Move convert GL state into renderer backend

This commit is contained in:
2026-06-04 20:27:43 +02:00
parent 51601adf6d
commit 967a15f15f
8 changed files with 172 additions and 25 deletions

View File

@@ -766,7 +766,10 @@ preserving the previous Windows/macOS program-point-size and line-smoothing
enablement while removing the Windows/macOS branch from `App::init`. The
Windows service now delegates the actual OpenGL program-point-size and
line-smooth enable sequence to a tested `pp_renderer_gl` dispatch helper, so
the platform shell no longer owns those backend state tokens.
the platform shell no longer owns those backend state tokens. The legacy
non-Windows fallback now consumes the same helper for retained macOS render
platform hints, keeping the debt-tracked adapter as a thin GL call bridge
rather than a source of backend state policy.
Windows OpenGL debug callback setup now dispatches through `PlatformServices`,
moving Win32 console coloring and debug-break callback behavior into
`WindowsPlatformServices` while keeping other platform adapters as no-ops; the
@@ -1029,7 +1032,9 @@ directly. App OpenGL initialization debug severity, debug output, GL info
string, renderer API viewport/scissor rect conversion, default
depth/program-point/line-smooth state, blend factor/equation, and UI
render-target RGBA8 format tokens are now also cataloged and tested in
`pp_renderer_gl`; the legacy convert command and resize path consume the same
`pp_renderer_gl`; the legacy convert command now applies its depth,
program-point-size, source-alpha blend, and add-equation startup policy through
a tested backend dispatch contract, and the resize path consumes the same
backend-owned mapping. App clear color-buffer masks, default framebuffer
binding, scissor state, and sampler filter/wrap tokens now share that backend
mapping too. OpenGL extension enumeration query tokens used before runtime