* perf(test): build the SQLite migration chain once per test binary (IDEA-1914)
internal/server's -race suite spent ~30 minutes replaying all 69
migrations + 3 backfills per test (~2.7s each, 622 store-backed tests,
BUG-1913). Add internal/store/storetest, which runs the full migration
chain once into a checkpointed, sidecar-free template DB (sync.Once)
and hands every test a plain file copy opened via store.New. Wire it
into internal/server's testServer/testServer_Stop_DrainsRateLimiterCleanup
and internal/store's own testStore (duplicated inline there — an
import cycle rules out sharing storetest with store's white-box
tests). Postgres-mode tests are untouched.
internal/server -race: 1819s -> 183s.
* fix(test): plug template-dir leak and Cleanup race in storetest fixture
Codex round 2 on IDEA-1914: buildTemplate/buildSQLiteTemplate left the
MkdirTemp'd template dir on disk if store.New/checkpoint/journal_mode
failed after mkdir succeeded — now removed via a disarm-on-success
defer in both mirrored copies. Also guard Cleanup()/removeSQLiteTemplate
against racing an in-flight build+copy with a sync.RWMutex (read-locked
across build+copy, write-locked for removal) in both places.
The full `internal/server` test suite under `-race` had grown past the
30m CI timeout, failing every push to main since ~TASK-1354. Diagnosis:
bcrypt at the production cost (12) takes ~3s per call under the race
detector, and dozens of tests now bootstrap a user via the loopback
HTTP path (`bootstrapFirstUser` → `store.CreateUser` →
`bcrypt.GenerateFromPassword`). Cumulative cost dominated the budget.
Two coordinated changes:
1. Lower bcrypt cost in test binaries. `bcryptCost` becomes a package
var (still package-private), and a new `SetBcryptCostForTesting`
helper lets each test binary's `TestMain` drop it to
`bcrypt.MinCost`. Production stays at 12 — only the test process
ever mutates the value.
2. Re-enable `-race` on pull requests. The `if: github.ref ==
'refs/heads/main'` gate was originally a GitHub Actions minutes
cost-control; the repo is public now, so PR minutes are free, and
we'd rather catch race regressions on the contributing branch than
after merge.
Measured impact:
- `go test -race ./internal/server`: 1800s timeout → 830s (13m51s).
- `go test ./internal/store`: 808s → 35s.
- `go test ./internal/server`: 192s → 60s.
The 30m timeout stays — it's headroom for genuine deadlocks, which
would still hit the goroutine-dump panic the way BUG-851 did.
Prior art: BUG-851 (10m → 30m bump, ipRateLimiter goroutine drain).
This is a different cause (bcrypt cumulative time) so the fix is
different.
xarmian