Matthew Stanley b509e8ffa7 gl-renderer phase 2b step 1: modern-GL loader + shader present pipeline ...

Replaces the legacy immediate-mode present quad with a modern-GL pipeline:
SDL_GL_GetProcAddress loader for the GL 2.0+/3.0 entry points, a compiled
vertex+fragment present program, and a fullscreen-triangle draw via an empty
VAO. Verified: Tomba renders correctly via the shader path (log "GL present
path = shader", GL 4.6 NVIDIA). On any loader/compile/link failure it degrades
to the immediate-mode quad, so present never black-screens.

Rasterization still delegates to software (vtable unchanged) — this isolates
and validates the shader/VAO plumbing that the GPU rasterizer builds on next:
VRAM as a GL texture + an FBO at internal resolution, primitive shaders.

Avoids GLchar/GLsizeiptr (absent in MinGW GL 1.1 headers) by spelling loaded
prototypes with char/ptrdiff_t (ABI-identical).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

2026-06-11 11:50:22 -07:00

..

include

gl-renderer phase 2a: real GL context + VRAM-texture present

2026-06-11 11:34:46 -07:00

src

gl-renderer phase 2b step 1: modern-GL loader + shader present pipeline

2026-06-11 11:50:22 -07:00

tests

fix: Bug B hard freeze — TOCTOU underflow in frame pacing slept ~24.7 days

2026-06-09 20:09:25 -07:00

CMakeLists.txt

fix: Bug B hard freeze — TOCTOU underflow in frame pacing slept ~24.7 days

2026-06-09 20:09:25 -07:00

runtime.cmake

gl-renderer phase 2a: real GL context + VRAM-texture present

2026-06-11 11:34:46 -07:00