Repository navigation
docs(contributing): install mage from the CI tools module, not @latest (celeris#838) - #485
Conversation
37b168f to
4b82e7c
Compare
|
Navigate logical layers of code changes, visualize relationships, and explore their blast radius. No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configuration
📒 Files selected for processing (1)
🔗 Linked repositories identifiedCodeRabbit considers these linked repositories for cross-repo context during reviews:
Included review availability: This review used your included allowance. Your plan provides up to 10 included reviews per hour; 6 remain after this review. 📜 Recent review details⏰ Context from checks skipped due to timeout. (25)
🧰 Additional context used🧠 Learnings (1)📓 Common learnings🔀 Multi-repo context goceleris/celeris, goceleris/loadgenLinked repositories findingsgoceleris/celeris
goceleris/loadgen
📝 SummarySummary by CodeRabbit
Walkthrough
ChangesContributor Setup
Priority: ⬇️ Low Estimated code review effort: 1 (Trivial) | ~4 minutes Merge Risk: ⚪ Minimal · up to Contributor setup now follows the repository’s pinned tool versions and matches the lint workflow; no actionable setup regression remains, so the change is mergeable after normal checks. Architecture SummaryArchitecture risk: 🔵 Low · up to The change affects 1 system. Changed systems: Architecture concerns Review detailsSystems and components
Before / after behavior
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
Comment |
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
…(celeris#838) CONTRIBUTING's setup step ran `go install github.com/magefile/mage@latest`, which follows whatever mage releases next, while CI installs the mage that .github/tools/go.mod pins. It now runs the same install as the lint workflow, `go install -modfile=.github/tools/go.mod tool`, which also gives the actionlint the PR template asks for.
…it works next to a local go.work (celeris#838) go refuses -modfile in workspace mode, and .gitignore lists go.work, so the setup hint stopped with '-modfile cannot be used in workspace mode' for a contributor who has one.
4b82e7c to
ae3cf19
Compare
…he version it ended with; pin the mage and benchstat hints (celeris#838) (#910) mage Tools now installs h2spec v2.6.0 on every platform: the go install fallback, which resolved to v2.2.1+incompatible ("Version: 2.0.0"), is gone or pinned, and the target prints the version it ended up with. CONTRIBUTING.md and .github/scripts/bench-ab.sh no longer suggest @latest installs that differ from the versions CI pins. This is the celeris part of #838; the probatorium part is goceleris/probatorium#485.
Part of goceleris/celeris#838: the probatorium part. The celeris part is goceleris/celeris#910. goceleris/celeris#838 needs both PRs: once both have merged, close it by hand.
Defect
CONTRIBUTING.md:7 at a814a5e, this PR's base, tells a contributor to run
go install github.com/magefile/mage@latest. CI installs the mage that.github/tools/go.modpins (probatorium#458), withgo install -modfile="$GITHUB_WORKSPACE/.github/tools/go.mod" tool. The lint workflow and every tier workflow do this. So the hint follows whatever mage releases next, and a localmagecan drift from CI's without any change in this repository.Today they agree:
mage@latestresolves to v1.17.2, the same as the pin (round 0,logs/hint-check.log). The defect is the floating hint, not a wrong version today.Fix
GOWORK=off go install -modfile=.github/tools/go.mod tool. It installs mage v1.17.2 and actionlint v1.7.12. actionlint is the one the PR template asks for..github/toolspins both, and whyGOWORK=offis there.GOWORK=off. go refuses-modfilein workspace mode, and.gitignore:50-51listsgo.work, so a contributor who has one hitgo: -modfile cannot be used in workspace mode. The oldmage@latesthint never had that problem, becausepkg@versionignores workspaces. CI has no go.work, so CI was never affected.Check
fix-r1/scripts/hint-gowork-485.sh(logfix-r1/logs/hint-gowork-485-summary.log, under the lane evidence dirlanes-20261003/CI1/):shblock rather than retyping it.go.workat the clone root (use .) as a contributor would have it, into a private GOBIN.go version -m.mage@latest)-modfile cannot be used in workspace mode, nothing installedcpThe same script also greps for any
@latesttool install. It finds CONTRIBUTING.md:7 at origin/main and nothing on this branch (grep rc 1). This branch changes only CONTRIBUTING.md, so there is no Go package or workflow to test.Cost
None.
Review (round 2)
Two reviews of ae3cf19 found nothing against this PR, so it is unchanged (head still ae3cf19).
GOWORK=off, the same line installed nothing and failed with "-modfile cannot be used in workspace mode". Log:review-correctness-r1/r1x-logs/hint485-summary.log, under the lane evidence dirlanes-20261003/CI1/.git merge-treemerges main 3f3a403 with this head, with fix(stress): the shard job's runner comes from the workflow, not from the pull request's plan (celeris#864) #486's new head 22337a7, and with both, all with no conflict.fix-r3/scripts/merge-check.sh, logfix-r3/logs/merge-check.log.