test: align output sizing with the render-target pool floor #1254

Closed
monsterlag wants to merge 1 commit from fix/output-pool-contract into main
monsterlag commented 2026-11-09 18:58:24 +00:00 (Migrated from github.com)

The output-resolution integration test still expected the old 305 MiB pool for non-4K modes. Merged #1202 deliberately raised that default minimum to 512 MiB while preserving larger scaled pools and explicit overrides. Update that exact expectation and explain its source; all seven resolution/menu cases and their other assertions remain unchanged.

Validation:

  • Windows LLVM-MinGW and Linux g++: targeted strict runner: 2 passed, 0 failed, 0 skipped on each platform, exercising the seven output-resolution cases and existing 12-case pool fixture.
  • The original integration test fails on its first Windows case. The unchanged 12-case fixture rejects the exact pre-#1202 sizing function on both platforms at the three below-floor cases; scale and override controls remain intact.

This corrects test coverage only. It changes no runtime allocation or performance behavior and makes no GPU/game validation claim.

Hosted CI on fff881d: all eight validation checks passed, including Windows and Linux source checks, Windows renderer and Builder, and Linux AppImage. The two release-publishing jobs were skipped.

The output-resolution integration test still expected the old 305 MiB pool for non-4K modes. Merged #1202 deliberately raised that default minimum to 512 MiB while preserving larger scaled pools and explicit overrides. Update that exact expectation and explain its source; all seven resolution/menu cases and their other assertions remain unchanged. Validation: - Windows LLVM-MinGW and Linux g++: targeted strict runner: 2 passed, 0 failed, 0 skipped on each platform, exercising the seven output-resolution cases and existing 12-case pool fixture. - The original integration test fails on its first Windows case. The unchanged 12-case fixture rejects the exact pre-#1202 sizing function on both platforms at the three below-floor cases; scale and override controls remain intact. This corrects test coverage only. It changes no runtime allocation or performance behavior and makes no GPU/game validation claim. Hosted CI on `fff881d`: all eight validation checks passed, including [Windows and Linux source checks](https://github.com/NotRexed/Hookshot/actions/runs/42678861405), [Windows renderer and Builder](https://github.com/NotRexed/Hookshot/actions/runs/42678861080), and [Linux AppImage](https://github.com/NotRexed/Hookshot/actions/runs/42678861395). The two release-publishing jobs were skipped.
NotRexed commented 2026-11-09 19:19:35 +00:00 (Migrated from github.com)

This is in main now: the same 512 MiB expectation in tests/test_output_resolution.py landed with a1c4bcd (the merge of #841), so closing. Thanks!

This is in `main` now: the same 512 MiB expectation in `tests/test_output_resolution.py` landed with a1c4bcd (the merge of #841), so closing. Thanks!
Commenting is not possible because the repository is archived.
No description provided.