test(cuda.bindings): restore thread_unsafe marker for cuFile stats level - #2812
Conversation
|
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
github-actions
Bot
added
the
cuda.bindings
|
/ok to test f77b79f |
rwgk
added
bug
ChatGPT Codex Connector
rwgk
added this to the
cuda.bindings 13.5.0 & 12.9.10 milestone
ChatGPT Codex Connector
This comment has been minimized.
This comment has been minimized.
github-actions
Bot
added
the
cuda.pathfinder
|
/ok to test aac6277 |
seberg
left a comment
seberg
left a comment
There was a problem hiding this comment.
Thanks, LGTM. The test feels redundant, I might be tempted to even drop it; but also nothing wrong with keeping it.
| ([SimpleNamespace(metadata={"Name": "unrelated-package"}, version="2.8.0")], False), | ||
| ], | ||
| ) | ||
| @pytest.mark.agent_authored(model="gpt-5.6-sol") |
There was a problem hiding this comment.
This test feels a bit redundant, but maybe some small value for have_distribution getting this logic wrong.
(It might make sense to check on successful load that _is_expected_load_nvidia_dynamic_lib_failure isn't True. But I think the function isn't strict it just says if a failure is acceptable).
There was a problem hiding this comment.
This came from #2809, which I had to include here for the CI to pass, but it was merged before this PR. The net change in this PR is only the one-line addition of the marker (git show 51ec66b after this PR was merged).
I agree the test is more defensive than what a human would write, but it's not wrong, and trimming back the defensiveness takes time, so I often/usually leave it as-is; it's clearly marked as agent-generated.
|
2 participants
Description
Restore
@pytest.mark.thread_unsafe(reason="cuFile stats level is process-global")ontest_set_stats_level.#2782 added the marker because cuFile's stats level is process-global, but it was unintentionally dropped while reconciling the 13.4.x release branch into
mainin #2789. Without it, this test can race with other cuFile stats tests under parallel execution.This restores the exact marker from #2782; no production code changes.
To keep CI viable while #2809 is awaiting merge, this branch also cherry-picks its exact cuTENSOR 2.8 test fix. Once #2809 lands, updating this branch from
mainwill absorb those changes.Testing
Checklist