[Arham-Begani]

Fixes #172.

Thread timeouts can lose output buffered by capsys, capfd, capsysbinary or capfdbinary when the process exits. With --capture=no, fixture capture can also swallow the timeout diagnostics.

This flushes the active capture fixture before reading captured output, then keeps global and fixture capture disabled while reporting the timeout. The existing stack dump and exit code 1 are preserved.

Nine subprocess tests cover the four fixtures with fd capture and --capture=no, plus running without the capture plugin. They use a real timer and process exit. Upstream CI passed all ten Python 3.10–3.14 Linux/Windows jobs at commit 2252ccb.

[Arham-Begani]

Could a maintainer review the focused thread-timeout capture fix when convenient? At 2252ccb0d3ac8419923040857ef9c46dfddd5ad6, all ten upstream Python 3.10–3.14 Linux/Windows jobs pass.

The change preserves output held by capsys/capfd and their binary variants before the process exits, including --capture=no, while keeping the existing stack dump and exit code. The nine subprocess regressions use the real timer and process exit; eight reproduce the loss before the fix, and the no-capture-plugin control already passes. The scope stays limited to thread-timeout teardown.

No additional code change is proposed here. The implementation, validation and this follow-up used AI assistance, as disclosed in the PR. Thanks for considering the review.

[Arham-Begani]

Following up on the October 20 review request for 2252ccb0. The upstream Test workflow passes, and the thread-timeout output-capture fix and its subprocess regressions are unchanged.

Could a maintainer review when convenient, or let me know if an adjustment is needed?

This follow-up was prepared with AI assistance.

[flub]

If you want to help this issue along please re-write the original into something reasonable human-like. The current description is just way to long and difficult to parse and make sense of.

[Arham-Begani]

Thanks, @flub. The description is now shortened to the problem, fix and regression coverage. The code is unchanged at 2252ccb, and all ten upstream Linux/Windows test jobs pass.

Could you take another look when convenient?

Prepared with AI assistance.

[flub]

Please fixup the changelog. otherwise lgtm

[flub]

Please start a new section, titled X.Y.Z. Can't add something to a previous version.