- beachcomber and lighthouse released their savepoint whenever the inner
function reported no error; a Gate that counts a failed read as off,
or a channel function or delete snapshot that swallows one, left the
caller's Postgres transaction aborted (25P02) and failed the write
- a failed RELEASE now rolls back to the savepoint, as the READMEs promise
- beachcomber gets its testcontainers harness and sync tests
(TestSyncGates, TestSyncAfterCommit, TestSyncDeleteAndSoftDelete,
TestSyncFailuresNonFatal, TestServiceSetup); lighthouse gets
TestBroadcastSwallowedReadFailure
- lighthouse:after_create/update/delete also declare
Before(gorm:commit_or_rollback_transaction); an After-only anchor put
them past GORM's own commit, so a plain gdb.Create enqueued its
broadcast job after the commit on the pool (deferred from 11-05)
- lighthouse gets the testcontainers Postgres harness and TestBroadcastTx
(commit publishes once, rollback nothing, single-statement write
enqueues on its own transaction, failed write enqueues nothing)
- Broadcastable contract and Bind[T] bindings; event {action}.{alias},
default {model, actor, timestamp, ttl} payload, delete snapshot taken
before the row goes
- GORM callbacks installed via lagoon.OnDatabase enqueue a summer.broadcast
job on the write's *sql.Tx inside a savepoint; failures are logged and
never abort the write; zero-key batch writes are skipped
- WithoutBroadcasting[T] (ctx-scoped, per type) and Service.Emit for one
summary event; the one-attempt job namespaces channels and publishes or
broadcasts; the payload travels as a JSON string so JSONB keeps its order
- no jobs for the null driver or Centrifugo without an API key
