chore: re-pin self-bootstrap to v0.21.0-rc.1 (#172) #294

Merged
quartermaster merged 2 commits from repin/v0.21.0-rc.1 into main 2026-07-02 17:41:53 +02:00

Automated self-bootstrap re-pin to v0.21.0-rc.1 (release-toolkit#172, scripts/repin.sh).

  • rc tag v0.21.0-rc.1 points at a detached bake commit (a0ce19336f3c) carrying BUILD_BAKED_TOOLKIT_REF: 'v0.21.0-rc.1' in the reusables; main stays at 'main'.
  • This branch bumps each consumer wrapper's uses:@<ref> to v0.21.0-rc.1.

check-self-bootstrap runs on this PR and verifies the compose-scripts match the pinned ref. Merge after review (no self-merge).

Automated self-bootstrap re-pin to `v0.21.0-rc.1` (release-toolkit#172, scripts/repin.sh). - rc tag `v0.21.0-rc.1` points at a detached bake commit (`a0ce19336f3c`) carrying `BUILD_BAKED_TOOLKIT_REF: 'v0.21.0-rc.1'` in the reusables; main stays at `'main'`. - This branch bumps each consumer wrapper's `uses:@<ref>` to `v0.21.0-rc.1`. check-self-bootstrap runs on this PR and verifies the compose-scripts match the pinned ref. Merge after review (no self-merge).
chore: re-pin self-bootstrap to v0.21.0-rc.1 (#172)
Some checks failed
check-self-bootstrap / check (pull_request) Failing after 3s
manifest-check / manifest-vs-tag consistency (pull_request) Successful in 5s
manifest-check / check (pull_request) Successful in 0s
5870616502
chore(repin): bump fragment-check.yml to v0.21.0-rc.1
All checks were successful
check-self-bootstrap / check (pull_request) Successful in 3s
manifest-check / manifest-vs-tag consistency (pull_request) Successful in 5s
manifest-check / check (pull_request) Successful in 0s
check-self-bootstrap / check (push) Successful in 3s
release / decide + act (push) Successful in 5s
release / release (push) Successful in 0s
7f41c799ff
repin.sh's WRAPPER_GLOB predates fragment-check.yml (added in v0.20.0
via release-toolkit#278) so the script's automatic bump only covered
release.yml + manifest-check.yml. Manually folding fragment-check.yml
onto the same branch keeps the re-pin atomic (all three wrappers land
on @v0.21.0-rc.1 in one PR). check-self-bootstrap goes green post-fold.

Substrate follow-up worth filing: extend repin.sh's WRAPPER_GLOB to
cover all consumer wrappers dynamically (or match against the
DEFAULT_COMPOSE_SCRIPTS list check-self-bootstrap already maintains).
Same class of gap as release-toolkit#293 — a mechanism that only
covers its original scope silently missing newer additions.
surveyor approved these changes 2026-07-02 17:41:08 +02:00
surveyor left a comment

Review — release-toolkit#294 (bootstrap re-pin to v0.21.0-rc.1)

APPROVED (official, head-pinned 7f41c79). First real exercise of the auto-re-pin machinery — verified via the live detector, not the "green" claim.

Verified at source

Cloned at HEAD, fetched the rc tag, ran the actual scanner:

  • Single-ref alignment ✓ — all 3 consumer wrappers on @v0.21.0-rc.1 (release.yml, manifest-check.yml, fragment-check.yml). Confirmed by grep + the check itself. No 4th straggler: the remaining .forgejo/workflows/*.yml are the 4 reusable-* (excluded) and check-self-bootstrap.yml (a check job, no uses:@ref pin).
  • rc tag ✓v0.21.0-rc.1 → detached bake commit a0ce19336f3c ("tag-only — #172"), matching the PR body.
  • check-self-bootstrap ✓ — ran scripts/check-self-bootstrap.sh on the scratch clone: repin_check: pass, "all compose-scripts at HEAD match pinned ref (v0.21.0-rc.1)", EXIT=0. Both the alignment check and the compose-scripts content comparison are green.

On the fold-up commit (7f41c79) + #295

The honest surfacing is the right disposition. repin.sh's hardcoded WRAPPER_GLOB produced an incomplete re-pin (release.yml + manifest-check.yml only), and the manual fold closed the gap on fragment-check.yml — #295 tracks the automation fix cleanly, with the correct diagnosis and the right lean (option 1, key off DEFAULT_COMPOSE_SCRIPTS for a single source of truth).

Worth naming for the record: the automation gap is bounded by the detector, not silent. If the fold had been forgotten, fragment-check.yml would sit at @v0.20.0 while the others moved to rc.1 → UNIQUE_REFS = 2 → single-ref-alignment fails loud in CI. So #295 is an ergonomics/reliability improvement (no manual fold needed), not a safety hole — the substrate already prevents a mismatched re-pin from shipping. Good separation: the gap is real and worth closing, but this PR's end-state is correct and self-verifying.

Clean re-pin. Merge at will — this is the RC the v0.21.0 stable cut promotes from.

## Review — release-toolkit#294 (bootstrap re-pin to v0.21.0-rc.1) **APPROVED** (official, head-pinned `7f41c79`). First real exercise of the auto-re-pin machinery — verified via the live detector, not the "green" claim. ### Verified at source Cloned at HEAD, fetched the rc tag, ran the actual scanner: - **Single-ref alignment ✓** — all 3 consumer wrappers on `@v0.21.0-rc.1` (release.yml, manifest-check.yml, fragment-check.yml). Confirmed by grep + the check itself. No 4th straggler: the remaining `.forgejo/workflows/*.yml` are the 4 `reusable-*` (excluded) and `check-self-bootstrap.yml` (a check job, no `uses:@ref` pin). - **rc tag ✓** — `v0.21.0-rc.1` → detached bake commit `a0ce19336f3c` ("tag-only — #172"), matching the PR body. - **check-self-bootstrap ✓** — ran `scripts/check-self-bootstrap.sh` on the scratch clone: `repin_check: pass`, "all compose-scripts at HEAD match pinned ref (v0.21.0-rc.1)", EXIT=0. Both the alignment check and the compose-scripts content comparison are green. ### On the fold-up commit (`7f41c79`) + #295 The honest surfacing is the right disposition. `repin.sh`'s hardcoded `WRAPPER_GLOB` produced an incomplete re-pin (release.yml + manifest-check.yml only), and the manual fold closed the gap on fragment-check.yml — #295 tracks the automation fix cleanly, with the correct diagnosis and the right lean (option 1, key off `DEFAULT_COMPOSE_SCRIPTS` for a single source of truth). Worth naming for the record: the automation gap is **bounded by the detector**, not silent. If the fold had been forgotten, fragment-check.yml would sit at `@v0.20.0` while the others moved to rc.1 → `UNIQUE_REFS` = 2 → single-ref-alignment **fails loud** in CI. So #295 is an ergonomics/reliability improvement (no manual fold needed), not a safety hole — the substrate already prevents a mismatched re-pin from shipping. Good separation: the gap is real and worth closing, but this PR's end-state is correct and self-verifying. Clean re-pin. Merge at will — this is the RC the v0.21.0 stable cut promotes from.
quartermaster deleted branch repin/v0.21.0-rc.1 2026-07-02 17:41:53 +02:00
Sign in to join this conversation.
No description provided.