fix(ship): harden Step 12 against whitespace + invalid REPAIR_VERSION

Claude adversarial subagent surfaced three correctness risks in the
Step 12 state machine:

- CURRENT_VERSION and BASE_VERSION were not stripped of CR/whitespace
  on read. A CRLF VERSION file would mismatch the clean package.json
  version, falsely classify as DRIFT_STALE_PKG, then propagate the
  carriage return into package.json via the repair path.

- REPAIR_VERSION was unvalidated. The bump path validates NEW_VERSION
  against the 4-digit semver pattern, but the drift-repair path wrote
  whatever cat VERSION returned directly into package.json. A
  manually-corrupted VERSION file would silently poison the repair.

- Empty-string CURRENT_VERSION (0-byte VERSION, directory-at-VERSION)
  fell through to "not equal to base" and misclassified as
  ALREADY_BUMPED.

Template fix strips \r/newlines/whitespace on every VERSION read,
guards against empty-string results, and applies the same semver
regex gate in the repair path that already protects the bump path.

Adds two regression tests (trailing-CR idempotency + invalid-semver
repair rejection). Total Step 12 coverage: 14 tests, 14/14 pass.

Opens two follow-up TODOs flagged but not fixed in this branch:
test/template drift risk (the tests still reimplement template bash)
and BASE_VERSION silent fallback on git-show failure.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
Garry Tan
2026-04-18 23:20:29 +08:00
parent 07b8983e77
commit c3bee68a60
3 changed files with 61 additions and 5 deletions
+24
View File
@@ -437,6 +437,30 @@ Linux cookie import shipped in v0.11.11.0 (Wave 3). Supports Chrome, Chromium, B
## Ship
### /ship Step 12 test harness should exec the actual template bash, not a reimplementation
**What:** `test/ship-version-sync.test.ts` currently reimplements the bash from `ship/SKILL.md.tmpl` Step 12 inside template literals. When the template changes, both sides must be updated — exactly the drift-risk pattern the Step 12 fix is meant to prevent, applied to our own testing strategy. Replace with a helper that extracts the fenced bash blocks from the template at test time and runs them verbatim (similar to the `skill-parser.ts` pattern).
**Why:** Surfaced by the Claude adversarial subagent during the v1.0.1.0 ship. Today the tests would stay green while the template regresses, because the error-message strings already differ between test and template. It's a silent-drift bug waiting to happen.
**Context:** The fixed test file is at `test/ship-version-sync.test.ts` (branched off garrytan/ship-version-sync). Existing precedent for extracting-from-skill-md is at `test/helpers/skill-parser.ts`. Pattern: read the template, slice from `## Step 12` to the next `---`, grep fenced bash, feed to `/bin/bash` with substituted fixtures.
**Effort:** S (human: ~2h / CC: ~30min)
**Priority:** P2
**Depends on:** None.
### /ship Step 12 BASE_VERSION silent fallback to 0.0.0.0 when git show fails
**What:** `BASE_VERSION=$(git show origin/<base>:VERSION 2>/dev/null || echo "0.0.0.0")` silently defaults to `0.0.0.0` in any failure mode — detached HEAD, no origin, offline, base branch renamed. In such states, a real drift could be misclassified or silently repaired with the wrong value. Distinguish "origin/<base> unreachable" from "origin/<base>:VERSION absent" and fail loudly on the former.
**Why:** Flagged as CRITICAL (confidence 8/10) by the Claude adversarial subagent during the v1.0.1.0 ship. Low practical risk because `/ship` Step 3 already fetches origin before Step 12 runs — any reachability failure would abort Step 3 long before this code runs. Still, defense in depth: if someone invokes Step 12 bash outside the full /ship pipeline (e.g., via a standalone helper), the fallback masks a real problem.
**Context:** Fix: wrap with `git rev-parse --verify origin/<base>` probe; if that fails, error out rather than defaulting. Touches `ship/SKILL.md.tmpl` Step 12 idempotency block (around line 409). Tests need a case where `git show` fails.
**Effort:** S (human: ~1h / CC: ~15min)
**Priority:** P3
**Depends on:** None.
### GitLab support for /land-and-deploy
**What:** Add GitLab MR merge + CI polling support to `/land-and-deploy` skill. Currently uses `gh pr view`, `gh pr checks`, `gh pr merge`, and `gh run list/view` in 15+ places — each needs a GitLab conditional path using `glab ci status`, `glab mr merge`, etc.