[etseidl]

Which issue does this PR close?

• Closes [Discuss] Mac testing on fork branches keeps failing #11976

Rationale for this change

CI checks in contributors' forks are timing out frequently on macOS.

What changes are included in this PR?

• Restrict the macOS workspace test job to upstream runs, preserving its existing lint-only guard.
• Update CI configuration validation and add regression cases for both guards.
• Document the fork exclusions and retained upstream coverage.

Windows, MSRV, and Linux testing are unchanged.

Are these changes tested?

Added configuration regressions covering removal of the macOS guard or either clause. All configuration/routing and fork-status tests pass, along with
spelling and whitespace checks.

AI assistance was used for the workflow, validator, regression tests, and documentation changes.

Are there any significant user-facing changes?

No

[etseidl]

@sdf-jkl I had pi whip this up for me. Can you look it over please 🙏

[Rich-T-kid]

issue + code make sense but do we know the reason as to why the mac runners keep failing? why do they timeout on user forks but not in the merge queue CI?

[etseidl]

issue + code make sense but do we know the reason as to why the mac runners keep failing? why do they timeout on user forks but not in the merge queue CI?

I reran the job when it happened in my fork, so I can't find the log, but @HippoBaro quoted the reason given to him when it happened #11838 (comment)

Due to capacity constraints, jobs targeting macOS arm64 runners may experience longer queue times.

Lack of timeout in the merge queue may be due to using ASF vs free GitHub 🤷 It's all voodoo to me 😅

[sdf-jkl]

gh probably has less macos-latest runners 🤷

Though why they would fail more on fork runners vs ASF's runner is beyond me 😕

[sdf-jkl]

Thanks @etseidl

[HippoBaro]

Here's an example of a CI run that failed. That's where I found the message that @etseidl quoted.

[etseidl]

macos test was skipped https://github.com/etseidl/arrow-rs/actions/runs/40895531684

Thanks all. @Rich-T-kid any objections to merging?

[Rich-T-kid]

Looks good to me, thanks @etseidl