control{},
Extensions::KHR::no_error{},
Extensions::KHR::texture_compression_astc_hdr{},
Extensions::KHR::texture_compression_astc_sliced_3d{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::NV::fragment_shader_barycentric{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::NV::polygon_mode{},
Extensions::NV::read_buffer_front{},
Extensions::NV::read_depth{},
Extensions::NV::read_depth_stencil{},
Extensions::NV::read_stencil{},
Extensions::NV::sample_locations{},
external: expose a bunch of new GL extensions. ARB_sample_locations and AMD_shader_explicit_vertex_parameter / NV_fragment_shader_barycentric for barycentic coords in the vertex shader. The ARB extension mentions GLES but doesn't have any assigned number, so I'm treating it as desktop-only. There's a NV variant that does have an ES extension number, but I don't really want to add it just because that.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::NV::shader_noperspective_interpolation{},
external: expose a bunch of new GL extensions. ARB_sample_locations and AMD_shader_explicit_vertex_parameter / NV_fragment_shader_barycentric for barycentic coords in the vertex shader. The ARB extension mentions GLES but doesn't have any assigned number, so I'm treating it as desktop-only. There's a NV variant that does have an ES extension number, but I don't really want to add it just because that.
7 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::NV::texture_border_clamp{},
Extensions::OES::depth32{},
Extensions::OES::draw_elements_base_vertex{},
Extensions::OES::mapbuffer{},
Extensions::OES::stencil1{},
Extensions::OES::stencil4{},
ASTC: the pain never ends (3D formats). GL has an extension, but only for ES, not on desktop. Vulkan has nothing yet (due to there being just ARM that implements it, no other vendor), except those being listed in a KTX format specification.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OES::texture_compression_astc{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OES::texture_float_linear{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OVR::multiview{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OVR::multiview2{},
ASTC: the pain never ends (3D formats). GL has an extension, but only for ES, not on desktop. Vulkan has nothing yet (due to there being just ARM that implements it, no other vendor), except those being listed in a KTX format specification.
7 years ago
#endif
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionListES300[]{
#ifdef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::ANGLE::depth_texture{},
Extensions::ANGLE::framebuffer_blit{},
Extensions::ANGLE::framebuffer_multisample{},
Extensions::ANGLE::instanced_arrays{},
Extensions::APPLE::framebuffer_multisample{},
Extensions::APPLE::texture_max_level{},
Extensions::ARM::rgba8{},
Extensions::EXT::blend_minmax{},
Extensions::EXT::discard_framebuffer{},
Extensions::EXT::draw_buffers{},
Extensions::EXT::draw_instanced{},
Extensions::EXT::instanced_arrays{},
Extensions::EXT::map_buffer_range{},
Extensions::EXT::occlusion_query_boolean{},
Extensions::EXT::sRGB{},
Extensions::EXT::shader_texture_lod{},
Extensions::EXT::shadow_samplers{},
Extensions::EXT::texture_rg{},
Extensions::EXT::texture_storage{},
Extensions::EXT::texture_type_2_10_10_10_REV{},
Extensions::EXT::unpack_subimage{},
#endif
#ifndef MAGNUM_TARGET_GLES2
Extensions::MAGNUM::shader_vertex_id{},
GL: don't include MAGNUM_shader_vertex_id on ES2 builds. It makes one think such feature *could* be supported there, while it couldn't.
7 years ago
#endif
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#ifdef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::NV::draw_buffers{},
Extensions::NV::draw_instanced{},
Extensions::NV::fbo_color_attachments{},
Extensions::NV::framebuffer_blit{},
Extensions::NV::framebuffer_multisample{},
Extensions::NV::instanced_arrays{},
Extensions::NV::pack_subimage{},
Extensions::NV::read_buffer{},
Extensions::NV::shadow_samplers_array{},
Extensions::NV::shadow_samplers_cube{},
Extensions::OES::depth24{},
Extensions::OES::depth_texture{},
Extensions::OES::element_index_uint{},
Extensions::OES::fbo_render_mipmap{},
Extensions::OES::packed_depth_stencil{},
Extensions::OES::required_internalformat{},
Extensions::OES::rgb8_rgba8{},
Extensions::OES::standard_derivatives{},
Extensions::OES::surfaceless_context{},
Extensions::OES::texture_3D{},
Extensions::OES::texture_float{},
Extensions::OES::texture_half_float{},
Extensions::OES::texture_half_float_linear{},
Extensions::OES::texture_npot{},
Extensions::OES::vertex_array_object{},
Extensions::OES::vertex_half_float{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#endif
GL: new pseudo-extension MAGNUM_shader_vertex_id. Will be used to distinguish if the driver supports gl_VertexID. This is not the case on SwiftShader ES3 contexts (and could be elsewhere, probably).
7 years ago
};
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
constexpr Extension ExtensionListES320[]{
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::color_buffer_float{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::color_buffer_half_float{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::copy_image{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::draw_buffers_indexed{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::geometry_shader{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::gpu_shader5{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::primitive_bounding_box{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::shader_io_blocks{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::tessellation_shader{},
Don't expose unneeded ES2 extensions under ES3. They provide the same or slightly different functionality as is present in ES3 itself, causing confusion. Fortunately this change required only one modification in Context test and nowhere else.
13 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::texture_border_clamp{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::texture_buffer{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::EXT::texture_cube_map_array{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::KHR::blend_equation_advanced{},
Extensions::KHR::debug{},
Extensions::KHR::robust_buffer_access_behavior{},
Extensions::KHR::robustness{},
Extensions::KHR::texture_compression_astc_ldr{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OES::sample_shading{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OES::sample_variables{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OES::shader_image_atomic{},
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
#endif
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OES::shader_multisample_interpolation{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#endif
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OES::texture_stencil8{},
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#ifndef MAGNUM_TARGET_GLES2
GL: make compile-time extensions convertible to the Extension class. Will need this for supplying disabled extensions to Context configuration.
5 years ago
Extensions::OES::texture_storage_multisample_2d_array{}
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
#endif
};
#endif
GL: MSVC 2015, don't die yet.
5 years ago
#ifdef CORRADE_MSVC2015_COMPATIBILITY
/* MSVC 2015 ICEs in the loop below if this is constexpr. Don't, then. */
const
#else
constexpr
#endif
struct {
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
Version version;
Containers::ArrayView<const Extension> extensions;
} KnownExtensionsForVersion[]{
#ifndef MAGNUM_TARGET_GLES
{Version::GL300, Containers::arrayView(ExtensionList300)},
{Version::GL310, Containers::arrayView(ExtensionList310)},
{Version::GL320, Containers::arrayView(ExtensionList320)},
{Version::GL330, Containers::arrayView(ExtensionList330)},
{Version::GL400, Containers::arrayView(ExtensionList400)},
{Version::GL410, Containers::arrayView(ExtensionList410)},
{Version::GL420, Containers::arrayView(ExtensionList420)},
{Version::GL430, Containers::arrayView(ExtensionList430)},
{Version::GL440, Containers::arrayView(ExtensionList440)},
{Version::GL450, Containers::arrayView(ExtensionList450)},
{Version::GL460, Containers::arrayView(ExtensionList460)},
#else
{Version::GLES300, Containers::arrayView(ExtensionListES300)},
#ifndef MAGNUM_TARGET_WEBGL
/* No extensions in ES 3.1 */
{Version::GLES320, Containers::arrayView(ExtensionListES320)},
#endif
#endif
{Version::None, Containers::arrayView(ExtensionList)}
};
const Extension* findExtension(const Containers::StringView extension, const std::size_t since = 0) {
for(std::size_t i = since; i != Containers::arraySize(KnownExtensionsForVersion); ++i) {
const auto found = std::lower_bound(KnownExtensionsForVersion[i].extensions.begin(), KnownExtensionsForVersion[i].extensions.end(), extension, [](const Extension& a, const Containers::StringView& b) {
return a.string() < b;
});
if(found != KnownExtensionsForVersion[i].extensions.end() && found->string() == extension)
return found;
}
return {};
}
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
}
Classes for handling version and extension information.
14 years ago
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
Containers::ArrayView<const Extension> Extension::extensions(Version version) {
Classes for handling version and extension information.
14 years ago
switch(version) {
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
/* GCC 4.8 needs an explicit cast */
case Version::None: return Containers::arrayView(ExtensionList);
Updated OpenGL ES support in Context and related files.
14 years ago
#ifndef MAGNUM_TARGET_GLES
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
case Version::GL210: return {};
case Version::GL300: return Containers::arrayView(ExtensionList300);
case Version::GL310: return Containers::arrayView(ExtensionList310);
case Version::GL320: return Containers::arrayView(ExtensionList320);
case Version::GL330: return Containers::arrayView(ExtensionList330);
case Version::GL400: return Containers::arrayView(ExtensionList400);
case Version::GL410: return Containers::arrayView(ExtensionList410);
case Version::GL420: return Containers::arrayView(ExtensionList420);
case Version::GL430: return Containers::arrayView(ExtensionList430);
case Version::GL440: return Containers::arrayView(ExtensionList440);
case Version::GL450: return Containers::arrayView(ExtensionList450);
case Version::GL460: return Containers::arrayView(ExtensionList460);
Make the ES versions first-class on desktop OpenGL. What this changes: * Before, a Version::GLES310 shader was created with `#version 450` directive, which is wrong, because an implementation might support GLES 3.1 but not GL 4.5 so this was not working and also the syntax is a bit different so this wasn't helpful at all. * Similarly, Context::isVersionSupported(Version::GLES310) was checking for support of OpenGL 4.5. Now it just checks for `ARB_ES3_1_compatibility` extension. * Added a small isVersionES() utility that just tells whether given version is ES or not.
10 years ago
case Version::GLES200:
case Version::GLES300:
Initial support for OpenGL ES 3.2. A bunch of extensions formerly in AEP are now part of ES 3.2, which means they were reordered in the extension lists. While at it, also added corresponding new GL and WebGL extensions and fixed a few wrongly categorized extensions in WebGL.
9 years ago
case Version::GLES310:
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
case Version::GLES320: return {};
Updated OpenGL ES support in Context and related files.
14 years ago
#else
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
case Version::GLES200: return {};
case Version::GLES300: return Containers::arrayView(ExtensionListES300);
First-class WebGL support, part 13: reduce version enum.
11 years ago
#ifndef MAGNUM_TARGET_WEBGL
Audio, GL: make the extension lists constexpr. The Audio::Extension::extensions() and GL::Extension::extensions() functions now return an ArrayView instead of a reference to std::vector. Having a function-local static std::vector is the worst thing ever. Ugh.
7 years ago
case Version::GLES310: return {};
case Version::GLES320: return Containers::arrayView(ExtensionListES320);
Updated OpenGL ES support in Context and related files.
14 years ago
#endif
First-class WebGL support, part 13: reduce version enum.
11 years ago
#endif
Classes for handling version and extension information.
14 years ago
}
Don't have `default:` in switches where all cases need to be handled. Also added assert for cases which shouldn't be reached.
14 years ago
Adapted to Corrade changes.
6 years ago
CORRADE_INTERNAL_ASSERT_UNREACHABLE(); /* LCOV_EXCL_LINE */
Classes for handling version and extension information.
14 years ago
}
Introduce a BUILD_STATIC_UNIQUE_GLOBALS option. Same as in Corrade. Because BUILD_STATIC is independent between Corrade and Magnum this option is also independent -- the corner cases and bad interactions would be otherwise too complex to handle (e.g., in case of a dynamic Corrade and static Magnum it would be impossible to enable this option for Magnum etc etc).
6 years ago
#if !defined(MAGNUM_BUILD_STATIC_UNIQUE_GLOBALS) || defined(CORRADE_TARGET_WINDOWS)
Implement globals shared across DLLs on Windows.
7 years ago
/* (Of course) can't be in an unnamed namespace in order to export it below
(except for Windows, where we do extern "C" so this doesn't matter) */
Introduce BUILD_MULTITHREADED CMake option. Enabled by default, makes the current Magnum context a thread-local variable instead of a global one, so it's possible to have multiple thread-local contexts. Might have some performance implications, that's why it's possible to disable it (but enabled by default is the safer option). GCC 4.7 and Apple platforms don't support thread_local, but __thread does the job too (though on iOS not until Xcode 7.3). Also had to move it to file-local because MSVC doesn't like having thread local variables as part of DLL interface. (And there is *of course* no way to disable exporting one particular member. F' that.)
10 years ago
namespace {
Export globals and make them weak on static builds so they work across SOs. Well, except on Windows -- not yet. Added an XFAIL for that.
7 years ago
#endif
#ifdef CORRADE_BUILD_MULTITHREADED
CORRADE_THREAD_LOCAL
#endif
Introduce a BUILD_STATIC_UNIQUE_GLOBALS option. Same as in Corrade. Because BUILD_STATIC is independent between Corrade and Magnum this option is also independent -- the corner cases and bad interactions would be otherwise too complex to handle (e.g., in case of a dynamic Corrade and static Magnum it would be impossible to enable this option for Magnum etc etc).
6 years ago
#if defined(MAGNUM_BUILD_STATIC_UNIQUE_GLOBALS) && !defined(CORRADE_TARGET_WINDOWS)
Export globals and make them weak on static builds so they work across SOs. Well, except on Windows -- not yet. Added an XFAIL for that.
7 years ago
/* On static builds that get linked to multiple shared libraries and then used
in a single app we want to ensure there's just one global symbol. On Linux
it's apparently enough to just export, macOS needs the weak attribute.
Implement globals shared across DLLs on Windows.
7 years ago
Windows handled differently below. */
Export globals and make them weak on static builds so they work across SOs. Well, except on Windows -- not yet. Added an XFAIL for that.
7 years ago
CORRADE_VISIBILITY_EXPORT
#ifdef __GNUC__
__attribute__((weak))
#else
/* uh oh? the test will fail, probably */
Introduce BUILD_MULTITHREADED CMake option. Enabled by default, makes the current Magnum context a thread-local variable instead of a global one, so it's possible to have multiple thread-local contexts. Might have some performance implications, that's why it's possible to disable it (but enabled by default is the safer option). GCC 4.7 and Apple platforms don't support thread_local, but __thread does the job too (though on iOS not until Xcode 7.3). Also had to move it to file-local because MSVC doesn't like having thread local variables as part of DLL interface. (And there is *of course* no way to disable exporting one particular member. F' that.)
10 years ago
#endif
Export globals and make them weak on static builds so they work across SOs. Well, except on Windows -- not yet. Added an XFAIL for that.
7 years ago
#endif
Context* currentContext = nullptr;
Introduce a BUILD_STATIC_UNIQUE_GLOBALS option. Same as in Corrade. Because BUILD_STATIC is independent between Corrade and Magnum this option is also independent -- the corner cases and bad interactions would be otherwise too complex to handle (e.g., in case of a dynamic Corrade and static Magnum it would be impossible to enable this option for Magnum etc etc).
6 years ago
#if !defined(MAGNUM_BUILD_STATIC_UNIQUE_GLOBALS) || defined(CORRADE_TARGET_WINDOWS)
Implement globals shared across DLLs on Windows.
7 years ago
}
#endif
/* Windows can't have a symbol both thread-local and exported, moreover there
isn't any concept of weak symbols. Exporting thread-local symbols can be
worked around by exporting a function that then returns a reference to a
non-exported thread-local symbol; and finally GetProcAddress() on
GetModuleHandle(nullptr) "emulates" the weak linking as it's guaranteed to
pick up the same symbol of the final exe independently of the DLL it was
called from. To avoid #ifdef hell in code below, the currentContext is
redefined to return a value from this uniqueness-ensuring function. */
Introduce a BUILD_STATIC_UNIQUE_GLOBALS option. Same as in Corrade. Because BUILD_STATIC is independent between Corrade and Magnum this option is also independent -- the corner cases and bad interactions would be otherwise too complex to handle (e.g., in case of a dynamic Corrade and static Magnum it would be impossible to enable this option for Magnum etc etc).
6 years ago
#if defined(CORRADE_TARGET_WINDOWS) && defined(MAGNUM_BUILD_STATIC_UNIQUE_GLOBALS) && !defined(CORRADE_TARGET_WINDOWS_RT)
GL: fix clang-cl warning about C-incompatible function signatures.
6 years ago
/* Clang-CL complains that the function has a return type incompatible with C.
I don't care, I only need an unmangled name to look up later at runtime. */
#ifdef CORRADE_TARGET_CLANG_CL
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Wreturn-type-c-linkage"
#endif
Audio, GL: (attempt to) fix MinGW GCC warnings on static Windows builds. The CI doesn't have this configuration in the build matrix, so hoping it's doing what it's supposed to.
7 years ago
extern "C" CORRADE_VISIBILITY_EXPORT Context*& magnumGLUniqueCurrentContext();
Implement globals shared across DLLs on Windows.
7 years ago
extern "C" CORRADE_VISIBILITY_EXPORT Context*& magnumGLUniqueCurrentContext() {
return currentContext;
}
GL: fix clang-cl warning about C-incompatible function signatures.
6 years ago
#ifdef CORRADE_TARGET_CLANG_CL
#pragma clang diagnostic pop
#endif
Implement globals shared across DLLs on Windows.
7 years ago
namespace {
Context*& windowsCurrentContext() {
/* A function-local static to ensure it's only initialized once without any
race conditions among threads */
A sad workaround for a sad misery with Python bindings on Windows.
7 years ago
static Context*&(*const uniqueGlobals)() = reinterpret_cast<Context*&(*)()>(Magnum::Implementation::windowsWeakSymbol("magnumGLUniqueCurrentContext", reinterpret_cast<void*>(magnumGLUniqueCurrentContext)));
Implement globals shared across DLLs on Windows.
7 years ago
return uniqueGlobals();
Introduce BUILD_MULTITHREADED CMake option. Enabled by default, makes the current Magnum context a thread-local variable instead of a global one, so it's possible to have multiple thread-local contexts. Might have some performance implications, that's why it's possible to disable it (but enabled by default is the safer option). GCC 4.7 and Apple platforms don't support thread_local, but __thread does the job too (though on iOS not until Xcode 7.3). Also had to move it to file-local because MSVC doesn't like having thread local variables as part of DLL interface. (And there is *of course* no way to disable exporting one particular member. F' that.)
10 years ago
}
Implement globals shared across DLLs on Windows.
7 years ago
}
#define currentContext windowsCurrentContext()
Export globals and make them weak on static builds so they work across SOs. Well, except on Windows -- not yet. Added an XFAIL for that.
7 years ago
#endif
Classes for handling version and extension information.
14 years ago
Introduce BUILD_MULTITHREADED CMake option. Enabled by default, makes the current Magnum context a thread-local variable instead of a global one, so it's possible to have multiple thread-local contexts. Might have some performance implications, that's why it's possible to disable it (but enabled by default is the safer option). GCC 4.7 and Apple platforms don't support thread_local, but __thread does the job too (though on iOS not until Xcode 7.3). Also had to move it to file-local because MSVC doesn't like having thread local variables as part of DLL interface. (And there is *of course* no way to disable exporting one particular member. F' that.)
10 years ago
bool Context::hasCurrent() { return currentContext; }
Context::current() now returns reference instead of pointer. 103% of use cases use the returned value directly without checking, so we might as well do the check ourselves. Added new function hasCurrent() and added deprecated backward-compatibility conversion and -> operators. Wow, that creeped to a lot of places. Last dinosaur from the pointer age.
10 years ago
Context& Context::current() {
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
CORRADE_ASSERT(currentContext, "GL::Context::current(): no current context", *currentContext);
Introduce BUILD_MULTITHREADED CMake option. Enabled by default, makes the current Magnum context a thread-local variable instead of a global one, so it's possible to have multiple thread-local contexts. Might have some performance implications, that's why it's possible to disable it (but enabled by default is the safer option). GCC 4.7 and Apple platforms don't support thread_local, but __thread does the job too (though on iOS not until Xcode 7.3). Also had to move it to file-local because MSVC doesn't like having thread local variables as part of DLL interface. (And there is *of course* no way to disable exporting one particular member. F' that.)
10 years ago
return *currentContext;
Context::current() now returns reference instead of pointer. 103% of use cases use the returned value directly without checking, so we might as well do the check ourselves. Added new function hasCurrent() and added deprecated backward-compatibility conversion and -> operators. Wow, that creeped to a lot of places. Last dinosaur from the pointer age.
10 years ago
}
GL: added Context::makeCurrent().
7 years ago
void Context::makeCurrent(Context* context) { currentContext = context; }
external: pass GL::Context to flextGLInit(). To make it possible to put driver-specific workaround there.
7 years ago
Context::Context(NoCreateT, Int argc, const char** argv, void functionLoader(Context&)): Context{NoCreate, Utility::Arguments{"magnum"}, argc, argv, functionLoader} {}
GL: make it possible to pass a pre-filled Utility::Arguments to Context. For example to support additional app-specific configuration.
8 years ago
external: pass GL::Context to flextGLInit(). To make it possible to put driver-specific workaround there.
7 years ago
Context::Context(NoCreateT, Utility::Arguments& args, Int argc, const char** argv, void functionLoader(Context&)): _functionLoader{functionLoader}, _version{Version::None} {
Initial support for passing arguments directly to the engine. The parameterless Platform::Context::Context() constructor and Platform::Context::tryCreate() function are deprecated in favor functions that take argc/argv pair. The Context class now accepts arguments starting with --magnum-* prefix. Currently there are none except for the implicit --magnum-help, but that will change with the following commits.
11 years ago
/* Parse arguments */
GL: make it possible to pass a pre-filled Utility::Arguments to Context. For example to support additional app-specific configuration.
8 years ago
CORRADE_INTERNAL_ASSERT(args.prefix() == "magnum");
GL: minor.
6 years ago
args.addOption("disable-workarounds").setHelp("disable-workarounds", "driver workarounds to disable\n (see https://doc.magnum.graphics/magnum/opengl-workarounds.html for detailed info)", "LIST")
Audio: make it possible to disable OpenAL extensions as well. Finally, a feature parity.
7 years ago
.addOption("disable-extensions").setHelp("disable-extensions", "API extensions to disable", "LIST")
GL: implement --magnum-gpu-validation no-error. I wonder if this actually makes any difference on any drivers.
5 years ago
.addOption("gpu-validation", "off").setHelp("gpu-validation", "GPU validation using KHR_debug (if present)", "off|on|no-error")
GL: update the --magnum-log option to include "verbose".
8 years ago
.addOption("log", "default").setHelp("log", "console logging", "default|quiet|verbose")
Ability to disable workarounds and extensions from environment. Use MAGNUM_DISABLE_WORKAROUNDS and MAGNUM_DISABLE_EXTENSIONS environment variables.
10 years ago
.setFromEnvironment("disable-workarounds")
.setFromEnvironment("disable-extensions")
GL: new --magnum-gpu-validation command-line option. Enables KHR_debug, if available. This should have been here since ages.
7 years ago
.setFromEnvironment("gpu-validation")
Ability to control the amount of console output in engine initialization.
10 years ago
.setFromEnvironment("log")
More transparent way to handle driver workarounds. The engine can now list all driver workarounds that were used during the initialization. Any listed workaround can then be disabled from the command-line using `--magnum-disable-workarounds` command-line parameter. The disabling and querying API is private and undocumented, because the driver workarounds should be disabled only by end-users and not application developers. The workaround list is now empty, but will be filled up in the following commits and the workarounds will be probably documented only privately in Implementation/driverSpecific.hpp, as it is really something that should be used only to debug driver problems.
11 years ago
.parse(argc, argv);
GL: print a message about GPU validation in verbose log output.
7 years ago
/* Decide how to display initialization log */
if(args.value("log") == "verbose" || args.value("log") == "VERBOSE")
GL: no need for separate Context::InternalFlags anymore. The public Context::Configuration::Flags can supplement that, and that also makes option merging easier.
5 years ago
_configurationFlags |= Configuration::Flag::VerboseLog;
else if(args.value("log") == "quiet" || args.value("log") == "QUIET")
_configurationFlags |= Configuration::Flag::QuietLog;
Ability to control the amount of console output in engine initialization.
10 years ago
GL: implement --magnum-gpu-validation no-error. I wonder if this actually makes any difference on any drivers.
5 years ago
/* Decide whether to enable GPU validation / no error context */
GL: new --magnum-gpu-validation command-line option. Enables KHR_debug, if available. This should have been here since ages.
7 years ago
if(args.value("gpu-validation") == "on" || args.value("gpu-validation") == "ON")
GL: no need for separate Context::InternalFlags anymore. The public Context::Configuration::Flags can supplement that, and that also makes option merging easier.
5 years ago
_configurationFlags |= Configuration::Flag::GpuValidation;
GL: implement --magnum-gpu-validation no-error. I wonder if this actually makes any difference on any drivers.
5 years ago
else if(args.value("gpu-validation") == "no-error")
_configurationFlags |= Configuration::Flag::GpuValidationNoError;
GL: new --magnum-gpu-validation command-line option. Enables KHR_debug, if available. This should have been here since ages.
7 years ago
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
/* If there are any disabled workarounds, save them until tryCreate() uses
them. The disableWorkaround() function saves the internal string view
instead of the one passed from the command line so we don't need to
bother with String allocations. */
const Containers::StringView disabledWorkarounds = args.value<Containers::StringView>("disable-workarounds");
if(!disabledWorkarounds.isEmpty()) {
const Containers::Array<Containers::StringView> split = disabledWorkarounds.splitWithoutEmptyParts();
arrayReserve(_driverWorkarounds, split.size());
for(const Containers::StringView workaround: split)
disableDriverWorkaround(workaround);
}
Ability to disable OpenGL extensions from command-line. Added `--magnum-disable-extensions` option next to the already existing `--magnum-disable-workarounds`. Specified extensions are not used automatically and also report as disabled and unsupported when asked using Context::isExtensionSupported() and Context::isExtensionDisabled().
11 years ago
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
/* Disable extensions. Here we search for them among the known extensions
and store the Extension objects instead, which avoids the string copying
and another binary search in tryCreate(). */
const Containers::StringView disabledExtensions = args.value<Containers::StringView>("disable-extensions");
if(!disabledExtensions.isEmpty()) {
const Containers::Array<Containers::StringView> split = disabledExtensions.splitWithoutEmptyParts();
arrayReserve(_disabledExtensions, split.size());
for(const Containers::StringView extension: split) {
if(const Extension* found = findExtension(extension)) {
arrayAppend(_disabledExtensions, *found);
}
}
}
Initial support for passing arguments directly to the engine. The parameterless Platform::Context::Context() constructor and Platform::Context::tryCreate() function are deprecated in favor functions that take argc/argv pair. The Context class now accepts arguments starting with --magnum-* prefix. Currently there are none except for the implicit --magnum-help, but that will change with the following commits.
11 years ago
}
GL: fix Context move constructor. The missing pieces made the delayed creation broken, it worked properly when the context was created *and then* moved.
5 years ago
Context::Context(Context&& other) noexcept:
_functionLoader{other._functionLoader},
_version{other._version},
Initial support for passing arguments directly to the engine. The parameterless Platform::Context::Context() constructor and Platform::Context::tryCreate() function are deprecated in favor functions that take argc/argv pair. The Context class now accepts arguments starting with --magnum-* prefix. Currently there are none except for the implicit --magnum-help, but that will change with the following commits.
11 years ago
#ifndef MAGNUM_TARGET_WEBGL
Modernize #1
9 years ago
_flags{other._flags},
Initial support for passing arguments directly to the engine. The parameterless Platform::Context::Context() constructor and Platform::Context::tryCreate() function are deprecated in favor functions that take argc/argv pair. The Context class now accepts arguments starting with --magnum-* prefix. Currently there are none except for the implicit --magnum-help, but that will change with the following commits.
11 years ago
#endif
Modernize #1
9 years ago
_extensionStatus{other._extensionStatus},
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
_extensionRequiredVersion{other._extensionRequiredVersion},
GL: deprecate Context::supportedExtensions(). I don't see a real use case for this API (and don't remember ever using it) and it only causes extra overhead during context creation (and then a ton of useless allocations at runtime).
5 years ago
#ifdef MAGNUM_BUILD_DEPRECATED
Initial support for passing arguments directly to the engine. The parameterless Platform::Context::Context() constructor and Platform::Context::tryCreate() function are deprecated in favor functions that take argc/argv pair. The Context class now accepts arguments starting with --magnum-* prefix. Currently there are none except for the implicit --magnum-help, but that will change with the following commits.
11 years ago
_supportedExtensions{std::move(other._supportedExtensions)},
GL: deprecate Context::supportedExtensions(). I don't see a real use case for this API (and don't remember ever using it) and it only causes extra overhead during context creation (and then a ton of useless allocations at runtime).
5 years ago
#endif
GL: Use Corrade::Containers::Pointer for Context _state The old raw pointer could cause a segfault if the context is never created, since the destructor unconditionally deleted _state.
7 years ago
_state{std::move(other._state)},
GL: fix Context move constructor. The missing pieces made the delayed creation broken, it worked properly when the context was created *and then* moved.
5 years ago
_detectedDrivers{std::move(other._detectedDrivers)},
_driverWorkarounds{std::move(other._driverWorkarounds)},
_disabledExtensions{std::move(other._disabledExtensions)},
GL: no need for separate Context::InternalFlags anymore. The public Context::Configuration::Flags can supplement that, and that also makes option merging easier.
5 years ago
_configurationFlags{other._configurationFlags}
Initial support for passing arguments directly to the engine. The parameterless Platform::Context::Context() constructor and Platform::Context::tryCreate() function are deprecated in favor functions that take argc/argv pair. The Context class now accepts arguments starting with --magnum-* prefix. Currently there are none except for the implicit --magnum-help, but that will change with the following commits.
11 years ago
{
Introduce BUILD_MULTITHREADED CMake option. Enabled by default, makes the current Magnum context a thread-local variable instead of a global one, so it's possible to have multiple thread-local contexts. Might have some performance implications, that's why it's possible to disable it (but enabled by default is the safer option). GCC 4.7 and Apple platforms don't support thread_local, but __thread does the job too (though on iOS not until Xcode 7.3). Also had to move it to file-local because MSVC doesn't like having thread local variables as part of DLL interface. (And there is *of course* no way to disable exporting one particular member. F' that.)
10 years ago
if(currentContext == &other) currentContext = this;
Initial support for passing arguments directly to the engine. The parameterless Platform::Context::Context() constructor and Platform::Context::tryCreate() function are deprecated in favor functions that take argc/argv pair. The Context class now accepts arguments starting with --magnum-* prefix. Currently there are none except for the implicit --magnum-help, but that will change with the following commits.
11 years ago
}
Move extension loading into Magnum::Context. 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.
13 years ago
Initial support for passing arguments directly to the engine. The parameterless Platform::Context::Context() constructor and Platform::Context::tryCreate() function are deprecated in favor functions that take argc/argv pair. The Context class now accepts arguments starting with --magnum-* prefix. Currently there are none except for the implicit --magnum-help, but that will change with the following commits.
11 years ago
Context::~Context() {
Introduce BUILD_MULTITHREADED CMake option. Enabled by default, makes the current Magnum context a thread-local variable instead of a global one, so it's possible to have multiple thread-local contexts. Might have some performance implications, that's why it's possible to disable it (but enabled by default is the safer option). GCC 4.7 and Apple platforms don't support thread_local, but __thread does the job too (though on iOS not until Xcode 7.3). Also had to move it to file-local because MSVC doesn't like having thread local variables as part of DLL interface. (And there is *of course* no way to disable exporting one particular member. F' that.)
10 years ago
if(currentContext == this) currentContext = nullptr;
Ability to do GL context creation fallbacks even from Context constructor. In some cases the GL context creation might success without error, but the created version is one that we don't want (e.g. software GDI rasterizer on Windows). Previously the Context class constructor just exited the application and it was impossible to react on that from the application side (for example reducing some context feature requirements). Now there is Platform::Context::tryCreate(), which returns either created instance or `nullptr` if the instance creation failed with some error. That is now used in all Platform::*Application implementations.
11 years ago
}
GL: introduce Context::Configuration mirroring --magnum-* arguments. Disabling engine startup log or modifying enabled extensions / workarounds from the application side was one of the common pain points and this should *finally* solve the problem. This Configuration is now inherited by the usual Platform::*Application::GLConfiguration / Platform::Windowless*Application::Configuration classes people are used to, so for the end user it's just as if these classes got a bunch new options. Having this, I also extended the ContextGLTest to verify that the Configuration and command-line options do what's expected because that hadn't automated tests until now. The test is mostly a copy of what I did for Vulkan already, nothing special. Additionally all Platform*ApplicationTest executables gained a new --quiet option to verify that the GL::Context::Configuration subset gets correctly passed from the Application code, because that's something we can't really verify in an automated way.
5 years ago
void Context::create(const Configuration& configuration) {
Ability to do GL context creation fallbacks even from Context constructor. In some cases the GL context creation might success without error, but the created version is one that we don't want (e.g. software GDI rasterizer on Windows). Previously the Context class constructor just exited the application and it was impossible to react on that from the application side (for example reducing some context feature requirements). Now there is Platform::Context::tryCreate(), which returns either created instance or `nullptr` if the instance creation failed with some error. That is now used in all Platform::*Application implementations.
11 years ago
/* Hard exit if the context cannot be created */
GL: introduce Context::Configuration mirroring --magnum-* arguments. Disabling engine startup log or modifying enabled extensions / workarounds from the application side was one of the common pain points and this should *finally* solve the problem. This Configuration is now inherited by the usual Platform::*Application::GLConfiguration / Platform::Windowless*Application::Configuration classes people are used to, so for the end user it's just as if these classes got a bunch new options. Having this, I also extended the ContextGLTest to verify that the Configuration and command-line options do what's expected because that hadn't automated tests until now. The test is mostly a copy of what I did for Vulkan already, nothing special. Additionally all Platform*ApplicationTest executables gained a new --quiet option to verify that the GL::Context::Configuration subset gets correctly passed from the Application code, because that's something we can't really verify in an automated way.
5 years ago
if(!tryCreate(configuration)) std::exit(1);
Ability to do GL context creation fallbacks even from Context constructor. In some cases the GL context creation might success without error, but the created version is one that we don't want (e.g. software GDI rasterizer on Windows). Previously the Context class constructor just exited the application and it was impossible to react on that from the application side (for example reducing some context feature requirements). Now there is Platform::Context::tryCreate(), which returns either created instance or `nullptr` if the instance creation failed with some error. That is now used in all Platform::*Application implementations.
11 years ago
}
GL: introduce Context::Configuration mirroring --magnum-* arguments. Disabling engine startup log or modifying enabled extensions / workarounds from the application side was one of the common pain points and this should *finally* solve the problem. This Configuration is now inherited by the usual Platform::*Application::GLConfiguration / Platform::Windowless*Application::Configuration classes people are used to, so for the end user it's just as if these classes got a bunch new options. Having this, I also extended the ContextGLTest to verify that the Configuration and command-line options do what's expected because that hadn't automated tests until now. The test is mostly a copy of what I did for Vulkan already, nothing special. Additionally all Platform*ApplicationTest executables gained a new --quiet option to verify that the GL::Context::Configuration subset gets correctly passed from the Application code, because that's something we can't really verify in an automated way.
5 years ago
bool Context::tryCreate(const Configuration& configuration) {
Initial support for passing arguments directly to the engine. The parameterless Platform::Context::Context() constructor and Platform::Context::tryCreate() function are deprecated in favor functions that take argc/argv pair. The Context class now accepts arguments starting with --magnum-* prefix. Currently there are none except for the implicit --magnum-help, but that will change with the following commits.
11 years ago
CORRADE_ASSERT(_version == Version::None,
"Platform::Context::tryCreate(): context already created", false);
GL: introduce Context::Configuration mirroring --magnum-* arguments. Disabling engine startup log or modifying enabled extensions / workarounds from the application side was one of the common pain points and this should *finally* solve the problem. This Configuration is now inherited by the usual Platform::*Application::GLConfiguration / Platform::Windowless*Application::Configuration classes people are used to, so for the end user it's just as if these classes got a bunch new options. Having this, I also extended the ContextGLTest to verify that the Configuration and command-line options do what's expected because that hadn't automated tests until now. The test is mostly a copy of what I did for Vulkan already, nothing special. Additionally all Platform*ApplicationTest executables gained a new --quiet option to verify that the GL::Context::Configuration subset gets correctly passed from the Application code, because that's something we can't really verify in an automated way.
5 years ago
/* Merge the configuration with parameters passed on the command line /
environment. For the log command-line gets a priority -- if it says
quiet, it'll override the verbose setting from the configuration; if
it says verbose, the quiet setting from the configuration will be
ignored */
GL: no need for separate Context::InternalFlags anymore. The public Context::Configuration::Flags can supplement that, and that also makes option merging easier.
5 years ago
if((configuration.flags() & Configuration::Flag::VerboseLog) && !(_configurationFlags & Configuration::Flag::QuietLog))
_configurationFlags |= Configuration::Flag::VerboseLog;
else if((configuration.flags() & Configuration::Flag::QuietLog) && !(_configurationFlags & Configuration::Flag::VerboseLog))
_configurationFlags |= Configuration::Flag::QuietLog;
GL: introduce Context::Configuration mirroring --magnum-* arguments. Disabling engine startup log or modifying enabled extensions / workarounds from the application side was one of the common pain points and this should *finally* solve the problem. This Configuration is now inherited by the usual Platform::*Application::GLConfiguration / Platform::Windowless*Application::Configuration classes people are used to, so for the end user it's just as if these classes got a bunch new options. Having this, I also extended the ContextGLTest to verify that the Configuration and command-line options do what's expected because that hadn't automated tests until now. The test is mostly a copy of what I did for Vulkan already, nothing special. Additionally all Platform*ApplicationTest executables gained a new --quiet option to verify that the GL::Context::Configuration subset gets correctly passed from the Application code, because that's something we can't really verify in an automated way.
5 years ago
/* GPU validation is enabled if either enables it */
if(configuration.flags() & Configuration::Flag::GpuValidation)
GL: no need for separate Context::InternalFlags anymore. The public Context::Configuration::Flags can supplement that, and that also makes option merging easier.
5 years ago
_configurationFlags |= Configuration::Flag::GpuValidation;
GL: implement --magnum-gpu-validation no-error. I wonder if this actually makes any difference on any drivers.
5 years ago
if(configuration.flags() & Configuration::Flag::GpuValidationNoError)
_configurationFlags |= Configuration::Flag::GpuValidationNoError;
GL: introduce Context::Configuration mirroring --magnum-* arguments. Disabling engine startup log or modifying enabled extensions / workarounds from the application side was one of the common pain points and this should *finally* solve the problem. This Configuration is now inherited by the usual Platform::*Application::GLConfiguration / Platform::Windowless*Application::Configuration classes people are used to, so for the end user it's just as if these classes got a bunch new options. Having this, I also extended the ContextGLTest to verify that the Configuration and command-line options do what's expected because that hadn't automated tests until now. The test is mostly a copy of what I did for Vulkan already, nothing special. Additionally all Platform*ApplicationTest executables gained a new --quiet option to verify that the GL::Context::Configuration subset gets correctly passed from the Application code, because that's something we can't really verify in an automated way.
5 years ago
GL,Platform: introduce GL::Context::Configuration::Flag::Windowless. With this flag set (which is done implicitly for all windowless apps and, conversely, not done for all windowed apps), the default framebuffer state isn't touched in any way, which should avoid potential race conditions with default framebuffer on another thread.
5 years ago
/* Same for windowless */
if(configuration.flags() & Configuration::Flag::Windowless)
_configurationFlags |= Configuration::Flag::Windowless;
GL: introduce Context::Configuration mirroring --magnum-* arguments. Disabling engine startup log or modifying enabled extensions / workarounds from the application side was one of the common pain points and this should *finally* solve the problem. This Configuration is now inherited by the usual Platform::*Application::GLConfiguration / Platform::Windowless*Application::Configuration classes people are used to, so for the end user it's just as if these classes got a bunch new options. Having this, I also extended the ContextGLTest to verify that the Configuration and command-line options do what's expected because that hadn't automated tests until now. The test is mostly a copy of what I did for Vulkan already, nothing special. Additionally all Platform*ApplicationTest executables gained a new --quiet option to verify that the GL::Context::Configuration subset gets correctly passed from the Application code, because that's something we can't really verify in an automated way.
5 years ago
/* Driver workarounds get merged. Not using disableDriverWorkaround() here
since the Configuration already contains the internal string views. */
for(const Containers::StringView workaround: configuration.disabledWorkarounds())
arrayAppend(_driverWorkarounds, Containers::InPlaceInit, workaround, true);
/* Extensions get merged also. Here we had the chance to force users to
give us the predefined extension types so no need to search for their
IDs */
for(const Extension& extension: configuration.disabledExtensions())
arrayAppend(_disabledExtensions, extension);
external: pass GL::Context to flextGLInit(). To make it possible to put driver-specific workaround there.
7 years ago
/* Load GL function pointers. Pass this instance to it so it can use it for
potential driver-specific workarounds. */
if(_functionLoader) _functionLoader(*this);
Initial support for passing arguments directly to the engine. The parameterless Platform::Context::Context() constructor and Platform::Context::tryCreate() function are deprecated in favor functions that take argc/argv pair. The Context class now accepts arguments starting with --magnum-* prefix. Currently there are none except for the implicit --magnum-help, but that will change with the following commits.
11 years ago
Initialize version to (0, 0) to avoid crashes on broken GL contexts.
10 years ago
/* Initialize to something predictable to avoid crashes on improperly
created contexts */
GLint majorVersion = 0, minorVersion = 0;
Remove unused Context::{major,minor}Version() functions. Superseeded by Context::version() and various utilities around Version enum.
11 years ago
Don't query GL_{MMAJOR,MINOR}_VERSION on WebGL 2. It doesn't make any sense -- WebGL 2 doesn't provide any equivalent functionality for this and so I don't even know what values should this return (3.0? 2.0?). Instead I just check for "WebGL 2" string prefix and treat it as ES 3.0.
11 years ago
/* Get version on ES 3.0+/WebGL 2.0+ */
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
#if defined(MAGNUM_TARGET_GLES) && !defined(MAGNUM_TARGET_GLES2)
Don't query GL_{MMAJOR,MINOR}_VERSION on WebGL 2. It doesn't make any sense -- WebGL 2 doesn't provide any equivalent functionality for this and so I don't even know what values should this return (3.0? 2.0?). Instead I just check for "WebGL 2" string prefix and treat it as ES 3.0.
11 years ago
/* ES 3.0+ */
#ifndef MAGNUM_TARGET_WEBGL
Remove unused Context::{major,minor}Version() functions. Superseeded by Context::version() and various utilities around Version enum.
11 years ago
glGetIntegerv(GL_MAJOR_VERSION, &majorVersion);
glGetIntegerv(GL_MINOR_VERSION, &minorVersion);
Don't query GL_{MMAJOR,MINOR}_VERSION on WebGL 2. It doesn't make any sense -- WebGL 2 doesn't provide any equivalent functionality for this and so I don't even know what values should this return (3.0? 2.0?). Instead I just check for "WebGL 2" string prefix and treat it as ES 3.0.
11 years ago
/* WebGL 2.0, treat it as ES 3.0 */
If targetting exactly ES 2.0, hide everything not available there. Some target platforms supply their own OpenGL headers, thus we cannot use our own from ES 3.0 and compilation fails. On the other hand, this will be better for users as usage of unsupported features will be catched right during compilation and not at runtime.
14 years ago
#else
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
const Containers::StringView version = versionString();
if(!version.contains("WebGL 2"_s)) {
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
Error{} << "GL::Context: unsupported version string:" << version;
Ability to do GL context creation fallbacks even from Context constructor. In some cases the GL context creation might success without error, but the created version is one that we don't want (e.g. software GDI rasterizer on Windows). Previously the Context class constructor just exited the application and it was impossible to react on that from the application side (for example reducing some context feature requirements). Now there is Platform::Context::tryCreate(), which returns either created instance or `nullptr` if the instance creation failed with some error. That is now used in all Platform::*Application implementations.
11 years ago
return false;
Don't query GL_{MMAJOR,MINOR}_VERSION on WebGL 2. It doesn't make any sense -- WebGL 2 doesn't provide any equivalent functionality for this and so I don't even know what values should this return (3.0? 2.0?). Instead I just check for "WebGL 2" string prefix and treat it as ES 3.0.
11 years ago
}
Remove unused Context::{major,minor}Version() functions. Superseeded by Context::version() and various utilities around Version enum.
11 years ago
majorVersion = 3;
minorVersion = 0;
Don't query GL_{MMAJOR,MINOR}_VERSION on WebGL 2. It doesn't make any sense -- WebGL 2 doesn't provide any equivalent functionality for this and so I don't even know what values should this return (3.0? 2.0?). Instead I just check for "WebGL 2" string prefix and treat it as ES 3.0.
11 years ago
#endif
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
/* On GL 2.1 and ES 2.0 there is no GL_{MAJOR,MINOR}_VERSION, we have to
parse version string. On desktop GL we have no way to check version
without version (duh) so we work around that by checking for invalid
enum error. */
Don't query GL_{MMAJOR,MINOR}_VERSION on WebGL 2. It doesn't make any sense -- WebGL 2 doesn't provide any equivalent functionality for this and so I don't even know what values should this return (3.0? 2.0?). Instead I just check for "WebGL 2" string prefix and treat it as ES 3.0.
11 years ago
#else
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
#ifndef MAGNUM_TARGET_GLES2
Remove unused Context::{major,minor}Version() functions. Superseeded by Context::version() and various utilities around Version enum.
11 years ago
glGetIntegerv(GL_MAJOR_VERSION, &majorVersion);
Fix ES3 build. Gah. And I tested it seven times.
12 years ago
const auto versionNumberError = Renderer::error();
if(versionNumberError == Renderer::Error::NoError)
Remove unused Context::{major,minor}Version() functions. Superseeded by Context::version() and various utilities around Version enum.
11 years ago
glGetIntegerv(GL_MINOR_VERSION, &minorVersion);
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
else
If targetting exactly ES 2.0, hide everything not available there. Some target platforms supply their own OpenGL headers, thus we cannot use our own from ES 3.0 and compilation fails. On the other hand, this will be better for users as usage of unsupported features will be catched right during compilation and not at runtime.
14 years ago
#endif
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
{
#ifndef MAGNUM_TARGET_GLES2
Fix ES3 build. Gah. And I tested it seven times.
12 years ago
CORRADE_ASSERT(versionNumberError == Renderer::Error::InvalidEnum,
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
"GL::Context: cannot retrieve OpenGL version:" << versionNumberError, false);
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
#endif
Be less strict in OpenGL ES context version. I didn't yet have time to add more complete support for ES 3.1, but the engine should at least work on it. If the driver reports ES 3.0 or 3.1 it is taken as valid value in both ES2 and ES3 builds (i.e. the engine can work in ES2 mode also on ES3 driver).
12 years ago
/* Allow ES2 context on driver that reports ES3 as supported */
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
const Containers::StringView version = versionString();
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
#ifndef MAGNUM_TARGET_GLES
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
if(version.hasPrefix("2.1"_s))
Use more general MAGNUM_TARGET_WEBGL instead of CORRADE_TARGET_EMSCRIPTEN. Only where it matters, though.
12 years ago
#elif defined(MAGNUM_TARGET_WEBGL)
Support Internet Explorer. It claims to have WebGL 0.94.
11 years ago
/* Internet Explorer currently has 0.94 */
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
if(version.contains("WebGL 1"_s) ||
version.contains("WebGL 0"_s))
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
#else
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
if(version.contains("OpenGL ES 2.0"_s) ||
Document the non-obvious behavior.
11 years ago
/* It is possible to use Magnum compiled for ES2 on ES3 contexts */
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
version.contains("OpenGL ES 3."_s))
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
#endif
{
Remove unused Context::{major,minor}Version() functions. Superseeded by Context::version() and various utilities around Version enum.
11 years ago
majorVersion = 2;
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
#ifndef MAGNUM_TARGET_GLES
Remove unused Context::{major,minor}Version() functions. Superseeded by Context::version() and various utilities around Version enum.
11 years ago
minorVersion = 1;
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
#else
Remove unused Context::{major,minor}Version() functions. Superseeded by Context::version() and various utilities around Version enum.
11 years ago
minorVersion = 0;
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
#endif
} else {
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
Error{} << "GL::Context: unsupported version string:" << version;
Ability to do GL context creation fallbacks even from Context constructor. In some cases the GL context creation might success without error, but the created version is one that we don't want (e.g. software GDI rasterizer on Windows). Previously the Context class constructor just exited the application and it was impossible to react on that from the application side (for example reducing some context feature requirements). Now there is Platform::Context::tryCreate(), which returns either created instance or `nullptr` if the instance creation failed with some error. That is now used in all Platform::*Application implementations.
11 years ago
return false;
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
}
}
#endif
/* Compose the version enum */
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
_version = GL::version(majorVersion, minorVersion);
Classes for handling version and extension information.
14 years ago
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
/* Check that version retrieval went right */
Print more information when cannot retrieve OpenGL version. At least the OpenGL error should be enough to pinpoint what went wrong.
12 years ago
#ifndef CORRADE_NO_ASSERT
Fix ES3 build. Gah. And I tested it seven times.
12 years ago
const auto error = Renderer::error();
Print more information when cannot retrieve OpenGL version. At least the OpenGL error should be enough to pinpoint what went wrong.
12 years ago
CORRADE_ASSERT(error == Renderer::Error::NoError,
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
"GL::Context: cannot retrieve OpenGL version:" << error, false);
Print more information when cannot retrieve OpenGL version. At least the OpenGL error should be enough to pinpoint what went wrong.
12 years ago
#endif
Verify that version retrieval went right. Sometimes the context is created improperly and all gl* functions are causing errors. Added check for that case.
13 years ago
Fix version checks on OpenGL 2.1 and OpenGL ES 2.0. Both don't have GL_{MAJOR,MINOR}_VERSION enums so we must decide whether the version is older than OpenGL 3.0 / ES 3.0 either using MAGNUM_TARGET_GLES2 preprocessor definition or checking for invalid enum error and then we have to parse version string to ensure that the version is not too old. On ES2 the version check wasn't present at all (only optimistic hope), this makes it more hardened.
12 years ago
/* Check that the version is supported (now it probably is, but be sure) */
Sanity check for OpenGL version in Context. I though this check was present since forever, but isn't. Maybe it's hidden somewhere else.
13 years ago
#ifndef MAGNUM_TARGET_GLES
It probably isn't good idea to have version checks as assertions. In particular when building with CORRADE_NO_ASSERT the version doesn't get checked at all, silently failing on null pointer somewhere else.
12 years ago
if(!isVersionSupported(Version::GL210))
Sanity check for OpenGL version in Context. I though this check was present since forever, but isn't. Maybe it's hidden somewhere else.
13 years ago
#elif defined(MAGNUM_TARGET_GLES2)
Hardened checks for OpenGL ES version. There is no ES 2.5 or 3.01 and this way we can properly check for garbage returned by `glGet(GL_VERSION_M*OR)` (e.g. version 2674.-2O8539 would pass here).
12 years ago
if(_version != Version::GLES200)
Sanity check for OpenGL version in Context. I though this check was present since forever, but isn't. Maybe it's hidden somewhere else.
13 years ago
#else
Be less strict in OpenGL ES context version. I didn't yet have time to add more complete support for ES 3.1, but the engine should at least work on it. If the driver reports ES 3.0 or 3.1 it is taken as valid value in both ES2 and ES3 builds (i.e. the engine can work in ES2 mode also on ES3 driver).
12 years ago
if(!isVersionSupported(Version::GLES300))
Sanity check for OpenGL version in Context. I though this check was present since forever, but isn't. Maybe it's hidden somewhere else.
13 years ago
#endif
It probably isn't good idea to have version checks as assertions. In particular when building with CORRADE_NO_ASSERT the version doesn't get checked at all, silently failing on null pointer somewhere else.
12 years ago
{
#ifndef MAGNUM_TARGET_GLES
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
Error{} << "GL::Context: unsupported OpenGL version" << std::make_pair(majorVersion, minorVersion);
It probably isn't good idea to have version checks as assertions. In particular when building with CORRADE_NO_ASSERT the version doesn't get checked at all, silently failing on null pointer somewhere else.
12 years ago
#else
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
Error{} << "GL::Context: unsupported OpenGL ES version" << std::make_pair(majorVersion, minorVersion);
It probably isn't good idea to have version checks as assertions. In particular when building with CORRADE_NO_ASSERT the version doesn't get checked at all, silently failing on null pointer somewhere else.
12 years ago
#endif
Initial support for passing arguments directly to the engine. The parameterless Platform::Context::Context() constructor and Platform::Context::tryCreate() function are deprecated in favor functions that take argc/argv pair. The Context class now accepts arguments starting with --magnum-* prefix. Currently there are none except for the implicit --magnum-help, but that will change with the following commits.
11 years ago
/* Reset the version so the context is not marked as successfully created */
_version = Version::None;
Ability to do GL context creation fallbacks even from Context constructor. In some cases the GL context creation might success without error, but the created version is one that we don't want (e.g. software GDI rasterizer on Windows). Previously the Context class constructor just exited the application and it was impossible to react on that from the application side (for example reducing some context feature requirements). Now there is Platform::Context::tryCreate(), which returns either created instance or `nullptr` if the instance creation failed with some error. That is now used in all Platform::*Application implementations.
11 years ago
return false;
It probably isn't good idea to have version checks as assertions. In particular when building with CORRADE_NO_ASSERT the version doesn't get checked at all, silently failing on null pointer somewhere else.
12 years ago
}
Sanity check for OpenGL version in Context. I though this check was present since forever, but isn't. Maybe it's hidden somewhere else.
13 years ago
Initial support for context flags.
13 years ago
/* Context flags are supported since GL 3.0 */
#ifndef MAGNUM_TARGET_GLES
/**
* @todo According to KHR_debug specs this should be also present in ES2
* if KHR_debug is available, but in headers it is nowhere to be found
*/
if(isVersionSupported(Version::GL300))
glGetIntegerv(GL_CONTEXT_FLAGS, reinterpret_cast<GLint*>(&_flags));
#endif
More robust support for extension-based driver bug workarounds. Previously the extensions were either disabled altogether (e.g. because we don't have yet extension wrangler on ES) or had manually adjusted minimal required versions to avoid issues on older versions (e.g. ARB_explicit_attrib_location is known to give compiler errors when used with GLSL < 1.50 on some drivers), both done basically at compile time unrelated to actual hardware/driver used. Now all the extensions are enabled exactly as the driver advertises them (or when their core version is not larger than the context version) and have minimal required versions as per specification. Given extension is supported in given version when it is marked as such and its minimal required version is not larger than the requested one. The extension disabling is thus done by simply increasing the minimal required version to larger value (or Version::None for disabling it for all versions). Currently no such disabling is put into place, but the existing workarounds scattered all over the place will be gradually converted to this. Simplified the code and tests by marking all extensions from previous versions as supported instead of additional "shorthand" checking whether extension's core version is supported. The check wasn't probably much of a speedup, as it was just another branch. This was also buggy previously, because when the extension would be reported as not being supported in older versions, which is not what we want. Hopefully this won't reintroduce the numerous issues with Mesa and OSX I had in the past :-) Also removed duplicate implementation of Context::isExtensionSupported(), one overload is now calling the other.
12 years ago
/* Get first future (not supported) version */
No `using namespace std` anywhere. Also added missing std:: prefix to remaining cases of std::size_t, std::[u]int[0-9]+_t, std::sin() etc., std::exit().
14 years ago
std::size_t future = 0;
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
while(KnownExtensionsForVersion[future].version != Version::None && isVersionSupported(KnownExtensionsForVersion[future].version))
Classes for handling version and extension information.
14 years ago
++future;
More robust support for extension-based driver bug workarounds. Previously the extensions were either disabled altogether (e.g. because we don't have yet extension wrangler on ES) or had manually adjusted minimal required versions to avoid issues on older versions (e.g. ARB_explicit_attrib_location is known to give compiler errors when used with GLSL < 1.50 on some drivers), both done basically at compile time unrelated to actual hardware/driver used. Now all the extensions are enabled exactly as the driver advertises them (or when their core version is not larger than the context version) and have minimal required versions as per specification. Given extension is supported in given version when it is marked as such and its minimal required version is not larger than the requested one. The extension disabling is thus done by simply increasing the minimal required version to larger value (or Version::None for disabling it for all versions). Currently no such disabling is put into place, but the existing workarounds scattered all over the place will be gradually converted to this. Simplified the code and tests by marking all extensions from previous versions as supported instead of additional "shorthand" checking whether extension's core version is supported. The check wasn't probably much of a speedup, as it was just another branch. This was also buggy previously, because when the extension would be reported as not being supported in older versions, which is not what we want. Hopefully this won't reintroduce the numerous issues with Mesa and OSX I had in the past :-) Also removed duplicate implementation of Context::isExtensionSupported(), one overload is now calling the other.
12 years ago
/* Mark all extensions from past versions as supported */
for(std::size_t i = 0; i != future; ++i)
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
for(const Extension& extension: KnownExtensionsForVersion[i].extensions)
GL: get rid of std::bitset and std::array in Context internals. We have our own lighter versions.
7 years ago
_extensionStatus.set(extension.index(), true);
More robust support for extension-based driver bug workarounds. Previously the extensions were either disabled altogether (e.g. because we don't have yet extension wrangler on ES) or had manually adjusted minimal required versions to avoid issues on older versions (e.g. ARB_explicit_attrib_location is known to give compiler errors when used with GLSL < 1.50 on some drivers), both done basically at compile time unrelated to actual hardware/driver used. Now all the extensions are enabled exactly as the driver advertises them (or when their core version is not larger than the context version) and have minimal required versions as per specification. Given extension is supported in given version when it is marked as such and its minimal required version is not larger than the requested one. The extension disabling is thus done by simply increasing the minimal required version to larger value (or Version::None for disabling it for all versions). Currently no such disabling is put into place, but the existing workarounds scattered all over the place will be gradually converted to this. Simplified the code and tests by marking all extensions from previous versions as supported instead of additional "shorthand" checking whether extension's core version is supported. The check wasn't probably much of a speedup, as it was just another branch. This was also buggy previously, because when the extension would be reported as not being supported in older versions, which is not what we want. Hopefully this won't reintroduce the numerous issues with Mesa and OSX I had in the past :-) Also removed duplicate implementation of Context::isExtensionSupported(), one overload is now calling the other.
12 years ago
Added Context::extensionStrings(). Returns list of _all_ extensions reported by the driver, even those that are not supported in Magnum and also those that are marked as disabled due to driver bugs. Usable mostly when doing fresh port to new platform to discover possible new features. The actual implementation was part of Context constructor (and now this function is called instead). It is rather ugly because we need to take care of both GL 2.1 and GL 3.0 (and also ES2 and ES3).
12 years ago
/* Check for presence of future and vendor extensions */
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
const Containers::Array<Containers::StringView> extensions = extensionStrings();
for(const Containers::StringView extension: extensions) {
if(const Extension* found = findExtension(extension, future)) {
GL: deprecate Context::supportedExtensions(). I don't see a real use case for this API (and don't remember ever using it) and it only causes extra overhead during context creation (and then a ton of useless allocations at runtime).
5 years ago
#ifdef MAGNUM_BUILD_DEPRECATED
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
arrayAppend(_supportedExtensions, *found);
GL: deprecate Context::supportedExtensions(). I don't see a real use case for this API (and don't remember ever using it) and it only causes extra overhead during context creation (and then a ton of useless allocations at runtime).
5 years ago
#endif
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
_extensionStatus.set(found->index(), true);
Context: use glGetString(GL_EXTENSIONS) on GL 2.1. glGetStringi() is since GL 3.0.
14 years ago
}
Added Context::shadingLanguageVersionStrings(), better extension query. Not relying on GL version when asking for extensions, i.e. Mesa might know glGetStringi() even if it reports OpenGL 2.1 only. Similarly in shadingLanguageVersionStrings(), not all HW is capable of 4.3, but GL_NUM_SHADING_LANGUAGE_VERSIONS might be supported on GL3 HW too (not my case, though).
13 years ago
}
Classes for handling version and extension information.
14 years ago
More robust support for extension-based driver bug workarounds. Previously the extensions were either disabled altogether (e.g. because we don't have yet extension wrangler on ES) or had manually adjusted minimal required versions to avoid issues on older versions (e.g. ARB_explicit_attrib_location is known to give compiler errors when used with GLSL < 1.50 on some drivers), both done basically at compile time unrelated to actual hardware/driver used. Now all the extensions are enabled exactly as the driver advertises them (or when their core version is not larger than the context version) and have minimal required versions as per specification. Given extension is supported in given version when it is marked as such and its minimal required version is not larger than the requested one. The extension disabling is thus done by simply increasing the minimal required version to larger value (or Version::None for disabling it for all versions). Currently no such disabling is put into place, but the existing workarounds scattered all over the place will be gradually converted to this. Simplified the code and tests by marking all extensions from previous versions as supported instead of additional "shorthand" checking whether extension's core version is supported. The check wasn't probably much of a speedup, as it was just another branch. This was also buggy previously, because when the extension would be reported as not being supported in older versions, which is not what we want. Hopefully this won't reintroduce the numerous issues with Mesa and OSX I had in the past :-) Also removed duplicate implementation of Context::isExtensionSupported(), one overload is now calling the other.
12 years ago
/* Reset minimal required version to Version::None for whole array */
for(auto& i: _extensionRequiredVersion) i = Version::None;
/* Initialize required versions from extension info */
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
for(const auto& version: KnownExtensionsForVersion)
GL,Audio: use plain old binary search for extension setup. Because the cost of hundreds of tiny allocations when filling a std::unordered_map (and then the cache unfriendliness of the whole thing) definitely overweights the potential speedup by searching in O(1) instead of O(log n). Same approach is chosen in the upcoming Vulkan wrapper, I just backported it here. This strips about 8 kB off the Emscripten build.
6 years ago
for(const Extension& extension: version.extensions)
GL: use public API of the Extension class. No need to friend and use underscored members.
8 years ago
_extensionRequiredVersion[extension.index()] = extension.requiredVersion();
More robust support for extension-based driver bug workarounds. Previously the extensions were either disabled altogether (e.g. because we don't have yet extension wrangler on ES) or had manually adjusted minimal required versions to avoid issues on older versions (e.g. ARB_explicit_attrib_location is known to give compiler errors when used with GLSL < 1.50 on some drivers), both done basically at compile time unrelated to actual hardware/driver used. Now all the extensions are enabled exactly as the driver advertises them (or when their core version is not larger than the context version) and have minimal required versions as per specification. Given extension is supported in given version when it is marked as such and its minimal required version is not larger than the requested one. The extension disabling is thus done by simply increasing the minimal required version to larger value (or Version::None for disabling it for all versions). Currently no such disabling is put into place, but the existing workarounds scattered all over the place will be gradually converted to this. Simplified the code and tests by marking all extensions from previous versions as supported instead of additional "shorthand" checking whether extension's core version is supported. The check wasn't probably much of a speedup, as it was just another branch. This was also buggy previously, because when the extension would be reported as not being supported in older versions, which is not what we want. Hopefully this won't reintroduce the numerous issues with Mesa and OSX I had in the past :-) Also removed duplicate implementation of Context::isExtensionSupported(), one overload is now calling the other.
12 years ago
/* Setup driver workarounds (increase required version for particular
extensions), see Implementation/driverWorkarounds.cpp */
setupDriverWorkarounds();
Classes for handling version and extension information.
14 years ago
/* Set this context as current */
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
CORRADE_ASSERT(!currentContext, "GL::Context: Another context currently active", false);
Introduce BUILD_MULTITHREADED CMake option. Enabled by default, makes the current Magnum context a thread-local variable instead of a global one, so it's possible to have multiple thread-local contexts. Might have some performance implications, that's why it's possible to disable it (but enabled by default is the safer option). GCC 4.7 and Apple platforms don't support thread_local, but __thread does the job too (though on iOS not until Xcode 7.3). Also had to move it to file-local because MSVC doesn't like having thread local variables as part of DLL interface. (And there is *of course* no way to disable exporting one particular member. F' that.)
10 years ago
currentContext = this;
First extension-aware functionality - Buffer::copy(). If EXT_direct_state_access is supported, it uses faster alternative to explicit binding.
14 years ago
Ability to control the amount of console output in engine initialization.
10 years ago
/* Decide whether to print the initialization output or not */
GL: no need for separate Context::InternalFlags anymore. The public Context::Configuration::Flags can supplement that, and that also makes option merging easier.
5 years ago
std::ostream* output = _configurationFlags & Configuration::Flag::QuietLog ? nullptr : Debug::output();
Ability to control the amount of console output in engine initialization.
10 years ago
Print some more info when initializing Context. Then also don't display this information duplicated in magnum-info.
12 years ago
/* Print some info and initialize state tracker (which also prints some
GL: trim vendor string in the engine startup log. Mesa has a space at the end and my OCD goes crazy about that.
7 years ago
more info). Mesa's renderer string has a space at the end, trim that. */
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
Debug{output} << "Renderer:" << rendererString().trimmed() << "by" << vendorString();
Ability to control the amount of console output in engine initialization.
10 years ago
Debug{output} << "OpenGL version:" << versionString();
Ability to disable OpenGL extensions from command-line. Added `--magnum-disable-extensions` option next to the already existing `--magnum-disable-workarounds`. Specified extensions are not used automatically and also report as disabled and unsupported when asked using Context::isExtensionSupported() and Context::isExtensionDisabled().
11 years ago
/* Disable extensions as requested by the user */
if(!_disabledExtensions.empty()) {
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
Debug{output} << "Disabling extensions:";
for(const Extension& extension: _disabledExtensions) {
_extensionRequiredVersion[extension.index()] = Version::None;
Debug{output} << " " << extension.string();
Ability to disable OpenGL extensions from command-line. Added `--magnum-disable-extensions` option next to the already existing `--magnum-disable-workarounds`. Specified extensions are not used automatically and also report as disabled and unsupported when asked using Context::isExtensionSupported() and Context::isExtensionDisabled().
11 years ago
}
}
GL: put all internal Context State substructures in a single allocation. This means that instead of 12 separate allocations we have just one, allocating everything together in a contiguous piece of memory. That should be also a bit more cache friendly when accessing the state as it's not scattered around the memory like crazy. Because there are no Pointer indirections needed anymore, the State members are just references now. That resulted in a lot of sweeping changes around the whole GL library, but they're all trivial, changing `->` to `.`, mostly. There's two more nested allocations in the TextureState struct, will take care of them in a separate commit.
5 years ago
std::pair<Containers::ArrayTuple, Implementation::State&> state = Implementation::State::allocate(*this, output);
_stateData = std::move(state.first);
_state = &state.second;
Private (implementation-only) OpenGL state tracker.
14 years ago
More transparent way to handle driver workarounds. The engine can now list all driver workarounds that were used during the initialization. Any listed workaround can then be disabled from the command-line using `--magnum-disable-workarounds` command-line parameter. The disabling and querying API is private and undocumented, because the driver workarounds should be disabled only by end-users and not application developers. The workaround list is now empty, but will be filled up in the following commits and the workarounds will be probably documented only privately in Implementation/driverSpecific.hpp, as it is really something that should be used only to debug driver problems.
11 years ago
/* Print a list of used workarounds */
Don't display "Using driver workarounds" when there aren't any.
11 years ago
if(!_driverWorkarounds.empty()) {
Ability to control the amount of console output in engine initialization.
10 years ago
Debug{output} << "Using driver workarounds:";
Don't display "Using driver workarounds" when there aren't any.
11 years ago
for(const auto& workaround: _driverWorkarounds)
Fix compilation on not-exactly-recent updates of MSVC 2015.
10 years ago
if(!workaround.second) Debug(output) << " " << workaround.first;
Don't display "Using driver workarounds" when there aren't any.
11 years ago
}
More transparent way to handle driver workarounds. The engine can now list all driver workarounds that were used during the initialization. Any listed workaround can then be disabled from the command-line using `--magnum-disable-workarounds` command-line parameter. The disabling and querying API is private and undocumented, because the driver workarounds should be disabled only by end-users and not application developers. The workaround list is now empty, but will be filled up in the following commits and the workarounds will be probably documented only privately in Implementation/driverSpecific.hpp, as it is really something that should be used only to debug driver problems.
11 years ago
Moved global data from Renderer class to per-context state.
12 years ago
/** @todo Get rid of these */
GL,Platform: introduce GL::Context::Configuration::Flag::Windowless. With this flag set (which is done implicitly for all windowless apps and, conversely, not done for all windowed apps), the default framebuffer state isn't touched in any way, which should avoid potential race conditions with default framebuffer on another thread.
5 years ago
/* Initialize functionality based on current OpenGL version and extensions.
If we are on a windowless context don't touch the default framebuffer
to avoid potential race conditions with default framebuffer on another
thread. */
if(!(_configurationFlags & Configuration::Flag::Windowless))
DefaultFramebuffer::initializeContextBasedFunctionality(*this);
Moved global data from Renderer class to per-context state.
12 years ago
Renderer::initializeContextBasedFunctionality();
Ability to do GL context creation fallbacks even from Context constructor. In some cases the GL context creation might success without error, but the created version is one that we don't want (e.g. software GDI rasterizer on Windows). Previously the Context class constructor just exited the application and it was impossible to react on that from the application side (for example reducing some context feature requirements). Now there is Platform::Context::tryCreate(), which returns either created instance or `nullptr` if the instance creation failed with some error. That is now used in all Platform::*Application implementations.
11 years ago
GL: new --magnum-gpu-validation command-line option. Enables KHR_debug, if available. This should have been here since ages.
7 years ago
/* Enable GPU validation, if requested */
GL: no need for separate Context::InternalFlags anymore. The public Context::Configuration::Flags can supplement that, and that also makes option merging easier.
5 years ago
if(_configurationFlags & Configuration::Flag::GpuValidation) {
GL: new --magnum-gpu-validation command-line option. Enables KHR_debug, if available. This should have been here since ages.
7 years ago
#ifndef MAGNUM_TARGET_WEBGL
GL: what?! why? no.
7 years ago
if(isExtensionSupported<Extensions::KHR::debug>()) {
GL: new --magnum-gpu-validation command-line option. Enables KHR_debug, if available. This should have been here since ages.
7 years ago
Renderer::enable(Renderer::Feature::DebugOutput);
Renderer::enable(Renderer::Feature::DebugOutputSynchronous);
DebugOutput::setDefaultCallback();
GL: warn if using --magnum-gpu-validation w/o debug context on AMD. That's so far the only driver where it matters I think, so not printing the warning elsewhere.
7 years ago
if((detectedDriver() & DetectedDriver::Amd) && !(flags() & Flag::Debug)) {
Warning{} << "GL::Context: GPU validation on AMD drivers requires debug context to work properly";
GL: no need for separate Context::InternalFlags anymore. The public Context::Configuration::Flags can supplement that, and that also makes option merging easier.
5 years ago
} else if(_configurationFlags & Configuration::Flag::VerboseLog) {
GL: print a message about GPU validation in verbose log output.
7 years ago
Debug{} << "GL::Context: enabling GPU validation";
GL: warn if using --magnum-gpu-validation w/o debug context on AMD. That's so far the only driver where it matters I think, so not printing the warning elsewhere.
7 years ago
}
GL: new --magnum-gpu-validation command-line option. Enables KHR_debug, if available. This should have been here since ages.
7 years ago
} else Warning{} << "GL::Context: GPU validation requested, but GL_KHR_debug not supported";
#else
Warning{} << "GL::Context: GPU validation is not available on WebGL";
#endif
}
Ability to do GL context creation fallbacks even from Context constructor. In some cases the GL context creation might success without error, but the created version is one that we don't want (e.g. software GDI rasterizer on Windows). Previously the Context class constructor just exited the application and it was impossible to react on that from the application side (for example reducing some context feature requirements). Now there is Platform::Context::tryCreate(), which returns either created instance or `nullptr` if the instance creation failed with some error. That is now used in all Platform::*Application implementations.
11 years ago
/* Everything okay */
return true;
Classes for handling version and extension information.
14 years ago
}
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
Containers::StringView Context::vendorString() const {
return {reinterpret_cast<const char*>(glGetString(GL_VENDOR)), Containers::StringViewFlag::Global};
Properly handle cases when glGetString() returns nullptr.
10 years ago
}
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
Containers::StringView Context::rendererString() const {
return {reinterpret_cast<const char*>(glGetString(GL_RENDERER)), Containers::StringViewFlag::Global};
Properly handle cases when glGetString() returns nullptr.
10 years ago
}
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
Containers::StringView Context::versionString() const {
return {reinterpret_cast<const char*>(glGetString(GL_VERSION)), Containers::StringViewFlag::Global};
Properly handle cases when glGetString() returns nullptr.
10 years ago
}
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
Containers::StringView Context::shadingLanguageVersionString() const {
return {reinterpret_cast<const char*>(glGetString(GL_SHADING_LANGUAGE_VERSION)), Containers::StringViewFlag::Global};
Properly handle cases when glGetString() returns nullptr.
10 years ago
}
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
Containers::Array<Containers::StringView> Context::shadingLanguageVersionStrings() const {
Added Context::shadingLanguageVersionStrings(), better extension query. Not relying on GL version when asking for extensions, i.e. Mesa might know glGetStringi() even if it reports OpenGL 2.1 only. Similarly in shadingLanguageVersionStrings(), not all HW is capable of 4.3, but GL_NUM_SHADING_LANGUAGE_VERSIONS might be supported on GL3 HW too (not my case, though).
13 years ago
#ifndef MAGNUM_TARGET_GLES
GLint versionCount = 0;
glGetIntegerv(GL_NUM_SHADING_LANGUAGE_VERSIONS, &versionCount);
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
/* If zero, the implementation doesn't yet support this query (< GL4.3) */
/** @todo doesn't this throw a GL error? some better handling? */
if(versionCount) {
/* Get all of them */
Containers::Array<Containers::StringView> versions{std::size_t(versionCount)};
for(GLint i = 0; i != versionCount; ++i)
versions[i] = {reinterpret_cast<const char*>(glGetStringi(GL_SHADING_LANGUAGE_VERSION, i)), Containers::StringViewFlag::Global};
return versions;
}
Added Context::shadingLanguageVersionStrings(), better extension query. Not relying on GL version when asking for extensions, i.e. Mesa might know glGetStringi() even if it reports OpenGL 2.1 only. Similarly in shadingLanguageVersionStrings(), not all HW is capable of 4.3, but GL_NUM_SHADING_LANGUAGE_VERSIONS might be supported on GL3 HW too (not my case, though).
13 years ago
#endif
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
return Containers::array({shadingLanguageVersionString()});
}
Added Context::extensionStrings(). Returns list of _all_ extensions reported by the driver, even those that are not supported in Magnum and also those that are marked as disabled due to driver bugs. Usable mostly when doing fresh port to new platform to discover possible new features. The actual implementation was part of Context constructor (and now this function is called instead). It is rather ugly because we need to take care of both GL 2.1 and GL 3.0 (and also ES2 and ES3).
12 years ago
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
Containers::Array<Containers::StringView> Context::extensionStrings() const {
GL: don't call GL3+ APIs on GL 2.1. Makes Zink sad.
6 years ago
/* If we have GL 3.0 / GLES 3.0 at least, ask the new way. Otherwise don't
even attempt to query GL_NUM_EXTENSIONS as that would cause a GL error
on GL 2.1. Happens with Mesa's zink that's just 2.1 currently (Apr 2020)
even though for other backends Mesa exposes this. */
Use glGetStringi() on Emscripten/WebGL 2. Available in recent versions.
11 years ago
#ifndef MAGNUM_TARGET_GLES2
Added Context::extensionStrings(). Returns list of _all_ extensions reported by the driver, even those that are not supported in Magnum and also those that are marked as disabled due to driver bugs. Usable mostly when doing fresh port to new platform to discover possible new features. The actual implementation was part of Context constructor (and now this function is called instead). It is rather ugly because we need to take care of both GL 2.1 and GL 3.0 (and also ES2 and ES3).
12 years ago
#ifndef MAGNUM_TARGET_GLES3
GL: don't call GL3+ APIs on GL 2.1. Makes Zink sad.
6 years ago
if(isVersionSupported(Version::GL300))
Added Context::extensionStrings(). Returns list of _all_ extensions reported by the driver, even those that are not supported in Magnum and also those that are marked as disabled due to driver bugs. Usable mostly when doing fresh port to new platform to discover possible new features. The actual implementation was part of Context constructor (and now this function is called instead). It is rather ugly because we need to take care of both GL 2.1 and GL 3.0 (and also ES2 and ES3).
12 years ago
#endif
{
GL: don't call GL3+ APIs on GL 2.1. Makes Zink sad.
6 years ago
GLint extensionCount = 0;
glGetIntegerv(GL_NUM_EXTENSIONS, &extensionCount);
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
Containers::Array<Containers::StringView> extensions{std::size_t(extensionCount)};
Added Context::extensionStrings(). Returns list of _all_ extensions reported by the driver, even those that are not supported in Magnum and also those that are marked as disabled due to driver bugs. Usable mostly when doing fresh port to new platform to discover possible new features. The actual implementation was part of Context constructor (and now this function is called instead). It is rather ugly because we need to take care of both GL 2.1 and GL 3.0 (and also ES2 and ES3).
12 years ago
for(GLint i = 0; i != extensionCount; ++i)
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
extensions[i] = {reinterpret_cast<const char*>(glGetStringi(GL_EXTENSIONS, i)), Containers::StringViewFlag::Global};
return extensions;
Added Context::extensionStrings(). Returns list of _all_ extensions reported by the driver, even those that are not supported in Magnum and also those that are marked as disabled due to driver bugs. Usable mostly when doing fresh port to new platform to discover possible new features. The actual implementation was part of Context constructor (and now this function is called instead). It is rather ugly because we need to take care of both GL 2.1 and GL 3.0 (and also ES2 and ES3).
12 years ago
}
#ifndef MAGNUM_TARGET_GLES3
else
#endif
#endif
#ifndef MAGNUM_TARGET_GLES3
/* OpenGL 2.1 / OpenGL ES 2.0 doesn't have glGetStringi() */
GL: rework Context to not use any STL containers. We're going to eventually include this class in all Application classes (need that in order to inherit a to-be-created Configuration class) and the <string> and <vector> would be just too much. This change caused magnum-gl-info.wasm (WebGL 2 build) to go down from 247 to 245 kB. Not much, but that's I guess because there's still a lot other vectors of strings elsewhere. There's a lot more places to clean up, will do those in separate commits. This change is the most atomic I could do, and it introduces a breaking change to all APIs that returned a std::vector or a std::string. Fortunately (or as I hope) those weren't used that much, so it shouldn't cause build breakages for that many people. Quite a lot of the optimization ideas is borrowed from the new Vk library -- such as "interning" the driver workaround strings to avoid allocating their copies.
5 years ago
return Containers::StringView{reinterpret_cast<const char*>(glGetString(GL_EXTENSIONS)), Containers::StringViewFlag::Global}.splitWithoutEmptyParts();
Added Context::extensionStrings(). Returns list of _all_ extensions reported by the driver, even those that are not supported in Magnum and also those that are marked as disabled due to driver bugs. Usable mostly when doing fresh port to new platform to discover possible new features. The actual implementation was part of Context constructor (and now this function is called instead). It is rather ugly because we need to take care of both GL 2.1 and GL 3.0 (and also ES2 and ES3).
12 years ago
#endif
}
Added Context::isCoreProfile(). It's funny when implementing a new feature immediately involves doing a NV-specific workaround. Sigh.
9 years ago
#ifndef MAGNUM_TARGET_GLES
bool Context::isCoreProfile() {
GL: put all internal Context State substructures in a single allocation. This means that instead of 12 separate allocations we have just one, allocating everything together in a contiguous piece of memory. That should be also a bit more cache friendly when accessing the state as it's not scattered around the memory like crazy. Because there are no Pointer indirections needed anymore, the State members are just references now. That resulted in a lot of sweeping changes around the whole GL library, but they're all trivial, changing `->` to `.`, mostly. There's two more nested allocations in the TextureState struct, will take care of them in a separate commit.
5 years ago
return isCoreProfileInternal(_state->context);
Bind some default VAO in case we are on core profile with ARB_VAO disabled. Otherwise it's not possible to render anything. The original behavior was accepted by NVidia, but it's not accepted by Mesa, e.g.
9 years ago
}
bool Context::isCoreProfileInternal(Implementation::ContextState& state) {
Added Context::isCoreProfile(). It's funny when implementing a new feature immediately involves doing a NV-specific workaround. Sigh.
9 years ago
Implementation::ContextState::CoreProfile& value = state.coreProfile;
GL: don't ask for GL_CONTEXT_PROFILE_MASK on GL < 3.2. Makes Zink sad.
6 years ago
if(value == Implementation::ContextState::CoreProfile::Initial) {
/* GL < 3.2 is never a core profile, moreover querying
GL_CONTEXT_PROFILE_MASK would result in a GL error, so don't do
that */
if(_version < Version::GL320)
value = Implementation::ContextState::CoreProfile::Compatibility;
else value = (this->*state.isCoreProfileImplementation)() ?
Added Context::isCoreProfile(). It's funny when implementing a new feature immediately involves doing a NV-specific workaround. Sigh.
9 years ago
Implementation::ContextState::CoreProfile::Core :
Implementation::ContextState::CoreProfile::Compatibility;
GL: don't ask for GL_CONTEXT_PROFILE_MASK on GL < 3.2. Makes Zink sad.
6 years ago
}
Added Context::isCoreProfile(). It's funny when implementing a new feature immediately involves doing a NV-specific workaround. Sigh.
9 years ago
return value == Implementation::ContextState::CoreProfile::Core;
}
bool Context::isCoreProfileImplementationDefault() {
GLint value = 0;
glGetIntegerv(GL_CONTEXT_PROFILE_MASK, &value);
return value & GL_CONTEXT_CORE_PROFILE_BIT;
}
bool Context::isCoreProfileImplementationNV() {
auto extensions = extensionStrings();
return std::find(extensions.begin(), extensions.end(), "GL_ARB_compatibility") == extensions.end();
}
#endif
Make the ES versions first-class on desktop OpenGL. What this changes: * Before, a Version::GLES310 shader was created with `#version 450` directive, which is wrong, because an implementation might support GLES 3.1 but not GL 4.5 so this was not working and also the syntax is a bit different so this wasn't helpful at all. * Similarly, Context::isVersionSupported(Version::GLES310) was checking for support of OpenGL 4.5. Now it just checks for `ARB_ES3_1_compatibility` extension. * Added a small isVersionES() utility that just tells whether given version is ES or not.
10 years ago
bool Context::isVersionSupported(Version version) const {
#ifndef MAGNUM_TARGET_GLES
if(version == Version::GLES200)
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
return isExtensionSupported<Extensions::ARB::ES2_compatibility>();
Make the ES versions first-class on desktop OpenGL. What this changes: * Before, a Version::GLES310 shader was created with `#version 450` directive, which is wrong, because an implementation might support GLES 3.1 but not GL 4.5 so this was not working and also the syntax is a bit different so this wasn't helpful at all. * Similarly, Context::isVersionSupported(Version::GLES310) was checking for support of OpenGL 4.5. Now it just checks for `ARB_ES3_1_compatibility` extension. * Added a small isVersionES() utility that just tells whether given version is ES or not.
10 years ago
if(version == Version::GLES300)
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
return isExtensionSupported<Extensions::ARB::ES3_compatibility>();
Make the ES versions first-class on desktop OpenGL. What this changes: * Before, a Version::GLES310 shader was created with `#version 450` directive, which is wrong, because an implementation might support GLES 3.1 but not GL 4.5 so this was not working and also the syntax is a bit different so this wasn't helpful at all. * Similarly, Context::isVersionSupported(Version::GLES310) was checking for support of OpenGL 4.5. Now it just checks for `ARB_ES3_1_compatibility` extension. * Added a small isVersionES() utility that just tells whether given version is ES or not.
10 years ago
if(version == Version::GLES310)
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
return isExtensionSupported<Extensions::ARB::ES3_1_compatibility>();
Initial support for OpenGL ES 3.2. A bunch of extensions formerly in AEP are now part of ES 3.2, which means they were reordered in the extension lists. While at it, also added corresponding new GL and WebGL extensions and fixed a few wrongly categorized extensions in WebGL.
9 years ago
if(version == Version::GLES320)
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
return isExtensionSupported<Extensions::ARB::ES3_2_compatibility>();
Make the ES versions first-class on desktop OpenGL. What this changes: * Before, a Version::GLES310 shader was created with `#version 450` directive, which is wrong, because an implementation might support GLES 3.1 but not GL 4.5 so this was not working and also the syntax is a bit different so this wasn't helpful at all. * Similarly, Context::isVersionSupported(Version::GLES310) was checking for support of OpenGL 4.5. Now it just checks for `ARB_ES3_1_compatibility` extension. * Added a small isVersionES() utility that just tells whether given version is ES or not.
10 years ago
#endif
return _version >= version;
}
No `using namespace std` anywhere. Also added missing std:: prefix to remaining cases of std::size_t, std::[u]int[0-9]+_t, std::sin() etc., std::exit().
14 years ago
Version Context::supportedVersion(std::initializer_list<Version> versions) const {
Added convenience function Context::supportedVersion().
14 years ago
for(auto version: versions)
if(isVersionSupported(version)) return version;
#ifndef MAGNUM_TARGET_GLES
return Version::GL210;
#else
return Version::GLES200;
#endif
}
Ability to reset internal state tracker.
12 years ago
void Context::resetState(const States states) {
GL: state reset needs to unbind PBOs when entering external code. Qt is not PBO-aware and it messes with its state when resizing a window.
6 years ago
#ifndef MAGNUM_TARGET_GLES2
/* Unbind a PBO (if any) to avoid confusing external GL code that is not
aware of those. Doing this before all buffer state is reset so we can
reuse the knowledge in our state tracker and unbind only if Magnum
actually bound a PBO before. */
if(states & State::UnbindPixelBuffer) {
Buffer::unbindInternal(Buffer::TargetHint::PixelPack);
Buffer::unbindInternal(Buffer::TargetHint::PixelUnpack);
}
#endif
Ability to reset internal state tracker.
12 years ago
if(states & State::Buffers)
GL: put all internal Context State substructures in a single allocation. This means that instead of 12 separate allocations we have just one, allocating everything together in a contiguous piece of memory. That should be also a bit more cache friendly when accessing the state as it's not scattered around the memory like crazy. Because there are no Pointer indirections needed anymore, the State members are just references now. That resulted in a lot of sweeping changes around the whole GL library, but they're all trivial, changing `->` to `.`, mostly. There's two more nested allocations in the TextureState struct, will take care of them in a separate commit.
5 years ago
_state->buffer.reset();
Ability to reset internal state tracker.
12 years ago
if(states & State::Framebuffers)
GL: put all internal Context State substructures in a single allocation. This means that instead of 12 separate allocations we have just one, allocating everything together in a contiguous piece of memory. That should be also a bit more cache friendly when accessing the state as it's not scattered around the memory like crazy. Because there are no Pointer indirections needed anymore, the State members are just references now. That resulted in a lot of sweeping changes around the whole GL library, but they're all trivial, changing `->` to `.`, mostly. There's two more nested allocations in the TextureState struct, will take care of them in a separate commit.
5 years ago
_state->framebuffer.reset();
Ability to reset internal state tracker.
12 years ago
if(states & State::Meshes)
GL: put all internal Context State substructures in a single allocation. This means that instead of 12 separate allocations we have just one, allocating everything together in a contiguous piece of memory. That should be also a bit more cache friendly when accessing the state as it's not scattered around the memory like crazy. Because there are no Pointer indirections needed anymore, the State members are just references now. That resulted in a lot of sweeping changes around the whole GL library, but they're all trivial, changing `->` to `.`, mostly. There's two more nested allocations in the TextureState struct, will take care of them in a separate commit.
5 years ago
_state->mesh.reset();
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
#ifndef MAGNUM_TARGET_GLES
/* Bind a scratch VAO for external GL code that is not VAO-aware and just
enables vertex attributes on the default VAO. Generate it on-demand as
we don't expect this case to be used very often. */
if(states & State::BindScratchVao) {
GL: put all internal Context State substructures in a single allocation. This means that instead of 12 separate allocations we have just one, allocating everything together in a contiguous piece of memory. That should be also a bit more cache friendly when accessing the state as it's not scattered around the memory like crazy. Because there are no Pointer indirections needed anymore, the State members are just references now. That resulted in a lot of sweeping changes around the whole GL library, but they're all trivial, changing `->` to `.`, mostly. There's two more nested allocations in the TextureState struct, will take care of them in a separate commit.
5 years ago
if(!_state->mesh.scratchVAO)
glGenVertexArrays(1, &_state->mesh.scratchVAO);
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
GL: put all internal Context State substructures in a single allocation. This means that instead of 12 separate allocations we have just one, allocating everything together in a contiguous piece of memory. That should be also a bit more cache friendly when accessing the state as it's not scattered around the memory like crazy. Because there are no Pointer indirections needed anymore, the State members are just references now. That resulted in a lot of sweeping changes around the whole GL library, but they're all trivial, changing `->` to `.`, mostly. There's two more nested allocations in the TextureState struct, will take care of them in a separate commit.
5 years ago
_state->mesh.bindVAOImplementation(_state->mesh.scratchVAO);
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
/* Otherwise just unbind the current VAO and leave the the default */
} else
#endif
Improve state tracker resetting functionality.
8 years ago
if(states & State::MeshVao)
GL: put all internal Context State substructures in a single allocation. This means that instead of 12 separate allocations we have just one, allocating everything together in a contiguous piece of memory. That should be also a bit more cache friendly when accessing the state as it's not scattered around the memory like crazy. Because there are no Pointer indirections needed anymore, the State members are just references now. That resulted in a lot of sweeping changes around the whole GL library, but they're all trivial, changing `->` to `.`, mostly. There's two more nested allocations in the TextureState struct, will take care of them in a separate commit.
5 years ago
_state->mesh.bindVAOImplementation(0);
Ability to reset internal state tracker.
12 years ago
Pixel storage support, part 7: state tracking for glPixelStore().
11 years ago
if(states & State::PixelStorage) {
GL: put all internal Context State substructures in a single allocation. This means that instead of 12 separate allocations we have just one, allocating everything together in a contiguous piece of memory. That should be also a bit more cache friendly when accessing the state as it's not scattered around the memory like crazy. Because there are no Pointer indirections needed anymore, the State members are just references now. That resulted in a lot of sweeping changes around the whole GL library, but they're all trivial, changing `->` to `.`, mostly. There's two more nested allocations in the TextureState struct, will take care of them in a separate commit.
5 years ago
_state->renderer.unpackPixelStorage.reset();
_state->renderer.packPixelStorage.reset();
Pixel storage support, part 7: state tracking for glPixelStore().
11 years ago
}
Ability to reset internal state tracker.
12 years ago
/* Nothing to reset for renderer yet */
if(states & State::Shaders) {
/* Nothing to reset for shaders */
GL: put all internal Context State substructures in a single allocation. This means that instead of 12 separate allocations we have just one, allocating everything together in a contiguous piece of memory. That should be also a bit more cache friendly when accessing the state as it's not scattered around the memory like crazy. Because there are no Pointer indirections needed anymore, the State members are just references now. That resulted in a lot of sweeping changes around the whole GL library, but they're all trivial, changing `->` to `.`, mostly. There's two more nested allocations in the TextureState struct, will take care of them in a separate commit.
5 years ago
_state->shaderProgram.reset();
Ability to reset internal state tracker.
12 years ago
}
if(states & State::Textures)
GL: put all internal Context State substructures in a single allocation. This means that instead of 12 separate allocations we have just one, allocating everything together in a contiguous piece of memory. That should be also a bit more cache friendly when accessing the state as it's not scattered around the memory like crazy. Because there are no Pointer indirections needed anymore, the State members are just references now. That resulted in a lot of sweeping changes around the whole GL library, but they're all trivial, changing `->` to `.`, mostly. There's two more nested allocations in the TextureState struct, will take care of them in a separate commit.
5 years ago
_state->texture.reset();
Initial transform feedback implementation. Full support for EXT_transform_feedback, transform feedback objects from ARB_transform_feedback2 and equivalent OpenGL ES 3.0 functionality. Example usage is in src/Magnum/Test/TransformFeedbackGLTest.cpp, I'll add some example later.
12 years ago
#ifndef MAGNUM_TARGET_GLES2
if(states & State::TransformFeedback)
GL: put all internal Context State substructures in a single allocation. This means that instead of 12 separate allocations we have just one, allocating everything together in a contiguous piece of memory. That should be also a bit more cache friendly when accessing the state as it's not scattered around the memory like crazy. Because there are no Pointer indirections needed anymore, the State members are just references now. That resulted in a lot of sweeping changes around the whole GL library, but they're all trivial, changing `->` to `.`, mostly. There's two more nested allocations in the TextureState struct, will take care of them in a separate commit.
5 years ago
_state->transformFeedback.reset();
Initial transform feedback implementation. Full support for EXT_transform_feedback, transform feedback objects from ARB_transform_feedback2 and equivalent OpenGL ES 3.0 functionality. Example usage is in src/Magnum/Test/TransformFeedbackGLTest.cpp, I'll add some example later.
12 years ago
#endif
Ability to reset internal state tracker.
12 years ago
}
GL: introduce Context::Configuration mirroring --magnum-* arguments. Disabling engine startup log or modifying enabled extensions / workarounds from the application side was one of the common pain points and this should *finally* solve the problem. This Configuration is now inherited by the usual Platform::*Application::GLConfiguration / Platform::Windowless*Application::Configuration classes people are used to, so for the end user it's just as if these classes got a bunch new options. Having this, I also extended the ContextGLTest to verify that the Configuration and command-line options do what's expected because that hadn't automated tests until now. The test is mostly a copy of what I did for Vulkan already, nothing special. Additionally all Platform*ApplicationTest executables gained a new --quiet option to verify that the GL::Context::Configuration subset gets correctly passed from the Application code, because that's something we can't really verify in an automated way.
5 years ago
Context::Configuration::Configuration() = default;
Context::Configuration::Configuration(const Configuration& other): _flags{other._flags} {
addDisabledWorkarounds(other._disabledWorkarounds);
addDisabledExtensions(other._disabledExtensions);
}
Context::Configuration::Configuration(Configuration&&) noexcept = default;
Context::Configuration::~Configuration() = default;
Context::Configuration& Context::Configuration::operator=(const Configuration& other) {
_flags = other._flags;
arrayResize(_disabledWorkarounds, 0);
/** @todo arrayClear(), ffs */
arrayResize(_disabledExtensions, Containers::NoInit, 0);
addDisabledWorkarounds(other._disabledWorkarounds);
addDisabledExtensions(other._disabledExtensions);
return *this;
}
Context::Configuration& Context::Configuration::operator=(Configuration&&) noexcept = default;
Containers::ArrayView<const Containers::StringView> Context::Configuration::disabledWorkarounds() const {
return _disabledWorkarounds;
}
Containers::ArrayView<const Extension> Context::Configuration::disabledExtensions() const {
return _disabledExtensions;
}
Context::Configuration& Context::Configuration::addDisabledWorkarounds(std::initializer_list<Containers::StringView> workarounds) {
return addDisabledWorkarounds(Containers::arrayView(workarounds));
}
Context::Configuration& Context::Configuration::addDisabledExtensions(Containers::ArrayView<const Extension> extensions) {
arrayAppend(_disabledExtensions, extensions);
return *this;
}
Context::Configuration& Context::Configuration::addDisabledExtensions(std::initializer_list<Extension> extensions) {
return addDisabledExtensions(Containers::arrayView(extensions));
}
Doc++
12 years ago
#ifndef DOXYGEN_GENERATING_OUTPUT
Debug output for Context::DetectedDriver[s].
9 years ago
#ifndef MAGNUM_TARGET_WEBGL
Use new-style debug operators everywhere.
11 years ago
Debug& operator<<(Debug& debug, const Context::Flag value) {
Reduce string duplication in enum debug output operators. Bloaty says it saved 10 kB in Debug build of MagnumGL: VM SIZE FILE SIZE -------------- -------------- [ = ] 0 .debug_info +1.59Ki +0.0% +0.4% +1.50Ki .text +1.50Ki +0.4% [ = ] 0 .debug_str +409 +0.0% [ = ] 0 .debug_line +276 +0.1% [ = ] 0 .debug_abbrev +20 +0.0% -28.6% -2 [LOAD [RX]] -2 -28.6% [ = ] 0 [Unmapped] -4.28Ki -41.0% -22.7% -9.23Ki .rodata -9.23Ki -22.7% -0.8% -7.73Ki TOTAL -9.73Ki -0.1% And 4 kB in Release: VM SIZE FILE SIZE -------------- -------------- +1.1% +3.44Ki .text +3.44Ki +1.1% +1.7% +1.39Ki .eh_frame +1.39Ki +1.7% [ = ] 0 [Unmapped] +656 +51% -25.5% -9.47Ki .rodata -9.47Ki -25.5% -0.7% -4.64Ki TOTAL -4.00Ki -0.4% That's not negative, so I guess that's good. This change is of course more significant in the context of a minimal WebGL build, where the exe can be as little as 50 kB -- there 4 kB is almost 10% of the size.
7 years ago
debug << "GL::Context::Flag" << Debug::nospace;
Debug output operator for Context::Flag enum.
12 years ago
switch(value) {
Exclude debug operators from code coverage.
10 years ago
/* LCOV_EXCL_START */
Reduce string duplication in enum debug output operators. Bloaty says it saved 10 kB in Debug build of MagnumGL: VM SIZE FILE SIZE -------------- -------------- [ = ] 0 .debug_info +1.59Ki +0.0% +0.4% +1.50Ki .text +1.50Ki +0.4% [ = ] 0 .debug_str +409 +0.0% [ = ] 0 .debug_line +276 +0.1% [ = ] 0 .debug_abbrev +20 +0.0% -28.6% -2 [LOAD [RX]] -2 -28.6% [ = ] 0 [Unmapped] -4.28Ki -41.0% -22.7% -9.23Ki .rodata -9.23Ki -22.7% -0.8% -7.73Ki TOTAL -9.73Ki -0.1% And 4 kB in Release: VM SIZE FILE SIZE -------------- -------------- +1.1% +3.44Ki .text +3.44Ki +1.1% +1.7% +1.39Ki .eh_frame +1.39Ki +1.7% [ = ] 0 [Unmapped] +656 +51% -25.5% -9.47Ki .rodata -9.47Ki -25.5% -0.7% -4.64Ki TOTAL -4.00Ki -0.4% That's not negative, so I guess that's good. This change is of course more significant in the context of a minimal WebGL build, where the exe can be as little as 50 kB -- there 4 kB is almost 10% of the size.
7 years ago
#define _c(value) case Context::Flag::value: return debug << "::" #value;
Debug output operator for Context::Flag enum.
12 years ago
_c(Debug)
GL: exposing GL::Context::Flag::ForwardCompatible.
8 years ago
#ifndef MAGNUM_TARGET_GLES
_c(ForwardCompatible)
#endif
Implemented KHR_no_error.
10 years ago
_c(NoError)
Updates for robust buffer access in ES 3.2.
9 years ago
#ifndef MAGNUM_TARGET_GLES2
Debug output operator for Context::Flag enum.
12 years ago
_c(RobustAccess)
Fix ES build.
12 years ago
#endif
Missing undef.
12 years ago
#undef _c
Exclude debug operators from code coverage.
10 years ago
/* LCOV_EXCL_STOP */
Debug output operator for Context::Flag enum.
12 years ago
}
Reduce string duplication in enum debug output operators. Bloaty says it saved 10 kB in Debug build of MagnumGL: VM SIZE FILE SIZE -------------- -------------- [ = ] 0 .debug_info +1.59Ki +0.0% +0.4% +1.50Ki .text +1.50Ki +0.4% [ = ] 0 .debug_str +409 +0.0% [ = ] 0 .debug_line +276 +0.1% [ = ] 0 .debug_abbrev +20 +0.0% -28.6% -2 [LOAD [RX]] -2 -28.6% [ = ] 0 [Unmapped] -4.28Ki -41.0% -22.7% -9.23Ki .rodata -9.23Ki -22.7% -0.8% -7.73Ki TOTAL -9.73Ki -0.1% And 4 kB in Release: VM SIZE FILE SIZE -------------- -------------- +1.1% +3.44Ki .text +3.44Ki +1.1% +1.7% +1.39Ki .eh_frame +1.39Ki +1.7% [ = ] 0 [Unmapped] +656 +51% -25.5% -9.47Ki .rodata -9.47Ki -25.5% -0.7% -4.64Ki TOTAL -4.00Ki -0.4% That's not negative, so I guess that's good. This change is of course more significant in the context of a minimal WebGL build, where the exe can be as little as 50 kB -- there 4 kB is almost 10% of the size.
7 years ago
return debug << "(" << Debug::nospace << reinterpret_cast<void*>(GLint(value)) << Debug::nospace << ")";
Debug output operator for Context::Flag enum.
12 years ago
}
Debug operator for Context::Flags.
9 years ago
Debug& operator<<(Debug& debug, const Context::Flags value) {
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
return Containers::enumSetDebugOutput(debug, value, "GL::Context::Flags{}", {
Debug operator for Context::Flags.
9 years ago
Context::Flag::Debug,
GL: exposing GL::Context::Flag::ForwardCompatible.
8 years ago
#ifndef MAGNUM_TARGET_GLES
Context::Flag::ForwardCompatible,
#endif
Debug operator for Context::Flags.
9 years ago
Context::Flag::NoError,
Updates for robust buffer access in ES 3.2.
9 years ago
#ifndef MAGNUM_TARGET_GLES2
Debug operator for Context::Flags.
9 years ago
Context::Flag::RobustAccess
#endif
});
}
Doc++
12 years ago
#endif
Debug output for Context::DetectedDriver[s].
9 years ago
Debug& operator<<(Debug& debug, const Context::DetectedDriver value) {
Reduce string duplication in enum debug output operators. Bloaty says it saved 10 kB in Debug build of MagnumGL: VM SIZE FILE SIZE -------------- -------------- [ = ] 0 .debug_info +1.59Ki +0.0% +0.4% +1.50Ki .text +1.50Ki +0.4% [ = ] 0 .debug_str +409 +0.0% [ = ] 0 .debug_line +276 +0.1% [ = ] 0 .debug_abbrev +20 +0.0% -28.6% -2 [LOAD [RX]] -2 -28.6% [ = ] 0 [Unmapped] -4.28Ki -41.0% -22.7% -9.23Ki .rodata -9.23Ki -22.7% -0.8% -7.73Ki TOTAL -9.73Ki -0.1% And 4 kB in Release: VM SIZE FILE SIZE -------------- -------------- +1.1% +3.44Ki .text +3.44Ki +1.1% +1.7% +1.39Ki .eh_frame +1.39Ki +1.7% [ = ] 0 [Unmapped] +656 +51% -25.5% -9.47Ki .rodata -9.47Ki -25.5% -0.7% -4.64Ki TOTAL -4.00Ki -0.4% That's not negative, so I guess that's good. This change is of course more significant in the context of a minimal WebGL build, where the exe can be as little as 50 kB -- there 4 kB is almost 10% of the size.
7 years ago
debug << "GL::Context::DetectedDriver" << Debug::nospace;
Debug output for Context::DetectedDriver[s].
9 years ago
switch(value) {
/* LCOV_EXCL_START */
Reduce string duplication in enum debug output operators. Bloaty says it saved 10 kB in Debug build of MagnumGL: VM SIZE FILE SIZE -------------- -------------- [ = ] 0 .debug_info +1.59Ki +0.0% +0.4% +1.50Ki .text +1.50Ki +0.4% [ = ] 0 .debug_str +409 +0.0% [ = ] 0 .debug_line +276 +0.1% [ = ] 0 .debug_abbrev +20 +0.0% -28.6% -2 [LOAD [RX]] -2 -28.6% [ = ] 0 [Unmapped] -4.28Ki -41.0% -22.7% -9.23Ki .rodata -9.23Ki -22.7% -0.8% -7.73Ki TOTAL -9.73Ki -0.1% And 4 kB in Release: VM SIZE FILE SIZE -------------- -------------- +1.1% +3.44Ki .text +3.44Ki +1.1% +1.7% +1.39Ki .eh_frame +1.39Ki +1.7% [ = ] 0 [Unmapped] +656 +51% -25.5% -9.47Ki .rodata -9.47Ki -25.5% -0.7% -4.64Ki TOTAL -4.00Ki -0.4% That's not negative, so I guess that's good. This change is of course more significant in the context of a minimal WebGL build, where the exe can be as little as 50 kB -- there 4 kB is almost 10% of the size.
7 years ago
#define _c(value) case Context::DetectedDriver::value: return debug << "::" #value;
Debug output for Context::DetectedDriver[s].
9 years ago
#ifndef MAGNUM_TARGET_WEBGL
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
_c(Amd)
#endif
#ifdef MAGNUM_TARGET_GLES
_c(Angle)
#endif
#ifndef MAGNUM_TARGET_WEBGL
Debug output for Context::DetectedDriver[s].
9 years ago
_c(IntelWindows)
_c(Mesa)
_c(NVidia)
Detecting VMware SVGA3D driver. The SVGA3D driver is used on Windows as well, which means Mesa is now detected on Windows too.
9 years ago
_c(Svga3D)
GL: detect SwiftShader.
7 years ago
#ifdef MAGNUM_TARGET_GLES
_c(SwiftShader)
#endif
Detecting VMware SVGA3D driver. The SVGA3D driver is used on Windows as well, which means Mesa is now detected on Windows too.
9 years ago
#endif
GL: detecting ARM Mali drivers on Android.
7 years ago
#ifdef CORRADE_TARGET_ANDROID
_c(ArmMali)
#endif
Debug output for Context::DetectedDriver[s].
9 years ago
#undef _c
/* LCOV_EXCL_STOP */
}
Reduce string duplication in enum debug output operators. Bloaty says it saved 10 kB in Debug build of MagnumGL: VM SIZE FILE SIZE -------------- -------------- [ = ] 0 .debug_info +1.59Ki +0.0% +0.4% +1.50Ki .text +1.50Ki +0.4% [ = ] 0 .debug_str +409 +0.0% [ = ] 0 .debug_line +276 +0.1% [ = ] 0 .debug_abbrev +20 +0.0% -28.6% -2 [LOAD [RX]] -2 -28.6% [ = ] 0 [Unmapped] -4.28Ki -41.0% -22.7% -9.23Ki .rodata -9.23Ki -22.7% -0.8% -7.73Ki TOTAL -9.73Ki -0.1% And 4 kB in Release: VM SIZE FILE SIZE -------------- -------------- +1.1% +3.44Ki .text +3.44Ki +1.1% +1.7% +1.39Ki .eh_frame +1.39Ki +1.7% [ = ] 0 [Unmapped] +656 +51% -25.5% -9.47Ki .rodata -9.47Ki -25.5% -0.7% -4.64Ki TOTAL -4.00Ki -0.4% That's not negative, so I guess that's good. This change is of course more significant in the context of a minimal WebGL build, where the exe can be as little as 50 kB -- there 4 kB is almost 10% of the size.
7 years ago
return debug << "(" << Debug::nospace << reinterpret_cast<void*>(GLint(value)) << Debug::nospace << ")";
Debug output for Context::DetectedDriver[s].
9 years ago
}
Debug& operator<<(Debug& debug, const Context::DetectedDrivers value) {
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
return Containers::enumSetDebugOutput(debug, value, "GL::Context::DetectedDrivers{}", {
Debug output for Context::DetectedDriver[s].
9 years ago
#ifndef MAGNUM_TARGET_WEBGL
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
Context::DetectedDriver::Amd,
#endif
#ifdef MAGNUM_TARGET_GLES
Context::DetectedDriver::Angle,
#endif
#ifndef MAGNUM_TARGET_WEBGL
Debug output for Context::DetectedDriver[s].
9 years ago
Context::DetectedDriver::IntelWindows,
Context::DetectedDriver::Mesa,
Context::DetectedDriver::NVidia,
GL: detect SwiftShader.
7 years ago
Context::DetectedDriver::Svga3D,
#ifdef MAGNUM_TARGET_GLES
Context::DetectedDriver::SwiftShader
Debug output for Context::DetectedDriver[s].
9 years ago
#endif
Detecting VMware SVGA3D driver. The SVGA3D driver is used on Windows as well, which means Mesa is now detected on Windows too.
9 years ago
#endif
Debug output for Context::DetectedDriver[s].
9 years ago
});
}
First-class WebGL support, part 11: no context flags.
11 years ago
#endif
Debug output operator for Context::Flag enum.
12 years ago
Split the OpenGL layer out, pt 4: new Magnum::GL namespace. At the moment just the GL library itself w/o the tests, and without backwards compatibility aliases. The following types were left in the root namespace, despite being in the GL/ directory, as they will get moved back soon: * Image, CompressedImage and their dimensional typedefs * ImageView, CompressedImageView and their dimensional typedefs * PixelStorage Not PixelFormat etc., that one will stay in the GL namespace and a completely new PixelFormat enum will be provided in the root namespace.
8 years ago
}}