[thomhurst]

I went through the CI failures from the last day. Most were already fixed on main. Two problems were still open, and this PR fixes both.

1. main does not compile (CS1061)

#7195 renamed TestDetails.FailedDataRowObjects to UnreceivedDataRowObjects. #7191 merged after it and still used the old name in TestArgumentRegistrationService.TrackSkippedTestArguments:

src/TUnit.Engine/Services/TestArgumentRegistrationService.cs(54,25): error CS1061: 'TestDetails' does not contain a definition for 'FailedDataRowObjects'

Since then, CloudShop Example Tests has failed at its Build step on every push to main (runs 39408482674 and 39408696640). This PR switches the reference to the new name.

2. macOS PublishTestProjectAOTModule hits the 30-minute module timeout

Since #7131, the AOT test-app publish builds into its own artifacts tree and runs alongside the test modules. On the 3-core macOS runner it now takes 24.5–26.6 minutes even when it passes (jobs 118552950835, 118517073607, 118489901581). The default 30-minute ModularPipelines timeout killed it on six PR runs on 2026-10-20: 39364097783, 39378625835, 39380563945, 39396577708, 39399788520 and 39399993532. Before #7131 the publish took 5.5–8 minutes.

This PR sets the module's timeout to 60 minutes. The job has no timeout-minutes, so the job-level limit does not get in the way. A cheaper macOS publish (less contention) would be a better long-term fix, but the timeout stops healthy runs from failing in the meantime.

Already fixed on main (no change here)

• ParamsParameter in the assertions source-generator tests on Windows: Fix net472 compile gate failure in assertions source generator tests #7170
• ExecutionTimeTests.Completes_Within_Unhappy2 and the WaitsFor_* cancellation tests' wall-clock budgets on macOS: Flaky: ExecutionTimeTests.Completes_Within_Unhappy* exceed 1s wall-clock bound on macOS CI #7159 and Flaky test: WaitsFor_Propagates_OCE_When_Predicate_Observes_Supplied_Token exceeds 500ms budget on macOS CI #7169
• KeyedNotInParallelCrossPhaseTests and ConstraintKeySchedulerConcurrencyTests being killed at 30 s: test(engine): stop whole-process budgets killing healthy scheduling engine tests #7177

Seen once, not fixed

• Run 39399854072, macOS: the job exited with code 137 (SIGKILL) partway through. This is probably memory pressure from the same overlap.
• Run 39380335858: GitVersion threw InvalidOperationException in MainlineVersionStrategy.GetCommitsWasBranchedFrom on all three OSes for a single PR push.

Validation

• TUnit.Engine and TUnit.Pipeline build.
• I ran the TUnit.TestProject bug tests for Data-source objects for skipped tests/rows are never disposed #7156 (skipped rows), Data-source objects for rows that end as failed tests are never disposed #7123 (failed rows) and First class-data fetch per class row creates an untracked instance that is never disposed (SharedType.None) #7157 in source-generated and reflection modes. All match the counts their engine tests expect.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-20T11:12:09.898116Z	47a0e0d	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[coderabbitai]

Note

Currently processing new changes in this PR. This may take a few minutes, please wait...

⚙️ Run configuration

• Configuration used: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 97dd3b9a-68f5-4564-aaa9-0409f3aeaf42

📥 Commits

Reviewing files that changed from the base of the PR and between 6043ab8 and 47a0e0d.

📒 Files selected for processing: 2

• src/TUnit.Engine/Services/TestArgumentRegistrationService.cs
• tools/TUnit.Pipeline/Modules/PublishTestProjectAOTModule.cs

+--------------------------------------------------+
| Brb...ordering more GPUs. CPUs are so last year. |
+--------------------------------------------------+
  \
   \   (\__/)
       (•ㅅ•)
       / 　 づ

✨ Finishing Touches

📝 Generate docstrings

• Commit to this branch
• Create a new PR

🧪 Generate unit tests (beta)

• Commit to this branch
• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

Review

Small, focused PR. I checked both changes.

1. CS1061 fix: This switches FailedDataRowObjects to UnreceivedDataRowObjects in TestArgumentRegistrationService. On the base checkout, that is the only remaining use of the old name under src. The change is correct and restores the build.

2. AOT publish timeout: The Configure() override with ModuleConfiguration.Create().WithTimeout(...).Build() matches the pattern in the other modules, such as CreateReleaseModule and InstallPlaywrightModule. The using is added. The comment explains why the value is 60 minutes.

Suggestions (non-blocking)

• A 60-minute timeout hides slowdowns. If the publish regresses further, it will take an hour to fail. You could scope the longer timeout to macOS only, or keep a tracking issue for the contention fix. The PR description already calls this a stopgap, so a linked issue would keep it from becoming permanent.
• The two unfixed items (the exit 137 and the GitVersion exception) should have issues filed, per the repo's self-improvement guidance.

LGTM.

[greptile-apps]

Confidence Score: 5/5

[Medium impact] The PR appears safe to merge; no actionable defects were found in either change.

Summary

This PR fixes the stale property reference that broke the engine build and gives the AOT publish module more time.

• Skipped-test tracking checks the current row-object property.
• The AOT test app publish gets a 60-minute timeout.

Acknowledgment: thomhurst explicitly defers reducing macOS publish contention; this PR only extends the timeout.

_{Reviews (1) · Last reviewed commit: "ci: give the macOS AOT test-app publish ..." · Reviewed by Greptile}