You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
docs(readme): #502 dropped the coverage and Go Report Card badges, orphaning the CI badge pipeline #509
Area: docs · infra — regression · found via codebase/status sweep
Expected: the README badge row keeps the signals it had, and CI's coverage-badge pipeline has a consumer. Actual: PR #502 (docs: update README.md, merged 2026-08-20) rewrote the badge block at README.md:21-24 and, alongside adding a new GitHub Release badge, dropped two existing badges:
Go Coverage — img.shields.io/endpoint?url=…/Wave-RF/WaveHouse/badges/coverage-go.json
Go Report Card — goreportcard.com/badge/github.com/Wave-RF/WaveHouse
The Release badge and the Apache-2.0 badge are the only two left, and both were restyled to brand colors (#06B0BF / #086D77) — so the badge row was clearly intentional work; the two removals read as collateral from the reorder, not a decision.
Impact: two things, one public and one internal.
The public front door of a just-published repo lost its test-coverage and Go-report signals in the same week v0.1.0 shipped — exactly the two badges an evaluating adopter scans for.
The whole coverage-badge pipeline is now orphaned but still running.ci.yml's non-gating badge job (.github/workflows/ci.yml:439-463) still runs on every main push, still holds the repo's only contents: write, and still pushes coverage-go.json to the orphan badges branch via scripts/ci/publish-badge.sh — for a badge nothing renders. grep finds no other consumer of coverage-go.json. The job's own comment (ci.yml:430-432) now states something false: "which the README badge reads over raw.githubusercontent.com (tracking: re-evaluate coverage reporting / badge after OSS launch #133)".
Note the surviving Release badge uses ?filter=!client, which correctly dodges the clients/ts/v* tag family — that half is right, and it partly answers #501's "README — no release or npm-version badge" bullet (the npm-version half is still missing).
Scope: decide per badge — restore, or retire the badge and its pipeline (the badge job, publish-badge.sh, cov badge, the badges branch, and the .github/workflows/README.md:64,156-159 invariants that describe it). Don't leave a contents: write job publishing to nothing.
Related: #501 (README badge bullet, item 7), #133 (closed — the coverage-badge pipeline this consumed)
From a pm-triage status sweep of PRs merged since 9d4cef9; validated by code-read against 9ebd10c on 2026-08-21.
On the restore-or-retire question: the only reason that job needs contents: write is that the badge data lives in the repo. I build gocov (https://gocov.dev, source at https://github.com/gocov/gocov), open source, and with it CI just uploads the cover profile under contents: read, you get the badge back plus diff coverage on PRs, and the badges branch can go. Self-host or hosted, free for public repos. If you'd rather just retire the badge, fair enough too.
Thanks for the suggestion – I think we are just retiring the coverage badge entirely for now (closed by #520) – but we'll look at this if we ever reconsider adding it again.
Area: docs · infra — regression · found via codebase/status sweep
Expected: the README badge row keeps the signals it had, and CI's coverage-badge pipeline has a consumer.
Actual: PR #502 (
docs: update README.md, merged 2026-08-20) rewrote the badge block atREADME.md:21-24and, alongside adding a new GitHub Release badge, dropped two existing badges:img.shields.io/endpoint?url=…/Wave-RF/WaveHouse/badges/coverage-go.jsongoreportcard.com/badge/github.com/Wave-RF/WaveHouseThe Release badge and the Apache-2.0 badge are the only two left, and both were restyled to brand colors (
#06B0BF/#086D77) — so the badge row was clearly intentional work; the two removals read as collateral from the reorder, not a decision.Impact: two things, one public and one internal.
v0.1.0shipped — exactly the two badges an evaluating adopter scans for.ci.yml's non-gatingbadgejob (.github/workflows/ci.yml:439-463) still runs on every main push, still holds the repo's onlycontents: write, and still pushescoverage-go.jsonto the orphanbadgesbranch viascripts/ci/publish-badge.sh— for a badge nothing renders.grepfinds no other consumer ofcoverage-go.json. The job's own comment (ci.yml:430-432) now states something false: "which the README badge reads over raw.githubusercontent.com (tracking: re-evaluate coverage reporting / badge after OSS launch #133)".Note the surviving Release badge uses
?filter=!client, which correctly dodges theclients/ts/v*tag family — that half is right, and it partly answers #501's "README — no release or npm-version badge" bullet (the npm-version half is still missing).Scope: decide per badge — restore, or retire the badge and its pipeline (the
badgejob,publish-badge.sh,cov badge, thebadgesbranch, and the.github/workflows/README.md:64,156-159invariants that describe it). Don't leave acontents: writejob publishing to nothing.Related: #501 (README badge bullet, item 7), #133 (closed — the coverage-badge pipeline this consumed)
From a pm-triage status sweep of PRs merged since
9d4cef9; validated by code-read against9ebd10con 2026-08-21.