Follow-up from #82 review.
internal/state/db/statepsql DB-backed tests call newTestDB, which t.Skipfs when Postgres is unreachable. There is currently no .github/workflows in the repo, so nothing provisions PG and these tests are green no-ops in automation:
Ask
Add a CI workflow with a Postgres service and point PSQL/the test DSN at it so these guards actually run on PRs.
Note
The pure-in-memory batcher tests (incl. TestShardedBatcherFlushForShardSurfacesAsyncError, added in #82) do not need a DB and run anywhere — this issue is specifically about the DB-backed slice.
🤖 Generated with Claude Code
Follow-up from #82 review.
internal/state/db/statepsqlDB-backed tests callnewTestDB, whicht.Skipfs when Postgres is unreachable. There is currently no.github/workflowsin the repo, so nothing provisions PG and these tests are green no-ops in automation:TestCompleteFlushesQueuedMembers(the flush-before-stamp regression guard from fix(state): stop CompleteGuildBackfill from deleting un-refreshed members #82)TestReconcileGuildMembersnewTestDBcallers inbackfills_test.go/channels_test.goAsk
Add a CI workflow with a Postgres service and point
PSQL/the test DSN at it so these guards actually run on PRs.Note
The pure-in-memory batcher tests (incl.
TestShardedBatcherFlushForShardSurfacesAsyncError, added in #82) do not need a DB and run anywhere — this issue is specifically about the DB-backed slice.🤖 Generated with Claude Code