tests/run_all.py on a current Linux (Clang 23, only a .NET 10 SDK): three checks fail on main, and the crash fixtures leave core dumps #2334

Closed

opened 2026-11-11 15:10:01 +00:00 by mooki000 · 2 comments

mooki000 commented

2026-11-11 15:10:01 +00:00

(Migrated from github.com)

Builder version

main 49f75dc8

How you built

Command line (the game-free checks: python -I -B tests/run_all.py)

The error

tests/test_wayland_libdecor_hint.py   harness.cpp:25:12: error: variable 'sdl_ready' set but not used [-Werror,-Wunused-but-set-global]
tests/test_updater_asset_digest.py    Fixture.csproj : error NU1100: Unable to resolve 'Microsoft.NETCore.App.Ref (= 8.0.32)' for 'net8.0'.
tests/test_make_game_cache.py         Unexpected process launch   (3 of its 11 checks)

Four things a developer on a current Linux meets on main, none of them a fault of the game:

test_wayland_libdecor_hint.py fails with Clang 23, whose -Wall now includes -Wunused-but-set-global; the fixture is built with -Werror and sets a global it never reads. run_all.py takes clang++ before g++. (#2290)
test_updater_asset_digest.py fails where the only .NET SDK is newer than 8: its fixture targets net8.0 with NuGet's sources cleared, and that SDK has no 8.0 reference pack. .NET 8's support ended on 2026-11-10. (#2294)
test_make_game_cache.py fails in every clone that holds the play profile in its work/ folder, where CONTRIBUTING.md tells developers to put it: make_game starts clang to see whether it can read the profile, and the check refuses every process launch. (#2291)
Every run of the checks leaves three core dumps, and with a crash notifier three "a program crashed" messages: three fixtures crash on purpose and the kernel hands each crash to the core-dump handler. One machine that ran the checks many times in a night had 107 such notifications. (#2293)

Operating system and free space on the output drive

Arch Linux, Clang 23.1.3, GCC 16.2.1, .NET SDK 10.0.113, Python 3.14.8, systemd-coredump

Claim

I'm working on these: #2290, #2294, #2291 and #2293 are open, one topic each, tests only; nothing in the game or the builder changes. By docs/workflow.md: S3 (1) x reach (developers and agents who run the checks on Linux with a current toolchain) x confidence 1.0 (each fails on main here and passes with its change). I can test on Linux with an RTX 4090 (NVIDIA 615.78.08) and an Intel UHD 770 (Mesa 26.2.5), not on Windows or AMD.

### Builder version main 49f75dc8 ### How you built Command line (the game-free checks: `python -I -B tests/run_all.py`) ### The error ```text tests/test_wayland_libdecor_hint.py harness.cpp:25:12: error: variable 'sdl_ready' set but not used [-Werror,-Wunused-but-set-global] tests/test_updater_asset_digest.py Fixture.csproj : error NU1100: Unable to resolve 'Microsoft.NETCore.App.Ref (= 8.0.32)' for 'net8.0'. tests/test_make_game_cache.py Unexpected process launch (3 of its 11 checks) ``` Four things a developer on a current Linux meets on main, none of them a fault of the game: 1. `test_wayland_libdecor_hint.py` fails with Clang 23, whose `-Wall` now includes `-Wunused-but-set-global`; the fixture is built with `-Werror` and sets a global it never reads. `run_all.py` takes `clang++` before `g++`. (#2290) 2. `test_updater_asset_digest.py` fails where the only .NET SDK is newer than 8: its fixture targets net8.0 with NuGet's sources cleared, and that SDK has no 8.0 reference pack. .NET 8's support ended on 2026-11-10. (#2294) 3. `test_make_game_cache.py` fails in every clone that holds the play profile in its `work/` folder, where CONTRIBUTING.md tells developers to put it: make_game starts clang to see whether it can read the profile, and the check refuses every process launch. (#2291) 4. Every run of the checks leaves three core dumps, and with a crash notifier three "a program crashed" messages: three fixtures crash on purpose and the kernel hands each crash to the core-dump handler. One machine that ran the checks many times in a night had 107 such notifications. (#2293) ### Operating system and free space on the output drive Arch Linux, Clang 23.1.3, GCC 16.2.1, .NET SDK 10.0.113, Python 3.14.8, systemd-coredump ### Claim I'm working on these: #2290, #2294, #2291 and #2293 are open, one topic each, tests only; nothing in the game or the builder changes. By `docs/workflow.md`: S3 (1) x reach (developers and agents who run the checks on Linux with a current toolchain) x confidence 1.0 (each fails on main here and passes with its change). I can test on Linux with an RTX 4090 (NVIDIA 615.78.08) and an Intel UHD 770 (Mesa 26.2.5), not on Windows or AMD.

NotRexed commented

2026-11-11 21:28:57 +00:00

(Migrated from github.com)

Thanks for the report. Reports of the same problem are now collected in #1000, which lists this one with your details and links back here, so one fix closes them together and the open list stays readable. Follow #1000 for the fix, and add new details there (or here). If this turns out to be a separate problem it will be reopened.

mooki000 commented

2026-11-11 22:03:29 +00:00

(Migrated from github.com)

The four items in this report are now covered by main: #2171 for the Clang/libdecor fixture, #2294 for the local later-SDK fallback, #2291 for cache-test isolation, and #2293 for the deliberate crash fixtures. On clean main 940eb9ef, Clang 23.1.3 passes the libdecor check’s three cases without #2290, so that PR is being closed as superseded. This does not reopen the report or change its consolidation into #1000.

The four items in this report are now covered by main: #2171 for the Clang/libdecor fixture, #2294 for the local later-SDK fallback, #2291 for cache-test isolation, and #2293 for the deliberate crash fixtures. On clean main `940eb9ef`, Clang 23.1.3 passes the libdecor check’s three cases without #2290, so that PR is being closed as superseded. This does not reopen the report or change its consolidation into #1000.

Commenting is not possible because the repository is archived.