Repository navigation
ci: run the release-notary guards on feat-cmux-next - #18718
Conversation
The publication moves the release tag to the resolved build commit, so the delta republish stale check must compare against it, and both delta jobs need the resolver. tests/test_nightly_cmux_tui_resolve.py already fails on base for the same jobs. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The delta jobs checked out the tip and compared the release tag with it, but publication moves the tag to the resolved build commit. Whenever the build is behind the tip, every delta result was skipped as stale. The release notes head and the nightly-next recovery manifest and artifact name also used the tip instead of the commit that was built. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ci.yml skips its guards for feat-cmux-next and never runs on push, so the nightly and nightly-next guard tests ran nowhere on the branch nightly-next ships from. cmux-next-guards.yml calls ci-guards.yml with only the release-notary group (plus its always-on submodule forward-only guard) for feat-cmux-next pull requests and pushes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configuration
You can disable this status message by setting the Use the checkbox below for a quick retry:
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. Comment |
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
toolchain_notices.py reads ghostty-next/build.zig.zon and fails, by design, when it is missing; the guard checkout has no submodules. Only feat-cmux-next has that submodule, so main skips the step's work. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
bin/cmux also requires the Rust and Zig standard library notices since the bundle map names them, which a fixture Mach-O cannot carry, so the license compliance test failed on feat-cmux-next. The libexec server helper needs first-party notices only. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
The feat-cmux-next push run https://github.com/manaflow-ai/cmux/actions/runs/37871997540 at 9e3a1a2 failed: cmux-next swift test. |
Summary
ci.yml skips its guards for feat-cmux-next PRs (cmux-next.yml is that branch's CI, #16206) and has no push trigger. So the nightly, notarization and nightly-next guard tests in ci-guards.yml ran nowhere on the branch nightly-next ships from.
tests/test_nightly_cmux_tui_resolve.pystayed red there unseen (#18716).New
.github/workflows/cmux-next-guards.ymlcallsci-guards.ymlfor feat-cmux-next pull requests and pushes:release-notarygroup runs, plus ci-guards' always-on submodule forward-only guard and its Guard status rollup. Main's other groups are not enabled on the diverged branch.The first run of the new workflow (job 113627139893) surfaced two more base reds in the group, both fixed here:
ghostty-next/build.zig.zon(toolchain_notices.py fails on a missing one by design), but the guard checkout has no submodules. A release-notary-only step now shallow-initializes ghostty-next when .gitmodules lists it. It's a no-op on main.bin/cmux, which since the bundle-map update also requires the Rust and Zig std notices that a fixture Mach-O cannot name. It now uses the first-party-onlylibexec/cmux-server-helper.Stacked on #18716 (its two commits are here so this PR's own run is green). Once #18716 merges, the diff is the last two commits.
Testing
tests/test_cmux_next_guards_route.py, which fails without the workflow ("no workflow runs the release-notary guards for feat-cmux-next pull requests and pushes"). It passes with it. The test runs in the release-notary group itself.Changelog
none
🤖 Generated with Claude Code
Need help on this PR? Tag
@codesmith-botwith what you need. Autofix is disabled.Summary by cubic
Runs the release-notary guard tests on
feat-cmux-nextpull requests and pushes, and fixes the nightly delta jobs to compare against the resolved build commit instead of the branch tip.ci.ymlskips its guards forfeat-cmux-nextand never runs on push, so the nightly and nightly-next guard tests ran nowhere on that branch. A newcmux-next-guards.ymlworkflow callsci-guards.ymlwith only therelease-notarygroup forfeat-cmux-nextPRs and pushes, with no path filter so a base red shows on the push that caused it.ghostty-nextsubmodule in therelease-notarygroup, since the license compliance test readsghostty-next/build.zig.zonand fails when it is missing.bin/cmuxalso requires the Rust and Zig standard library notices.release-notarygroup that fails if no workflow routes those guards tofeat-cmux-nextor initializesghostty-nextbefore the compliance check.This stacks on #18716; once that merges, the diff here is the remaining commits.
Written for commit 0115151. Summary will update on new commits.
Note
Medium Risk
Touches nightly publication/delta republish tagging and adds branch-specific CI routing; mistakes could skip deltas or mis-attribute release metadata, but changes are guard-tested and read-only on the new workflow.
Overview
Adds
cmux-next-guards.ymlsofeat-cmux-nextpull requests and pushes run the reusableci-guardsworkflow with only therelease-notaryLinux guard group (nightly/notarization/nightly-next contract tests that main CI skipped on that branch).ci-guards.ymlnow shallow-inits theghostty-nextsubmodule before release-notary license checks when present, wirestests/test_cmux_next_guards_route.pyinto that group, and the license compliance shell test uses alibexec/cmux-server-helperfixture instead ofbin/cmuxso toolchain notices are not required for the positive Mach-O case.nightly.ymldelta generation, republish stale-tag guard, release notes head SHA, and notarization recovery artifact naming now useresolve-nightly-cmux-tui-client.outputs.build_sha(and short SHA) instead ofdecide.outputs.head_sha, with matching updates intest_nightly_universal_build.sh.Reviewed by Cursor Bugbot for commit 0115151. Bugbot is set up for automated code reviews on this repo. Configure here.