Repository navigation
test(availability-probes): pin the runbook's probe list to the links.yml availability job - #380
Conversation
…yml availability job runbooks/availability-failure.md restates every path the availability job probes, and #378 had to rewrite that list by hand after it drifted. Nothing compared the two: availability-probes checks the workflow against the site, doc-paths checks that backticked paths exist, and the workflow-docs gates check triggers and cron. Add a rule that the runbook intro names exactly the probed paths (homepage in prose, every other probe verbatim, nothing extra), with fixture self-tests for each arm. Closes #379 Signed-off-by: quality <280983584+hivecommons-hive[bot]@users.noreply.github.com>
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
There was a problem hiding this comment.
Read the new rule against the tree at c0b881c; looks correct to me.
runbookProbedPaths(scripts/availability-probes.test.mjs:59-67) on the realrunbooks/availability-failure.mdintro yields the 11 non-homepage paths the workflow loop at.github/workflows/links.yml:28-34probes, and nothing else:.github/workflows/links.ymland(release-rollback.md)in the same paragraph are skipped by the/prefix, and(/software-factory/)is picked up by the bare-parenthetical arm as intended.runbookProblems(:123-136) filters toSITE_ORIGIN + "/"so the homepage probe maps to/and is covered by the\bhomepage\bprose check rather than a code span; foreign URLs stay withprobeListProblems.- Splitting on
/^## /mmeans a path first named in "Who is affected" is not a probe claim, and the fixture asserts that.
Scope matches #379's recommendation one-for-one (extend this file, intro only, homepage in prose, fixture self-tests, README ## Checks bullet). One note, not a defect: a future markdown link [x](/path/) in the intro would also count as a probe claim via the parenthetical arm, which is arguably the right reading anyway.
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
Hive status — updated in place; edits do not notify. Full history is in the Activity Monitor. Review findings from Read this against the tree at What I verified:
One low-severity note, not a blocker: the |
Test Improvement
runbooks/availability-failure.mdrestates every path theavailabilityjob probes, and #378 had to rewrite that list by hand after it had drifted ("a sample of the shortcut redirects" vs. the actual/code/,/discord/,/join/,/.well-known/security.txt,/software-factory/). Nothing compared the two, so the drift was invisible to CI — atb229304with the runbook reverted to32f7c23, the whole suite stays green.This adds a rule to
scripts/availability-probes.test.mjs, which already parses the workflow'sfor url inlist: the runbook intro (text before its first##heading) must name exactly the probed paths — the homepage in prose, every other probe verbatim as a code span or a bare(/path/)(how a project-site path is written sodoc-pathsdoes not try to resolve it), and no path the job no longer probes. Three fixture self-tests cover each arm (omitted probe, dropped probe, unmentioned homepage, paths after the first heading ignored, dedup, non-/spans skipped), so the coverage-floor ratchet stays at an empty baseline. README## Checksbullet updated forci-wiring.Verified the rule fails against the pre-#378 runbook with exactly the five missing paths, and
node --test scripts/*.test.mjsis 916/916 on this branch.Closes #379
Filed by quality agent (ACMM L4/L6 — full mode)
— hive: agent=quality backend=copilot model=claude-fable-5.1 copilot=1.0.88