test(vk): execute startup pool placement and fallback checks #50

Open

mooki wants to merge 3 commits from mooki/Longshot:test/vk-start-blocks-constructors into main

pull from: mooki/Longshot:test/vk-start-blocks-constructors

merge into: MagicBeanDev:main

MagicBeanDev:main

Conversation 4 Commits 3 Files changed 3 +131 -5

mooki commented 2026-11-13 00:12:04 +00:00

First-time contributor

• Copy link
• Copy Markdown

Why

The startup-pool regression check inspected constructor wiring as text but did not execute allocation refusal. This test-only follow-up executes the production constructors, so misplaced initial blocks or a broken empty-pool retry cannot pass unnoticed. It changes no renderer behavior.

What changed

• Compile the real pool constructors against a controlled refusing allocator. Exercise default placement, compaction off, explicit/zero preallocation, retry success, retry failure and cached reuse.
• Keep every existing policy/wiring assertion. Add --source to test a named historical production tree with the same new fixture.
• Document the existing policy and the checks performed here. Preserve the four old Market observations explicitly as historical Hookshot builds, with their original commits and issue links; they are not Longshot game runs.

Flags and settings

No runtime default changes. The new control is test-only; the existing renderer switches below are documented or exercised by the new cases, not newly enabled.

Flag	What it does	Default now	Recommendation	Helps on	Doesn't help on	Untested on
Test --source	New: Select the production tree whose constructors are compiled	Own checkout	On by default: own checkout; override for negative control	Historical/main comparisons	Actual allocation pressure	Fresh Windows alternate-tree run (should help on: Windows source testing)
Test --cxx	Existing, also used by the added constructor unit: Select the C++ compiler	g++; the suite supplies its default compiler	On by default: suite-selected compiler	Portable fixture compilation	Game speed	Fresh Windows execution (should help on: Windows fixture testing)
RX_VK_IMAGE_PREALLOC	Existing policy documented/tested; unchanged: Choose initial 128-MiB blocks in the pool holding sampled textures; zero disables initial allocation	15% of real device-local heap, 2–24 blocks	Unchanged: automatic policy stays default	Correct pool-placement/fallback regression coverage	Emulating physical VRAM or OS paging	New Windows/AMD/real 8-GB-card runs (should help on: No new hardware claim)
RX_VK_COMPACT	Existing switch exercised/documented; unchanged: With zero, sampled textures use the image pool rather than their compacted pool	On	Unchanged: on	Testing both constructor placement paths	A game-free proof of performance	New hardware runtime runs (should help on: No new hardware claim)
RX_VK_TEST_ALLOWED_MB	Existing diagnostic documented; unchanged: Cap the allowance the renderer's eviction policy sees	Unset	Off by default: diagnostic only	Reproducing the historical policy model	Physical VRAM reduction or OS paging	New Windows/AMD runtime runs (should help on: No new hardware claim)

Measurements

Branch state: on main 5eaaf7fb9c548782a3dfabcf89ad37363ee76b1e; head 6b9748a4a890f9b6a87d572c85e554c2c3a42e65; whole game-free suite on this head: 510 passed, 0 failed, 19 skipped (main 5eaaf7fb9c548782a3dfabcf89ad37363ee76b1e: 510 passed, 0 failed, 19 skipped); the commits are unchanged by rebases apart from the results table. The updater test that failed locally on older mains left main together with the in-game updater.

main: d4a69456c9 (2026-11-12)
Baseline: main @ d4a69456
this PR: test/vk-start-blocks-constructors @ 8179355aae

Original handover record: this branch @ 1652ef660014b40c475643723b83bc65e7d11b59; executable test content tested at 8179355aaeff65e2163279b42466f88fdd352d50, followed only by the evidence-documentation commit.

The later main changes add PR-description tooling and a separate DLSS model-settle fix, neither of which changes the pool constructors or this fixture. The measured commits and original observations therefore stay as named; the table conflict preserves both sides' rows, and no game evidence is relabelled.
Hardware: Core i9-13900K, 94 GiB RAM, Arch Linux kernel 7.2.15-1-omarchy; Clang 23.1.3, Python 3.12.16. No GPU used by these checks.
Scene: game-free constructor/policy fixtures and the whole source suite; no save, window or game.
Method: one unmeasured fixture warm-up, then before/after/before/after with fresh HOME directories on the same PC. Report medians of executed-case counts, not speed. Separately run historical-parent/current-source/historical-parent/current-source with the new test's --source option. The whole suite uses --jobs 6 --strict and the repository's default compiler. The base suite was run once for all moved pull requests of this day, on this commit; the branch suite was run once. Those suite receipts are not a two-pair performance comparison.
How to reproduce: run python -I -B tests/test_blocks_at_start.py --cxx clang++ on the named main and branch trees, alternating twice after a warm-up. Run the same new driver with --source selecting b157c3e8 and then the pinned main tree, twice each, to repeat the historical control. The whole suite uses python -I -B tests/run_all.py --jobs 6 --strict.

	main @ d4a69456	this PR @ 8179355a
Existing policy/wiring test, run 1 / run 2	PASS / PASS	PASS / PASS
Real constructor cases executed, run 1 / run 2; median	0 / 0; 0	6 / 6; 6
Whole suite, one run each	507 passed, 2 failed, 20 skipped	508 passed, 1 failed, 20 skipped

Actual focused output (identical in both repetitions):

main's original test:
PASS

this branch's test:
PASS pool constructors: default
PASS pool constructors: compaction-off
PASS pool constructors: override
PASS pool constructors: prealloc-off
PASS pool constructors: fallback
PASS pool constructors: failed-pool
PASS

Why the negative control is older than Longshot: main already contains the fix from Hookshot pull request 2711, commit 1c0cd9803d5951889535edbfadfb2be89d56e432. Its actual parent, b157c3e817666cf8848a85fa993311495a64f885, exists in Longshot's ancestry but predates the rename. The new fixture was run today with --source selecting that parent's production tree; it was not a game build, a current-main failure, or a synthetic source mutation. It compiles and then fails four behavior assertions. The same new fixture passes against current main's source.

Baseline: Hookshot parent @ b157c3e8 for the historical control below; main already contains the fix, so that comparison must use its historical parent.

Historical differential control	Hookshot parent @ b157c3e8	main @ d4a69456
Constructor failures, run 1 / run 2; median	4 / 4; 4	0 / 0; 0

Actual parent output (both runs):

FAIL default: utility->info.minBlockCount == (compact || fail_prealloc ? 0 : wanted)
PASS pool constructors: compaction-off
FAIL override: utility->info.minBlockCount == (compact || fail_prealloc ? 0 : wanted)
PASS pool constructors: prealloc-off
FAIL fallback: calls == 3 && requested[1] == 9 && requested[2] == 0
FAIL failed-pool: calls == 3 && requested[0] == 0 && requested[1] == 9 && requested[2] == 0
4 pool constructor cases failed

Whole-suite output:

main:
507 passed, 2 failed, 20 skipped in 504 s (linux, 6 at once)
branch:
508 passed, 1 failed, 20 skipped in 175 s (linux, 6 at once)

Both suites fail tests/test_updater_asset_digest.py because this PC lacks the local .NET reference pack. Exact diagnostic, with its temporary path omitted:

error NU1100: Unable to resolve 'Microsoft.AspNetCore.App.Ref (= 10.0.13)' for 'net10.0'.

The shared main run also failed two timing-dependent assertions in tests/test_appimage_package_retry.py during heavy load: the enrollment retry sequence and 124 != 0. Its separately recorded isolated follow-up passed all nine cases; the full-suite failure is retained, not erased. This branch's whole suite passes that file. The 20 skips are declared platform/game/GPU exclusions, not missing dependencies hidden by non-strict mode. These elapsed times are receipts under shared load, not a speed claim.

Both Windows-target C++ translation units compile with warnings as errors; they were not run on Windows for this move. The fixture substitutes getenv with test_getenv before every extracted constructor, so no C runtime getenv is called; no warning flags or assertions were relaxed. Historical Hookshot fork run 43268735876, on 185a460b on 2026-11-12, executed this test on Ubuntu and Windows; that is earlier evidence, not a new Longshot Windows run.

Testing

Tested: the paired constructor and historical-parent checks on Arch Linux with Clang, the whole game-free suite on the exact head in Branch state, and both Windows-target translation units. The byte-identical test driver also executed successfully in the original GitHub head's Ubuntu and Windows checks. The historical picture was inspected; no performance measurement is inferred from it.

Not tested: No new Longshot game build/run: this PR changes only tests and documentation. No fresh Windows execution of either C++ translation unit; local cross-compilation is syntax only. No AMD, Steam Deck, physical 8-GB-card behavior, OS paging, long-session or frame-time claim. The historical ordinary Market runs were not repeated after the rename.

Needs other testers:
None for this unchanged test/docs-only scope: the driver is byte-identical to the GitHub head that passed both Windows and Ubuntu source checks, and it adds no hardware-dependent runtime behavior.

Screenshots

Nothing on screen changes in this PR. This inspected historical picture is context for the retained Hookshot results, not proof of a new Longshot game run or continuous flicker behavior.

Risk and rollback

The source-slicing fixture can fail when production constructors are reshaped, and an inaccurate test stand-in could miss a future allocation-policy regression. It does not ship in the game; reverting this test/documentation change rolls it back without changing saves, renderer behavior or defaults.

Reviewer notes

Start with compile_pools() and POOL_FIXTURE in tests/test_blocks_at_start.py: the production constructors are executed, every original assertion remains, and the historical negative control fails behavior assertions after compilation. Windows-target compilation is not fresh Windows execution evidence.

Issues

Original report: https://github.com/NotRexed/Hookshot/issues/2690. Original merged fix: https://github.com/NotRexed/Hookshot/issues/2711. This moves the test follow-up https://github.com/NotRexed/Hookshot/issues/2877. No corresponding issue was present in the forge's public all-issues listing checked on 2026-11-12; no old number is used as a forge closing reference.

Checklist

For everyone

• One topic.
• The title is a Conventional Commit (type(scope): description).
• Rebased on current main.
• python -I -B tests/run_all.py run and passing (any failure that also fails on main is named above).
• Tested in the game, or what wasn't is named under Testing. — Not applicable: tests/docs only; no runtime change.
• Testing filled in: what was tested, what wasn't, and what needs other testers (with test requests).
• Measured: main and this PR at named commits, on the same PC, scene and settings, at least 2 interleaved runs each, the median reported.
• No game files, dumps, build output or generated files.
• third_party/ untouched (a new folder with a PROVENANCE.md instead).
• The license is OK: GPL-3.0-or-later, nothing incompatible and nothing decompiled.

If applicable

• AI assistance recorded with Assisted-by: / Co-Authored-By: trailers.
• Flags noted, and the switch reference updated (docs/development.md, the README). — Existing runtime policy is documented, not changed.
• A row in docs/tested-configurations.md for each measured run.
• A check under tests/ that reads changed source updated in this PR, not weakened.
• Docs updated.

---

Moved from NotRexed/Hookshot#2877.

Assisted-by: GPT-6 Astra
Co-Authored-By: GPT-6 Astra [email protected]

## Why The startup-pool regression check inspected constructor wiring as text but did not execute allocation refusal. This test-only follow-up executes the production constructors, so misplaced initial blocks or a broken empty-pool retry cannot pass unnoticed. It changes no renderer behavior. ## What changed - Compile the real pool constructors against a controlled refusing allocator. Exercise default placement, compaction off, explicit/zero preallocation, retry success, retry failure and cached reuse. - Keep every existing policy/wiring assertion. Add `--source` to test a named historical production tree with the same new fixture. - Document the existing policy and the checks performed here. Preserve the four old Market observations explicitly as **historical Hookshot builds**, with their original commits and issue links; they are not Longshot game runs. ## Flags and settings No runtime default changes. The new control is test-only; the existing renderer switches below are documented or exercised by the new cases, not newly enabled. | Flag | What it does | Default now | Recommendation | Helps on | Doesn't help on | Untested on | |---|---|---|---|---|---|---| | Test `--source` | New: Select the production tree whose constructors are compiled | Own checkout | On by default: own checkout; override for negative control | Historical/main comparisons | Actual allocation pressure | Fresh Windows alternate-tree run (should help on: Windows source testing) | | Test `--cxx` | Existing, also used by the added constructor unit: Select the C++ compiler | `g++`; the suite supplies its default compiler | On by default: suite-selected compiler | Portable fixture compilation | Game speed | Fresh Windows execution (should help on: Windows fixture testing) | | `RX_VK_IMAGE_PREALLOC` | Existing policy documented/tested; unchanged: Choose initial 128-MiB blocks in the pool holding sampled textures; zero disables initial allocation | 15% of real device-local heap, 2–24 blocks | Unchanged: automatic policy stays default | Correct pool-placement/fallback regression coverage | Emulating physical VRAM or OS paging | New Windows/AMD/real 8-GB-card runs (should help on: No new hardware claim) | | `RX_VK_COMPACT` | Existing switch exercised/documented; unchanged: With zero, sampled textures use the image pool rather than their compacted pool | On | Unchanged: on | Testing both constructor placement paths | A game-free proof of performance | New hardware runtime runs (should help on: No new hardware claim) | | `RX_VK_TEST_ALLOWED_MB` | Existing diagnostic documented; unchanged: Cap the allowance the renderer's eviction policy sees | Unset | Off by default: diagnostic only | Reproducing the historical policy model | Physical VRAM reduction or OS paging | New Windows/AMD runtime runs (should help on: No new hardware claim) | ## Measurements <!-- branch-state --> Branch state: on main `5eaaf7fb9c548782a3dfabcf89ad37363ee76b1e`; head `6b9748a4a890f9b6a87d572c85e554c2c3a42e65`; whole game-free suite on this head: `510 passed, 0 failed, 19 skipped` (main `5eaaf7fb9c548782a3dfabcf89ad37363ee76b1e`: `510 passed, 0 failed, 19 skipped`); the commits are unchanged by rebases apart from the results table. The updater test that failed locally on older mains left main together with the in-game updater. <!-- /branch-state --> main: d4a69456c9de5131589af1fcf9a09adba4c403d1 (2026-11-12) Baseline: main @ d4a69456 this PR: test/vk-start-blocks-constructors @ 8179355aaeff65e2163279b42466f88fdd352d50 Original handover record: this branch @ `1652ef660014b40c475643723b83bc65e7d11b59`; executable test content tested at `8179355aaeff65e2163279b42466f88fdd352d50`, followed only by the evidence-documentation commit. The later main changes add PR-description tooling and a separate DLSS model-settle fix, neither of which changes the pool constructors or this fixture. The measured commits and original observations therefore stay as named; the table conflict preserves both sides' rows, and no game evidence is relabelled. Hardware: Core i9-13900K, 94 GiB RAM, Arch Linux kernel 7.2.15-1-omarchy; Clang 23.1.3, Python 3.12.16. No GPU used by these checks. Scene: game-free constructor/policy fixtures and the whole source suite; no save, window or game. Method: one unmeasured fixture warm-up, then before/after/before/after with fresh HOME directories on the same PC. Report medians of executed-case counts, not speed. Separately run historical-parent/current-source/historical-parent/current-source with the new test's `--source` option. The whole suite uses `--jobs 6 --strict` and the repository's default compiler. **The base suite was run once for all moved pull requests of this day, on this commit**; the branch suite was run once. Those suite receipts are not a two-pair performance comparison. How to reproduce: run `python -I -B tests/test_blocks_at_start.py --cxx clang++` on the named main and branch trees, alternating twice after a warm-up. Run the same new driver with `--source` selecting b157c3e8 and then the pinned main tree, twice each, to repeat the historical control. The whole suite uses `python -I -B tests/run_all.py --jobs 6 --strict`. | | main @ d4a69456 | this PR @ 8179355a | |---|---|---| | Existing policy/wiring test, run 1 / run 2 | PASS / PASS | PASS / PASS | | Real constructor cases executed, run 1 / run 2; median | 0 / 0; 0 | 6 / 6; 6 | | Whole suite, one run each | 507 passed, 2 failed, 20 skipped | 508 passed, 1 failed, 20 skipped | Actual focused output (identical in both repetitions): ```text main's original test: PASS this branch's test: PASS pool constructors: default PASS pool constructors: compaction-off PASS pool constructors: override PASS pool constructors: prealloc-off PASS pool constructors: fallback PASS pool constructors: failed-pool PASS ``` **Why the negative control is older than Longshot:** main already contains the fix from [Hookshot pull request 2711](https://github.com/NotRexed/Hookshot/issues/2711), commit `1c0cd9803d5951889535edbfadfb2be89d56e432`. Its actual parent, `b157c3e817666cf8848a85fa993311495a64f885`, exists in Longshot's ancestry but predates the rename. The new fixture was run today with `--source` selecting that parent's production tree; it was **not** a game build, a current-main failure, or a synthetic source mutation. It compiles and then fails four behavior assertions. The same new fixture passes against current main's source. Baseline: Hookshot parent @ b157c3e8 for the historical control below; main already contains the fix, so that comparison must use its historical parent. | Historical differential control | Hookshot parent @ b157c3e8 | main @ d4a69456 | |---|---|---| | Constructor failures, run 1 / run 2; median | 4 / 4; 4 | 0 / 0; 0 | Actual parent output (both runs): ```text FAIL default: utility->info.minBlockCount == (compact || fail_prealloc ? 0 : wanted) PASS pool constructors: compaction-off FAIL override: utility->info.minBlockCount == (compact || fail_prealloc ? 0 : wanted) PASS pool constructors: prealloc-off FAIL fallback: calls == 3 && requested[1] == 9 && requested[2] == 0 FAIL failed-pool: calls == 3 && requested[0] == 0 && requested[1] == 9 && requested[2] == 0 4 pool constructor cases failed ``` Whole-suite output: ```text main: 507 passed, 2 failed, 20 skipped in 504 s (linux, 6 at once) branch: 508 passed, 1 failed, 20 skipped in 175 s (linux, 6 at once) ``` Both suites fail `tests/test_updater_asset_digest.py` because this PC lacks the local .NET reference pack. Exact diagnostic, with its temporary path omitted: ```text error NU1100: Unable to resolve 'Microsoft.AspNetCore.App.Ref (= 10.0.13)' for 'net10.0'. ``` The shared main run also failed two timing-dependent assertions in `tests/test_appimage_package_retry.py` during heavy load: the enrollment retry sequence and `124 != 0`. Its separately recorded isolated follow-up passed all nine cases; the full-suite failure is retained, not erased. This branch's whole suite passes that file. The 20 skips are declared platform/game/GPU exclusions, not missing dependencies hidden by non-strict mode. These elapsed times are receipts under shared load, **not a speed claim**. Both Windows-target C++ translation units compile with warnings as errors; they were not run on Windows for this move. The fixture substitutes `getenv` with `test_getenv` before every extracted constructor, so no C runtime `getenv` is called; no warning flags or assertions were relaxed. Historical [Hookshot fork run 43268735876](https://github.com/mooki000/Hookshot/actions/runs/43268735876), on `185a460b` on 2026-11-12, executed this test on Ubuntu and Windows; that is earlier evidence, not a new Longshot Windows run. ## Testing **Tested:** the paired constructor and historical-parent checks on Arch Linux with Clang, the whole game-free suite on the exact head in Branch state, and both Windows-target translation units. The byte-identical test driver also executed successfully in the original GitHub head's Ubuntu and Windows checks. The historical picture was inspected; no performance measurement is inferred from it. **Not tested:** No new Longshot game build/run: this PR changes only tests and documentation. No fresh Windows execution of either C++ translation unit; local cross-compilation is syntax only. No AMD, Steam Deck, physical 8-GB-card behavior, OS paging, long-session or frame-time claim. The historical ordinary Market runs were not repeated after the rename. **Needs other testers:** None for this unchanged test/docs-only scope: the driver is byte-identical to the GitHub head that passed both Windows and Ubuntu source checks, and it adds no hardware-dependent runtime behavior. ## Screenshots Nothing on screen changes in this PR. This inspected historical picture is context for the retained Hookshot results, not proof of a new Longshot game run or continuous flicker behavior.  ## Risk and rollback The source-slicing fixture can fail when production constructors are reshaped, and an inaccurate test stand-in could miss a future allocation-policy regression. It does not ship in the game; reverting this test/documentation change rolls it back without changing saves, renderer behavior or defaults. ## Reviewer notes Start with `compile_pools()` and `POOL_FIXTURE` in `tests/test_blocks_at_start.py`: the production constructors are executed, every original assertion remains, and the historical negative control fails behavior assertions after compilation. Windows-target compilation is not fresh Windows execution evidence. ## Issues Original report: https://github.com/NotRexed/Hookshot/issues/2690. Original merged fix: https://github.com/NotRexed/Hookshot/issues/2711. This moves the test follow-up https://github.com/NotRexed/Hookshot/issues/2877. No corresponding issue was present in the forge's public all-issues listing checked on 2026-11-12; no old number is used as a forge closing reference. ## Checklist ### For everyone - [x] One topic. - [x] The title is a Conventional Commit (`type(scope): description`). - [x] Rebased on current `main`. - [x] `python -I -B tests/run_all.py` run and passing (any failure that also fails on `main` is named above). - [x] Tested in the game, or what wasn't is named under Testing. — Not applicable: tests/docs only; no runtime change. - [x] Testing filled in: what was tested, what wasn't, and what needs other testers (with test requests). - [x] Measured: main and this PR at named commits, on the same PC, scene and settings, at least 2 interleaved runs each, the median reported. - [x] No game files, dumps, build output or generated files. - [x] `third_party/` untouched (a new folder with a `PROVENANCE.md` instead). - [x] The license is OK: GPL-3.0-or-later, nothing incompatible and nothing decompiled. ### If applicable - [x] AI assistance recorded with `Assisted-by:` / `Co-Authored-By:` trailers. - [x] Flags noted, and the switch reference updated (docs/development.md, the README). — Existing runtime policy is documented, not changed. - [x] A row in `docs/tested-configurations.md` for each measured run. - [x] A check under `tests/` that reads changed source updated in this PR, not weakened. - [x] Docs updated. --- Moved from NotRexed/Hookshot#2877. Assisted-by: GPT-6 Astra Co-Authored-By: GPT-6 Astra <[email protected]>

startup-pool-history.jpg

1 MiB

mooki added 3 commits 2026-11-13 00:12:04 +00:00

test(vk): exercise startup pool placement and allocation fallback ... ae38c152c8

Keep the numeric and wiring checks from hckrman101. Run the real pool constructors against a refusing allocator, covering compaction off, overrides, no preallocation, retry without blocks and total failure. Keep both translation units reachable before execution for Windows-target preflight. Document the kept-block policy and the limits of the small-card model.

Assisted-by: GPT-6 Astra
Co-Authored-By: GPT-6 Astra <[email protected]>

docs(vk): record startup pool placement under the 8 GB model ... 8179355aae

Record the four ordinary Market memory and picture checks at their actual DLAA K render size. Distinguish these shared-card runs from timing evidence and retain both model limits and the explicit mirror limit.

Assisted-by: GPT-6 Astra
Co-Authored-By: GPT-6 Astra <[email protected]>

docs(test): record Longshot constructor coverage and historical provenance ... 1652ef6600

Record the paired constructor checks and whole source suite, keep the pre-rename game runs explicitly named Hookshot, and use their original issue URLs.

Assisted-by: GPT-6 Astra
Co-Authored-By: GPT-6 Astra <[email protected]>

mooki commented 2026-11-13 00:12:05 +00:00

Author

First-time contributor

• Copy link
• Copy Markdown

Discussion carried over from NotRexed/Hookshot#2877

Every comment that was readable on the original thread when it was copied on 2026-11-12, in order, unchanged except that issue and pull request numbers are written as links to the old repository and names carry no @. Comments by an account that has since been closed there could no longer be read and are missing; where a reply quotes them, the quote remains.

---

mooki000, 2026-11-12 18:42 UTC (comment, original):

Rebased onto main 3c2d80cf (77691c2a): the only conflict was the shared results table, where both sides' rows are kept; the test commit is unchanged (range-diff shows only moved context in the docs), and tests/test_blocks_at_start.py passes again locally.

## Discussion carried over from NotRexed/Hookshot#2877 Every comment that was readable on [the original thread](https://github.com/NotRexed/Hookshot/pull/2877) when it was copied on 2026-11-12, in order, unchanged except that issue and pull request numbers are written as links to the old repository and names carry no @. Comments by an account that has since been closed there could no longer be read and are missing; where a reply quotes them, the quote remains. --- **mooki000**, 2026-11-12 18:42 UTC (comment, [original](https://github.com/NotRexed/Hookshot/pull/2877#issuecomment-6832667254)): Rebased onto main `3c2d80cf` (`77691c2a`): the only conflict was the shared results table, where both sides' rows are kept; the test commit is unchanged (range-diff shows only moved context in the docs), and `tests/test_blocks_at_start.py` passes again locally.

mooki changed title from test(vk): execute startup pool placement and fallback checks to WIP: test(vk): execute startup pool placement and fallback checks 2026-11-13 00:58:47 +00:00

mooki commented 2026-11-13 00:58:47 +00:00

Author

First-time contributor

• Copy link
• Copy Markdown

Back to work in progress, for its description only. This was opened shortly before #48 merged and follows the earlier template, so it does not pass the new description check (no Flags and settings and no Risk and rollback section, table columns named Before and After). The code and the measurements are unchanged: the whole suite passes on this exact commit apart from one local .NET failure that also fails on main. The description is being moved to the new template and the branch rebased onto d725b363; it comes back through pr ready once the check passes.

Back to work in progress, for its description only. This was opened shortly before #48 merged and follows the earlier template, so it does not pass the new description check (no Flags and settings and no Risk and rollback section, table columns named Before and After). The code and the measurements are unchanged: the whole suite passes on this exact commit apart from one local .NET failure that also fails on main. The description is being moved to the new template and the branch rebased onto d725b363; it comes back through `pr ready` once the check passes.

mooki commented 2026-11-13 02:30:56 +00:00

Author

First-time contributor

• Copy link
• Copy Markdown

Updated the description to the current template, including settings, Testing and rollback, and kept the uploaded picture and the historical measurement record. Rebased onto main f41db065c keeping both sides' rows of the results table; the test code is unchanged. Whole suite on the new head 5921a0d2a: 509 passed, 0 failed, 19 skipped. Ready for review.

Updated the description to the current template, including settings, Testing and rollback, and kept the uploaded picture and the historical measurement record. Rebased onto main f41db065c keeping both sides' rows of the results table; the test code is unchanged. Whole suite on the new head 5921a0d2a: 509 passed, 0 failed, 19 skipped. Ready for review.

mooki changed title from WIP: test(vk): execute startup pool placement and fallback checks to test(vk): execute startup pool placement and fallback checks 2026-11-13 02:30:57 +00:00

mooki force-pushed test/vk-start-blocks-constructors from 1652ef6600 to 5921a0d2a5 2026-11-13 02:30:58 +00:00 Compare

mooki referenced this pull request 2026-11-13 02:36:25 +00:00

WIP: fix(docs): give tested runs independent result files #96

mooki commented 2026-11-13 02:58:27 +00:00

Author

First-time contributor

• Copy link
• Copy Markdown

Rebased onto main 5eaaf7fb9 after its pull request 77 added a switch row next to this one's rows in docs/development.md: both kept, this pull request's changes are otherwise identical. Whole suite on the new head 6b9748a4a: 510 passed, 0 failed, 19 skipped. The description follows the newest template (only the boxes that apply, the flags table's new columns).

Rebased onto main 5eaaf7fb9 after its pull request 77 added a switch row next to this one's rows in docs/development.md: both kept, this pull request's changes are otherwise identical. Whole suite on the new head 6b9748a4a: 510 passed, 0 failed, 19 skipped. The description follows the newest template (only the boxes that apply, the flags table's new columns).

mooki force-pushed test/vk-start-blocks-constructors from 5921a0d2a5 to 6b9748a4a8 2026-11-13 02:58:27 +00:00 Compare

Some checks are pending

Hide checks

PR description / Description follows the template (pull_request_target) Waiting to run

Details

Linux AppImage / appimage (pull_request) Blocked by required conditions

Details

Windows builder / Longshot Builder zip (pull_request) Blocked by required conditions

Details

Windows builder / Windows Vulkan back end (llvm-mingw) (pull_request) Blocked by required conditions

Details

Windows builder / dav1d for Windows (llvm-mingw) (pull_request) Blocked by required conditions

Details

CI / Conventional Commit title (pull_request) Blocked by required conditions

Details

CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Blocked by required conditions

Details

CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Blocked by required conditions

Details

CI / Python source checks (windows-2022, part 1/2) (pull_request) Blocked by required conditions

Details

CI / Python source checks (windows-2022, part 2/2) (pull_request) Blocked by required conditions

Details

CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Blocked by required conditions

Details

CI / Workflow lint (actionlint) (pull_request) Blocked by required conditions

Details

Third-party parts / As their notes say (pull_request) Blocked by required conditions

Details

Third-party parts / The same as upstream (pull_request) Blocked by required conditions

Details

This pull request doesn't have enough approvals yet. 0 of 1 approvals granted.

Some workflows are waiting to be reviewed.

This branch is out-of-date with the base branch

You are not authorized to merge this pull request.

View command line instructions

Checkout

From your project repository, check out a new branch and test the changes.

git fetch -u 'test/vk-start-blocks-constructors:mooki-test/vk-start-blocks-constructors'

git switch 'mooki-test/vk-start-blocks-constructors'

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

accessibility

Barrier affecting people with disabilities

  

autorelease: pending

  

autorelease: tagged

  

bug

Something isn't working

  

dependencies

Pull requests that update a dependency file

  

documentation

Improvements or additions to documentation

  

duplicate

This issue or pull request already exists

  

enhancement

New feature or request

  

github_actions

Pull requests that update GitHub Actions code

  

good first issue

Good for newcomers

  

help wanted

Extra attention is needed

  

invalid

This doesn't seem right

  

needs measurements

Waiting for before/after measurements

  

question

Further information is requested

  

testing wanted

Wanted: a test on other hardware (posts a test request in #test-requests)

  

wontfix

This will not be worked on

No labels

accessibility

autorelease: pending

autorelease: tagged

bug

dependencies

documentation

duplicate

enhancement

github_actions

good first issue

help wanted

invalid

needs measurements

question

testing wanted

wontfix

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

MagicBeanDev/Longshot!50

No description provided.