Skip to content

Release suite: sibling paths from the exported variables, a refusal to install over a linked node_modules, and the lanes a macOS host cannot run named instead of failed - #13

Draft
puskuruk wants to merge 5 commits into
mainfrom
fix/release-suite-sibling-paths
Draft

puskuruk wants to merge 5 commits into
mainfrom
fix/release-suite-sibling-paths

Conversation

@puskuruk

@puskuruk puskuruk commented Oct 19, 2026 •

Copy link
Copy Markdown
Collaborator

Four gaps the release suite showed on a macOS host whose checkouts are not laid out as siblings, found running npm run release:check for #7-#10. None changes what a lane tests; each removes a failure that was about the harness, not the release.

  • test/gea-commonjs.mjs, test/gea-host-namespace-members.mjs and test/native-rtc.mjs resolved core as ../core beside the compiler checkout, although the compiler lane exports GEA_CORE_ROOT. They now read the variable first and fall back to the sibling. Verified from a worktree with no core beside it: all three pass with the variable set.
  • scripts/release-setup.mjs ran npm ci in a worktree whose node_modules was a link to another checkout's install, and npm emptied that checkout. Setup now refuses such a directory, naming the link's target.
  • The perf lane required the benchmarks checkout on every host, then named itself as not run off Linux x86-64. The checkout is required only where the lane runs.
  • The parallel lane shipped the native fixtures to dashwin-geastack from macOS and failed when that machine was unreachable. When neither GEA_PARALLEL_HOST nor the default host answers an ssh -o BatchMode=yes probe, the fixtures build and run on this host, which has the sanitizers they need; the choice is printed. Verified: --only=parallel passes on this Mac with the host unreachable (656 s for the native step alone; beside a gate run and a container build the stress fixture under ThreadSanitizer exceeded the runner's 300 s cap, so the lane wants the machine to itself).
  • The three run-gea-companion-*.sh tests compile the generated C++ of a gea-companion macOS build under apple's targets/macos/generated/gea-companion, which no lane produces. They are named in the summary as not run, as the suite already does for host-bound tests, until a lane produces that build.

Related

@coderabbitai

coderabbitai Bot commented Oct 19, 2026 •

Copy link
Copy Markdown

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

  • Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true
  • 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

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

@puskuruk

Copy link
Copy Markdown
Collaborator Author

@coderabbitai review

@coderabbitai

coderabbitai Bot commented Oct 20, 2026 •

Copy link
Copy Markdown
⚠️ Action not completed

Review rate limited.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

…the three tests that compile core's own sources, with the sibling checkout as the fallback
…the lane is named as not run everywhere else
… machine is unreachable, instead of failing the lane on a host that has the sanitizers
…compile the generated C++ of a gea-companion macOS build that no lane produces
@puskuruk
puskuruk force-pushed the fix/release-suite-sibling-paths branch from d3b6c72 to 9a78921 Compare October 20, 2026 07:45
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant