ci: trim incident history from ci.yml comments

[tylerkron]

Summary

ci.yml's comments had grown into incident history (#444/#445, #635, #689, #641), plus a second copy of the ApiCompat essay that already lives in CONTRIBUTING.md ("The published package is the second opinion"). That buried the few constraints that actually keep the workflow correct. This is a comment-only trim that keeps each constraint and its reason, in one or two lines.

Removed:

• the #444 / #445 narration on the push backstop
• "until now the first machine…" (#635)
• #689 (on the test step and the coverage mv)
• #641 on the coverage threshold
• the duplicated ApiCompat write-up (#636 two-layer guard, narrower-not-stricter baseline). CONTRIBUTING.md on main has all of it.

Each non-obvious setting still says why it exists, so a maintainer won't "fix" it back:

• merge_group is the queue gate. push on main is the backstop for changes that skip the queue (a ruleset bypass, an admin push).
• Only pull_request runs are cancelled. merge_group and push runs are never redundant.
• The baseline-drift check is deliberately a failure, not a warning: an out-of-date baseline silently narrows ApiCompat. It goes red after every release until the bump lands. (Re-added in review: the first cut had dropped this.)
• The pack step is the second half of the public-API guard: it compares the package to the nuget.org release, which no PR can edit. RS0016/RS0017 only check the PR-editable PublicAPI.*.txt. (Re-added in review, with a pointer to CONTRIBUTING.md.)
• The pack step leaves out --no-build on purpose, because that flag skips the ApiCompat targets. release.yml packs with --no-build and so isn't this gate.
• dotnet pack defaults to Release, the configuration the Build step compiled (ci: build and test Release in CI #774), so it still runs the build targets and ApiCompat but skips recompiling Daqifi.Core. I checked this locally: after dotnet build -c Release, the pack log shows Skipping target "CoreCompile" for both TFMs, and ValidatePackageTask still runs.
• Prereleases must not become the baseline. Filtering them out also keeps the baseline a plain three-part version.
• Coverage must never fail the job. The mv is guarded because the step runs under set -e.

No workflow logic, matrix, run: content, action refs, timeouts or pack configuration changed. benchmarks.yml is untouched.

Merge note

#774 (Release build/test) and #768 (SHA pins, build timeout) have both landed on main, and this branch has origin/main merged in. The only conflict was the pack-step comment #774 rewrote. I resolved it by keeping main's -c Release run: lines and SHA pins and trimming the comment to the wording above. Against current main, the diff is still comments-only in ci.yml.

Test plan

• Ruby/Psych script parses ci.yml on main and on this branch and compares the data structures. They are identical, except that one run: string differs only in a shell # comment line (the coverage step's set -e note). The script then diffs the non-blank, non-comment source lines against current main (post-ci: build and test Release in CI #774/ci: SHA-pin actions, drop net9 from benchmarks, build timeout #768): 107 on each side, all identical. I checked the script against a pre-chore: GHA permissions/timeouts, Any-CPU-only solution, TFM comment #741 ref, where it fails as it should on the added permissions/timeout-minutes.
• actionlint .github/workflows/ci.yml: clean
• dotnet build -c Release Daqifi.Core.sln: 0 warnings, 0 errors
• PublicApiTrackingTests + CoverageReportingTests both read ci.yml: 16/16 pass on net9.0 and net10.0 (Release)
• CI green on this PR

🤖 Generated with Claude Code

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0)

Great, no issues found!

Qodo reviewed your code and found no material issues that require review

Tip of the day

💡 Did you know, you can turn on the rule miner and Qodo learns your standards from review history

More tips ↗ | Customize Qodo ↗ | Qodo docs ↗

Previous reviews

Review updated until commit 70f4475 🧠 Deep

Results up to commit 4d9be30 🚀 Fast

No changes from previous review

Results up to commit 920578f ⏭️ Skipped

No changes from previous review

Results up to commit 5b73629 ⏭️ Skipped

No changes from previous review

[qodo-code-review]

PR Summary by Qodo

Trim incident history from CI workflow comments

📝 Documentation 🕐 Less than 5 minutes

AI Description

• Removes historical incident narration from CI workflow comments.
• Retains concise explanations of constraints required for correct workflow behavior.

High-Level Assessment

The current approach is optimal: keep essential operational constraints beside the affected workflow steps while removing historical details already documented elsewhere. Moving all explanations out of the workflow would reduce valuable context for future maintainers.

Files changed (1) +25 / -62

Documentation (1) +25 / -62

ci.ymlCondense CI workflow guidance +25/-62

Condense CI workflow guidance

• Removes incident-specific history and duplicate explanations from workflow comments. Preserves concise rationale for merge-queue backstops, cancellation behavior, API compatibility validation, and non-blocking coverage reporting without changing workflow logic.

.github/workflows/ci.yml

[tylerkron]

/agentic_review

[qodo-code-review]

Code review by qodo was updated up to the latest commit 920578f

[tylerkron]

/agentic_review

[qodo-code-review]

Code review by qodo was updated up to the latest commit 5b73629

[tylerkron]

Qodo-clean, CI green — ready for review

[tylerkron]

/agentic_review

[qodo-code-review]

Code review by qodo was updated up to the latest commit 70f4475

[tylerkron]

Qodo-clean, CI green — ready for review