mirror of
https://github.com/vercel/workflow.git
synced 2026-09-14 19:59:43 +08:00
codex/retain-workflow-vm
80 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Joey Hotz
|
7d29babaef |
feat(world): add optional getMany() for batch run reads (#2915)
Signed-off-by: Joey Hotz <joeyhotz1@gmail.com> |
||
|
Casey Gowrie
|
7a1ea5a45a |
Fix namespaced active run recovery (#2888)
Signed-off-by: Casey Gowrie <ctgowrie@gmail.com> |
||
|
Nathan Colosimo
|
145835b647 |
Centralize workflow event semantics (#2790)
* Centralize workflow event semantics * Simplify centralized event helper usage * refactor: finish centralizing event semantics * refactor(world): derive Hook from its schema * fix(world): preserve event helper compatibility |
||
|
Nathan Colosimo
|
49a50e83d9 | Document configuration environment variables (v5) (#2468) | ||
|
Nathan Colosimo
|
239031ad9e |
fix(next): respect basePath for workflow routes (#2732)
* fix(next): respect basePath for workflow routes * docs(core): note workflow URL resolution gap * fix(next): expose workflow health route methods * test(utils): remove workflow route helper tests * test(builders): remove route handler string test * fix(next): defer basePath validation to Next.js * refactor(utils): remove workflow url helper wrappers * Test Next basePath builder wiring |
||
|
Nathan Colosimo
|
dd36e26962 |
Fix Postgres step lifecycle event ordering (#2714)
* Fix Postgres step start event ordering * Document Postgres step start transaction * Increase canary HMR e2e timeouts * Address Postgres lifecycle review comments |
||
|
Nathan Colosimo
|
97b8469020 | Fix workflow Postgres enum schemas (#2705) | ||
|
Nathan Colosimo
|
5718df8721 |
fix(world-postgres): defer loopback worker startup (#2657)
* fix(world-postgres): defer loopback worker startup * add changeset |
||
|
Karthik Kalyan
|
25c3df74f8 |
Send occurredAt with workflow events (#2580)
* Send occurredAt with workflow events * Fix occurredAt detail typing |
||
|
Pranay Prakash
|
e7ef9d823b |
perf(core): lazy inline step start (save one world round-trip per step) (#2478)
* perf(core): lazy inline step start to save a world round-trip per step The owned-inline runtime path used to write step_created (suspension handler) and then step_started (executeStep) as two separate world round-trips for a step it already owns and is about to run inline. This defers the step_created write: executeStep sends a single step_started carrying the step input, and the world creates the step on the fly (materializing the step entity plus a synthetic step_created event so replay still observes it). Mirrors the existing resilient run_started -> run_created pattern. Exactly-one ownership is preserved by the world's atomic create-claim: the loser of a concurrent lazy step_started gets EntityConflictError, which executeStep maps to `skipped`, so it never runs the body. A lazy step_started is only ever sent for a brand-new step (the suspension handler defers only steps with no prior step_created), so crash recovery still re-runs a `running` step via the normal non-lazy step_started. Worlds updated: world-local, world-postgres (implicit create + synthetic step_created event), world-vercel (routes the input as the v4 frame payload and threads the server's stepCreated flag). @workflow/world adds optional `input` to step_started and a `stepCreated` EventResult signal. Rollout: server-first. The matching workflow-server change must deploy before this ships; the Vercel world targets a single Vercel-operated backend (server always >= SDK). For local/postgres the world ships in the same package as the runtime, so there is no version skew. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * fix(core): materialize deferred step before failing unregistered step on lazy inline path The lazy inline step-start optimization defers a step's step_created write, expecting executeStep to materialize the step via a lazy step_started carrying its input. For an UNREGISTERED step, executeStep bails out before sending that step_started and writes step_failed directly — but the step entity was never created, so the world's "step must exist" ordering guard rejects the step_failed and the run wedges (times out). This regressed the StepNotRegisteredError e2e tests uniformly across every framework/world (the ghost step never reached `failed`). Fix: on the lazy path, send the lazy step_started first to materialize the step (entity + synthetic step_created, keeping replay correct), then write step_failed. The lazy step_started's atomic create-claim preserves exactly-one-owner: a concurrent winner makes ours reject with EntityConflictError → skipped, so the failure is never written twice. Adds world-level regression tests (world-local, world-postgres) asserting a lazy step_started followed by step_failed marks the step failed. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> |
||
|
Nathan Rajlich
|
f2a7bdeb0a |
fix(world-local,world-postgres): make duplicate hook_created idempotent (#2295)
* fix(world-local): make duplicate hook_created idempotent Duplicate processing of the same hook_created — same runId, hookId, and token, e.g. cross-process replay or queue redelivery — was being recorded as a hook_conflict in the event log, which then replayed as a self- conflict HookConflictError. The fix mirrors the existing step_created duplicate-correlation path: when the exclusive token claim fails and the existing claim has the same (runId, hookId), throw EntityConflictError so the runtime's existing concurrent-replay catch path swallows it. Different runId or hookId reusing the same token still produces a real hook_conflict. The persisted token claim already carried hookId; only the read schema was dropping it. The schema now preserves hookId (marked optional for backward compatibility with older claim files). Fixes #2283 * fix(world-postgres): make duplicate hook_created idempotent world-postgres has the same gap as world-local was just fixed for: the duplicate-token check in events.create unconditionally writes a hook_conflict event when an existing hook with the same token is found, even when the existing hook has the same (runId, hookId) as the incoming event. The unique partial index on workflow_events does not catch this because the duplicate path inserts hook_conflict, not hook_created. Mirror the world-local fix: when the existing hook's (runId, hookId) matches the incoming event, throw EntityConflictError so the runtime's existing concurrent-replay catch path swallows it. Different runId or hookId reusing the same token still produces a real hook_conflict. Refs #2283 * test(e2e): add regression test for hook_conflict from same-tick replay race Regression test for #1665 / #2283. A parent workflow awaits 6 child workflows with Promise.all; each child does a tiny step and creates one webhook. Awaited children flatten into the parent run, so all webhook creations land on the same workflow body. When their step resolutions align in the same tick the workflow body is re-walked and each pass submits hook_created with the same deterministic (correlationId, token). Before the world-side idempotency fix, the world wrote hook_conflict events for the duplicates and the workflow failed with HookConflictError. With the fix, duplicates throw EntityConflictError (swallowed by the suspension handler), no hook_conflict events appear in the log, and the webhooks resolve normally. Verified locally against world-local: the test fails reliably (3/3) on the unfixed code and passes reliably (5/5) on the fixed code. * test(e2e): rewrite parallelStepsThenWebhookWorkflow to match the actual #1665 repro The earlier version invoked another 'use workflow' function directly from inside the parent workflow, which is not a valid child-workflow invocation (child workflows must be spawned via start()) and didn't mirror the bug shape on #1665 anyway. Rewrite the workflow as a single 'use workflow' function that exactly mirrors Paolo's minimal repro: await Promise.all([stepA(), stepB()]); using webhook = createWebhook(); await webhook; The for-loop runs N independent iterations of that sequence in series, each disposing its webhook via 'using' before the next, to give the timing-sensitive race multiple chances to fire. The race is hard to force deterministically on fast local dev — but the same (runId, hookId) idempotency invariant is covered deterministically by the new unit tests in world-local and world-postgres. This e2e test serves as a higher-level regression net: its assertions (no hook_conflict event in the log, no HookConflictError-failed run) are correct whether the race fires or not, and will catch any future regression on a run that does hit it. * fix(world-local,world-postgres): recover crash-orphaned hook claims/rows instead of suppressing the retry Addresses review feedback on PR #2295. The original idempotency fix made duplicate same-(runId, hookId) hook_created submissions throw EntityConflictError so the suspension handler's concurrent-replay catch path swallows them. But the claim file (world-local) and hook row (world-postgres) are written before the durable hook_created event, and the writes are not atomic. A process / DB interruption between the claim/hook write and the event write leaves an orphaned claim/hook row; the retry then matched the same (runId, hookId), threw EntityConflictError, got swallowed, and the run was permanently left with no hook_created event in the log. world-local: - Add a per-(runId, hookId) in-process mutex (withHookLock) mirroring the existing withStepLock, so two same-tick concurrent calls serialize on the entity write and the dedup branch never observes an in-flight winner mid-write. - In the dedup branch, when the existing claim is for the same (runId, hookId) we are trying to create, check whether the durable hook entity actually exists on disk: - exists → real duplicate: throw EntityConflictError as before. - missing → orphaned claim from a prior crash: fall through and complete the partial write (write the hook entity with overwrite, then emit hook_created via the outer code path). world-postgres: - In the dedup branch, when the existing hook row matches the incoming (runId, hookId), check whether a hook_created event for this (runId, correlationId) already exists in the event log: - exists → real duplicate: throw EntityConflictError as before. - missing → orphaned hook row from a prior crash between hook INSERT and events INSERT: skip the hook insert (the row is already there) and let the outer code path emit hook_created, completing the partial write. Tests: - world-local: pre-seed an orphaned token claim with no matching hook entity, retry hook_created, assert hook entity and hook_created event both land (no hook_conflict, no EntityConflictError). - world-postgres: pre-seed an orphaned hook row with no matching hook_created event, retry, assert hook_created event lands (no hook_conflict, no EntityConflictError). Both tests fail on the prior implementation (EntityConflictError thrown on retry, exact symptom from the review). * fix(world-local): probe the event log (not the hook entity) to detect duplicate hook_created Addresses follow-up review on PR #2295. The previous dedup branch checked whether the durable hook entity existed on disk. But the hook entity is written before the `hook_created` event, and the two writes are not atomic, so a crash between them leaves both the claim file and the hook entity on disk with no event in the log. The dedup branch then matched on `(runId, hookId)`, found the hook entity, threw EntityConflictError, and the suspension handler swallowed the retry — permanently losing `hook_created` from the event log. The fix mirrors what the world-postgres branch already does: probe the run's event log for an existing `hook_created` event for the same `(runId, correlationId)`. The event is the durable record of a successful hook creation; the claim file and hook entity are partial- write artifacts that may exist without the event. - exists → real duplicate: throw EntityConflictError so the runtime's concurrent-replay catch path swallows it. - missing → orphaned partial write (crash at any point before the event landed): re-write the hook entity (with overwrite: true, in case a stale partial copy exists) and let the outer code path emit the hook_created event. Added a new helper findHookCreatedEvent that runs a filtered paginatedFileSystemQuery with limit:1 over the run's events. Regression test "should recover an orphaned hook entity with no matching hook_created event" added — pre-creates a hook, deletes just the hook_created event from disk to simulate a crash between the entity write and the event write, asserts the retry emits a fresh hook_created event (no hook_conflict, no swallowed EntityConflictError). I verified this test fails on the prior fix (throws `EntityConflictError: Hook "hook_orphan_entity_1" already created`, exactly as pranaygp reported) and passes on this commit. The previous test ("should recover an orphaned hook token claim with no matching hook entity") continues to pass — the event-log probe is a strict superset of the entity probe, since a missing entity always also implies a missing event. * fix(world-local): converge same-hook creation across workers via canonical eventId Addresses follow-up review on PR #2295. The previous fix made the dedup branch probe the event log to decide real-duplicate vs orphan-recovery, but the probe and the recovery write are not a single atomic operation. Two workers sharing a data directory (or two retries that lose `writeExclusive(constraintPath)` back to back) could both pass the probe (each observing no hook_created event yet), both fall through to the recovery write, and both append a hook_created event with a different eventId — producing two events in the log for the same (runId, hookId). The in-process `withHookLock` mutex does not help here because it is process-local and tag-specific. The fix persists `eventId` in the durable token claim file (written by the original `writeExclusive(constraintPath)`). On a same-(runId, hookId) dedup match, retries adopt that canonical eventId and rebuild the event with a deterministic createdAt derived from the eventId (a ULID). The outer event write switches from `writeJSON` (check-then-write, TOCTOU) to `writeExclusive` (O_CREAT|O_EXCL via temp-file + hard-link, atomic across processes). Either worker may win the publish; the other throws EntityConflictError which the runtime's existing concurrent-replay catch path swallows. Net result: exactly one hook_created event per logical creation. Backward compatibility: a claim file written before this commit lacks `eventId`. Retries that read such a claim fall back to the event-log probe + fresh-eventId recovery — the legacy behavior that does not converge across workers but cannot regress for freshly- written claims after upgrade. world-postgres already converges across workers via the partial unique index on workflow_events_entity_creation_unique (runId+correlationId+eventType for hook/step/wait_created): the loser's INSERT raises 23505 which is already translated to EntityConflictError. Regression tests: - world-local: `converges same-hook creation across workers to one event` uses two tagged storage instances sharing one data directory and fires 25 paired Promise.allSettled hook_created calls. Expected 25 hook_created events total; before this fix yielded 50. - world-postgres: `converges same-hook creation across concurrent calls to one event` exercises the same shape against the real Postgres unique index. Already converges; the test is a guard against future regressions to the catch path. Verified the world-local test fails on c7b23e1b5 with exactly the shape pranaygp reported (50 events for 25 logical creations) and passes on this commit. The earlier orphaned-claim and orphaned- entity recovery tests also continue to pass. * fix(world-local): converge legacy hook claims via recovery-marker sidecar; replace tag-proxy test with real subprocess workers Addresses follow-up review on PR #2295. Two distinct issues, both flagged by pranaygp as P1: 1. The fallback path for token claims written by versions before eventId was persisted inline (legacy claims after upgrade) still permitted the same cross-process corruption the inline fast path was fixed to prevent. Two processes both reading a legacy claim each generated their own eventId, landed their writeExclusive(eventPath) calls at different paths, and appended two hook_created events for the same (runId, hookId). Existing persisted claims after a real upgrade are exactly the state the crash-recovery branch needs to repair, so leaving the legacy path non-convergent is silent corruption, not backward compatibility. 2. The committed cross-worker convergence test used two tagged storage instances sharing one directory as a proxy for separate processes. But tags change the destination filename (events/wrun_X-evnt_Y.worker-a.json vs ...worker-b.json), so two tagged workers can each writeExclusive their own event at different paths and both fulfill. The Map-by-eventId deduplication in the assertion then masked the duplicate publication, so the test passed for the wrong reason. Implementation: - New HookRecoveryMarkerSchema (`{ eventId, hookId, runId }`) and HookRecoveryMarkerPath helper. The marker is a sidecar at hooks/tokens/<hash>.recovery.json, written via writeExclusive so the first cross-process retry pins its candidate eventId as canonical; subsequent retries read the marker and adopt that eventId. Together with the existing writeExclusive(eventPath) in the outer publish, this gives the legacy-fallback path the same single-event convergence guarantee as the inline-eventId fast path. - pinCanonicalEventIdForLegacyClaim() encapsulates the marker write-or-read. A stale marker for a different (runId, hookId) (token-reuse with leaked state) is overwritten best-effort — the common cross-worker race for the same hook still converges; only the narrow stale-token-reuse case loses convergence. - hook_disposed now also deletes the recovery marker when it deletes the token constraint file, preventing a future legacy recovery for a recycled token from latching onto a stale eventId. - The dedup branch unified: existingClaim.eventId for new claims, pinCanonicalEventIdForLegacyClaim() for legacy ones. Removed the now-redundant findHookCreatedEvent helper — the writeExclusive(eventPath) in the outer publish is the authoritative duplicate-vs-orphan detector. Tests: - New test fixture test-fixtures/hook-race-worker.ts (TypeScript, run via child_process.fork with tsx as execPath — tsx is a transitive dev dep via vitest). Each subprocess gets its own createStorage(testDir) so the in-process hookLocks Map cannot serialize across workers. - Replaced the tag-proxy test with "converges same-hook creation across separate OS processes to one event". Spawns workerCount subprocesses, releases them from a barrier into the same hook_created, asserts exactly one fulfilled + (N-1) rejected with EntityConflictError, and asserts directly on the raw events.list() result (no Map dedup) that the number of hook_created entries equals the number of logical creations. - Added "converges same-hook creation across processes when only a legacy token claim exists". Same shape, but pre-seeds the legacy claim format (`{ token, hookId, runId }` with no eventId) before each race. Verified to FAIL on 7ce66551b (both subprocesses fulfill, no convergence) and pass on this commit. - Also verified the new-eventId subprocess test FAILS when the event write is reverted to writeJSON (TOCTOU), confirming it exercises the writeExclusive-based cross-process arbitration. Both prior orphaned-claim / orphaned-entity recovery tests also continue to pass. * fix(world-local): per-lifetime recovery markers, restore event-log probe, fix CI tsx resolution Addresses three P1 review comments on PR #2295. 1. Stale recovery marker leaking across token-reuse lifetimes (pranaygp): The previous marker path used `hashToken(token)` so a stale marker for run A could leak into run B's recovery when the same token was reused after run A terminated through normal lifecycle. `deleteAllHooksForRun()` and tagged `world.clear()` deleted the token constraint and hook entity but NOT the marker sidecar, so the next legacy claim on the same token entered the stale-marker overwrite branch and the workers overwrote it non-atomically, yielding divergent publication. Fix: - Marker path now hashes `(token, runId, hookId)` together (`hookRecoveryMarkerPath` in storage/helpers.ts). Different lifetimes can never share a marker, so the stale-marker overwrite branch is removed entirely. - `hookRecoveryMarkerPath` is moved to helpers.ts and shared across events-storage.ts, hooks-storage.ts, and index.ts. - `deleteAllHooksForRun()` and tagged `world.clear()` now also delete the recovery marker for each hook (disk hygiene; per- lifetime identity makes leaks no longer corrupting). - `hook_disposed` now uses the new per-lifetime marker path too. 2. Duplicate `hook_created` event when a legacy claim's event was already published (VADE bot, also implied by pranaygp's analysis): Removing the event-log probe from the legacy fallback let a post- upgrade retry pin a new canonical eventId via the marker and publish a duplicate event at that path, even when the original pre-upgrade writer had already successfully published the event with its own (different) eventId. Fix: - Restore `findExistingHookCreatedEventId()` (renamed and made to return the eventId for clearer semantics). - Legacy fallback now probes the event log BEFORE pinning the marker; if a matching `hook_created` event already exists, throw `EntityConflictError` so the runtime's concurrent-replay catch path swallows the retry. - Inline-`eventId` fast path does NOT need the probe — the claim itself is the durable convergence key. 3. CI failure: tsx not resolvable under pnpm isolated linking (pranaygp; confirmed by ubuntu/windows unit test 60s timeouts): The previous test hard-coded `node_modules/.bin/tsx` assuming tsx would be hoisted there. But tsx was only a transitive peer dep via vitest, and pnpm's isolated linking does NOT link transitive peer deps into the workspace bin after a fresh install — so neither root nor package-local `.bin/tsx` existed in CI, the subprocess fork never started, and the barrier hung until vitest killed the test. Fix: - Add `tsx` as a direct `devDependency` of `@workflow/world- local` (pinned to 4.20.6 to match the existing transitive resolution). - Resolve via `import.meta.resolve('tsx/package.json')` and read the `bin` field dynamically, so we adapt to wherever pnpm links tsx for this package — not a hard-coded layout. - Lazy-init the resolver (no module-load IIFE) so an absent tsx fails only the convergence tests, not all 376 tests in the file. - Surface a clear error message if resolution fails, calling out the cause (transitive vs direct deps) for future readers. Also: harden the barrier helper so `error` events and pre-ready exits resolve BOTH `readyPromises` and `donePromises`, then `SIGKILL` siblings. Previously a broken child only resolved `donePromises`, leaving `Promise.all(readyPromises)` pending until the per-test timeout (60s in CI). Regression tests added: - `legacy claim whose hook_created event was already published does not append a duplicate event` — pre-seeds a legacy claim AND a pre-existing `hook_created` event with a different eventId, asserts the retry throws EntityConflictError and the log still has exactly the original event. - `converges legacy claim recovery across run lifetimes after token reuse` — runs pranaygp's full lifecycle path: race subprocess workers on run A's legacy claim, terminate run A via `run_completed` (triggers `deleteAllHooksForRun`), reuse the token in a legacy claim for run B, race subprocess workers again, asserts exactly one fulfillment + one `EntityConflictError` per race and exactly one `hook_created` event per run. Both new tests verified to fail on 2c673e436 (after rebuilding): the published-event test throws via duplicate publish instead of EntityConflictError, the token-reuse test sees both run B workers fulfill (2 events instead of 1). The existing orphaned-claim and orphaned-entity recovery tests also continue to pass. CI loop confirmed to be repaired locally by spawning subprocesses via the new resolver and intentionally breaking the worker fixture to verify the helper fails fast (~500ms) instead of hanging at the barrier. * fix(world-local): defer hook entity write until event publish commits Addresses karthikscale3's P1 review comment on PR #2295. The dedup-recovery path used to write the hook entity BEFORE the outer event publish proved whether the attempt was repairing a missing event or just colliding with an already-published `hook_created`. For already-committed duplicates, the event write then throws `EntityConflictError`, but the hook entity had already been overwritten with the retry's payload — leaving the durable hook entity and the event log inconsistent (e.g. the entity reflects the retry's metadata while the event still carries the original). karthikscale3 reproduced this on the prior head by creating `hook_created` with metadata `{ v: "a" }`, then retrying the same `(runId, hookId, token)` with metadata `{ v: "b" }` and `isWebhook: false`: the retry threw `EntityConflictError` but `hooks.get()` returned the retry's payload. Fix: defer the hook entity write until AFTER the outer `writeExclusive(eventPath)` commits. The branch now only captures the entity-to-write and its overwrite options; the actual write happens immediately after the event publish in the shared trailing block. A retry that ends in `EntityConflictError` (the event was already published) now leaves the entity untouched. The first-writer happy path and all recovery paths (orphaned- claim, orphaned-entity, cross-worker convergence, legacy claim, token-reuse across lifetimes) are unaffected — they all reach the event publish successfully, then the entity write runs as before. Regression test `does not mutate an already-committed hook entity when a duplicate hook_created retry collides` added to world-local: runs karthikscale3's exact scenario and asserts the persisted entity still carries the original metadata and isWebhook. Verified to fail on the prior commit (persisted metadata = 0xbb instead of 0xaa) and pass on this commit after rebuilding. Parallel guard test `does not mutate an already-committed hook entity when a duplicate hook_created retry collides` added to world-postgres. Postgres already protected this via `onConflictDoNothing()` on the hook INSERT, but the test guards against a future regression that adds an UPDATE/UPSERT to the dedup path. * refactor(world-local): per-instance in-process locks; drop tsx subprocess test plumbing You were right that the tsx subprocess machinery was overkill for a storage-level convergence test. Replaced with a simple two-instance in-process test that exercises the same cross-process semantics without spawning anything. The trick: `stepLocks` and `hookLocks` were module-level Maps shared by all `createEventsStorage` calls in the same process. Move them inside the function so each `createStorage(dir)` call gets its own lock map. Two storage instances sharing one data directory then behave exactly like two separate OS processes: - independent in-process `hookLocks` Maps (no in-process serialization between them), and - a shared filesystem (so the on-disk `writeExclusive` claim / marker / event publish primitives are the only thing arbitrating convergence). This is also a real architectural improvement — the global lock map was always a leaky abstraction that made unit-test simulation of the cross-process path awkward. Changes: - `stepLocks` and `hookLocks` moved from module scope into `createEventsStorage`. `withStepLock` and `withHookLock` wrappers collapsed into direct `withInProcessLock(map, key, fn)` calls at the two call sites that need them. - The three convergence regression tests in `storage.test.ts` now use `const workerA = createStorage(testDir); const workerB = createStorage(testDir);` and race `Promise.allSettled` of `events.create` from both — no subprocess, no IPC, no barrier helper, no `raceHookCreatedAcrossProcesses`. Same assertions (exactly one fulfillment + N-1 `EntityConflictError` per race, raw `events.list()` shows exactly one `hook_created` per logical creation — no Map dedup) so the regression catches are identical. - Removed: `tsx` devDep, `test-fixtures/hook-race-worker.ts`, `HOOK_RACE_WORKER` / `resolveTsxLoaderUrl` / `TSX_BIN` / `raceHookCreatedAcrossProcesses` and the `fork`/`fileURLToPath` imports they pulled in. Verified (after rebuilding world-local): - All 379 tests pass on macOS in ~1s (was ~6.7s with subprocesses). - Convergence tests confirmed to still catch the bugs: temporarily reverted the `eventId = canonicalEventId` adoption → both workers fulfilled (2 events instead of 1). Temporarily reverted the legacy-claim marker pin → same: both workers fulfilled. - No subprocess machinery means no Windows-specific quirks (cli.mjs shebang, .cmd wrappers, .bin hoisting under pnpm isolated linking, etc.) that produced the Windows CI 60s timeouts. - World-postgres still has its own parallel guard test for the karthikscale3 "no-mutate-on-duplicate" regression; that one exercises real DB concurrency and is unaffected by this change. Full repo `pnpm test` (43 packages) and the `parallelStepsThenWebhookWorkflow` e2e test against world-local both green. * fix(world-local): repair event-first hook orphans from the persisted event; skip #1665 e2e on world-postgres - A crash between the hook_created event publish and the deferred hook entity write left the event committed with the entity missing and unrepairable (retries threw EntityConflictError without materializing the entity). Retries now rebuild the entity from the PERSISTED event's payload — never the retry's eventData — via a race-safe writeExclusive, on both the canonical-eventId collision path and the legacy-claim probe path. - Skip parallelStepsThenWebhookWorkflow e2e on world-postgres: the same-tick replay pattern surfaces a separate pre-existing step_started ordering bug there (#2331). --------- Co-authored-by: Peter Wielander <peter.wielander@vercel.com> |
||
|
Pranay Prakash
|
ae8d6feeda |
Add native v4 workflow attribute events (#2226)
* Add native workflow attribute events * Fix abbreviated attributes docs sample * Document attribute replay ordering for step races * Address native attribute review feedback * Validate before claiming attr_set dedup lock; clearer start() attribute errors - world-local: claim the attr_set correlation lock only after validation, so a validation failure does not permanently mark the correlationId as written and wedge the run in a re-invoke loop on retry - world-postgres: distinguish a concurrently-deleted run from a cap violation when the guarded attributes update matches no rows - core: reject non-string initial attribute values in start() with a clear error instead of a downstream schema failure Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Add attribute edge-case tests across all layers - core: normalizeAttributeChanges unit tests (non-object inputs, FatalError wrapping, key/value/batch limits, boundary lengths, UTF-8 byte counting) - core: start() rejects reserved keys, oversized keys/values, and over-cap initial attribute batches before any write - world-local + world-postgres: per-run cap enforced against existing attributes (upsert-at-cap allowed, removal frees room), oversized values rejected on attr_set, invalid initial attributes rejected on run_created - e2e: validation DX workflow asserting every invalid write throws a catchable FatalError naming the violated rule and limit, with the run staying healthy; start() rejects invalid initial attributes client-side Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Remove accidentally committed local e2e diagnostics artifact Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Bump world-vercel to spec version 4 for native attributes The deployed workflow-server (vercel/workflow-server#469) materializes native attr_set events and accepts initial run attributes, but world-vercel still advertised spec v3 — so start(..., { attributes }) rejected itself client-side ('requires spec version 4') on every Vercel deployment, failing the new e2e seeding test across the prod matrix. New runs are now stamped v4. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Reject duplicate correlated attr_set before materializing in Postgres A redelivered duplicate — including one carrying different changes for the same correlationId — previously re-applied the run attributes update and only then failed the event insert, leaving the snapshot out of sync with the event log. Pre-check the event log for the correlationId before mutating; the unique index still guards the truly-concurrent race, which is idempotent (deterministic replay carries identical changes). Also apply the suggested docs wording for initial attributes. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Apply suggestion from @VaguelySerious Signed-off-by: Peter Wielander <mittgfu@gmail.com> * Fail the run on World-rejected attribute writes; un-nest runtime test Two fixes from review: - runtime.test.ts: the pre-existing test "propagates transient step_created failures..." was accidentally nested inside the new attribute-race test, failing the new test ("Calling the test function inside another test function is not allowed") and preventing the old test from running. Restored it verbatim at describe level. - A workflow-body attr_set the World rejects as invalid (e.g. the cumulative per-run attribute cap, which only the World can check) is deterministic: redelivering the orchestrator message replays the same write into the same rejection, wedging the run in redelivery with no terminal event. handleSuspension now wraps such rejections in FatalError, and workflowEntrypoint fails the run with the validation error instead of rejecting the delivery. Transient storage errors still propagate and retry via redelivery. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Signed-off-by: Peter Wielander <mittgfu@gmail.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com> Co-authored-by: Peter Wielander <mittgfu@gmail.com> Co-authored-by: Peter Wielander <peter.wielander@vercel.com> |
||
|
Will Sather
|
4670c4b92d |
feat(core): add optional namespace for queue topic prefix (#2305)
* feat(core): add optional namespace for queue prefix * fix(world-postgres): job queue prefix validation * fix: changeset description Co-authored-by: Peter Wielander <mittgfu@gmail.com> Signed-off-by: Will Sather <56037657+willsather@users.noreply.github.com> * fix: add world-postgres to changeset * fix: world-postgres handle namespaced job queue names * fix: resolve namespace via env var in core runtime * fix: world-postgres job queue name task handler * fix(world-postgres): honor namespace on consumer side * Fix namespaced queue routing reliability (#2340) * Fix namespaced queue routing reliability * Inline queue namespace in generated routes * Avoid loading Vercel functions during runtime import --------- Signed-off-by: Will Sather <56037657+willsather@users.noreply.github.com> Co-authored-by: Peter Wielander <mittgfu@gmail.com> Co-authored-by: JJ Kasper <jj@jjsweb.site> |
||
|
Peter Wielander
|
1e6b1fdea2 |
Attributes MVP (experimental and write-only) and CI hardening (#2134)
* fix(core): scan inline sourcemaps during error remapping * Attributes MVP (experimental and write-only) (#2088) |
||
|
Pranay Prakash
|
dc0be50618 |
[codex] Forward port stale wait replay fix (#2038)
* Forward port stale wait replay fix
* Guard V5 replay writes against stale events
* Revert "Guard V5 replay writes against stale events"
This reverts commit
|
||
|
Peter Wielander
|
738ec5e81d |
[world-postgres] Bootstrap graphile-worker schema in setup CLI (#2019)
* fix(world-postgres): bootstrap graphile-worker schema in setup CLI `workflow-postgres-setup` now installs the `graphile_worker` schema in addition to the drizzle migrations so that by the time any consumer calls `world.start()`, both schemas already exist. This eliminates the inter-process race on graphile-worker's `installSchema` where concurrent `CREATE SCHEMA IF NOT EXISTS` calls could both pass the MVCC-snapshotted existence check and one would fail with `duplicate key value violates unique constraint "pg_namespace_nspname_index"`. Reproduced locally against a fresh postgres:18-alpine with 8 parallel `makeWorkerUtils().migrate()` calls — 7/8 fail without the pre-bootstrap, 0/8 fail after running `workflow-postgres-setup` first. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * Apply suggestions from code review Co-authored-by: Peter Wielander <mittgfu@gmail.com> Signed-off-by: Peter Wielander <mittgfu@gmail.com> --------- Signed-off-by: Peter Wielander <mittgfu@gmail.com> Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> Co-authored-by: Nathan Rajlich <n@n8.io> |
||
|
Pranay Prakash
|
9d2a9261fd |
Expose conflicting run id on hook conflicts (#2012)
* Expose conflicting run id on hook conflicts * Mark hook conflict run id as future required * Address hook conflict docs review * Address hook conflict review comments * Fix hook conflict docs typecheck |
||
|
Pranay Prakash
|
aee56993c7 |
feat: serializable AbortController/AbortSignal (#1301)
* feat: add docs and test stubs for serializable AbortController/AbortSignal Adds documentation and test infrastructure for making AbortController and AbortSignal serializable across workflow and step boundaries. The feature uses a dual hook+stream backing: hooks for deterministic replay in the workflow context, streams for real-time propagation to running steps. Docs: - Cancellation guide (foundations) covering AbortSignal and run cancellation - How Cancellation Works (how-it-works) explaining hook+stream internals - AbortSignal.timeout() error page for the workflow VM restriction - Updated serialization docs with AbortController/AbortSignal section Tests (all .todo stubs for TDD): - VM behavior: AbortController API, static methods, hook integration - Step-side: stream reader setup, abort propagation, ops queue - Serialization round-trips: all boundaries, encryption, nested structures - Consistency: race conditions, partial failure, eventual convergence - E2E workflows: timeout, parallel, step-initiated, hook-triggered, replay Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: use correct frontmatter type for error page Change type from "error" to "troubleshooting" to match the valid frontmatter schema used by all other error pages. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * docs: address review feedback on cancellation docs - abort() in workflow does not synchronously update signal.aborted; instead it queues hook resumption and the replay handles state update - stream name and hook token are generated at serialization time (not deterministically in the workflow) and stored in the event log - use throwIfAborted() instead of manual signal.aborted checks Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * docs: document runtime change for processing abort queue items on completion The current runtime only processes invocation queue items on suspension. When abort() is called after the last suspension point and the workflow completes, the queue items are dropped with a warning. Document that the runtime needs to flush abort-related items on completion/failure too. Add test stubs for this behavior. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * docs: generalize queue processing on completion to all item types Processing pending invocations queue items on workflow completion/failure should apply to all queue item types (steps, hooks, waits, abort signals), not just abort-related ones. Update docs and tests accordingly. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * docs: abort errors in steps are automatically wrapped in FatalError When a step throws due to an abort (AbortError from fetch, throwIfAborted, etc.), the error is wrapped in FatalError so the step skips retries. An abort is intentional cancellation, not a transient failure. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * docs: remove contrived "aborting from within a step" example Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * docs: add meaningful step-initiated abort example (quota monitor) Replace the contrived example with a watchdog pattern where a monitoring step polls an external condition and aborts parallel work when triggered. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * docs: remove unnecessary "as const" from hook cancellation example Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat: implement serializable AbortController/AbortSignal Core serialization layer: - Add AbortController/AbortSignal to SerializableSpecial interface - Add reducers for all 4 contexts (external, workflow, step, common) - Add revivers for all 4 contexts with stream-backed propagation - Add reviveAbortController helper for step/external contexts - Guard instanceof checks for VMs without AbortController global Workflow VM: - New workflow/abort-controller.ts with createCreateAbortController factory - WorkflowAbortSignal class with hook-backed state - AbortSignal static methods (abort, any, timeout blocked) - Hook integration via invocations queue and events consumer Supporting changes: - Add ABORT_STREAM_NAME, ABORT_HOOK_TOKEN symbols - Add getAbortStreamId() for system stream namespace - Add isSystem, abortRequested, abortReason to HookInvocationQueueItem - Add isSystem to world Hook entity and events - Wrap AbortError in FatalError in step handler (skip retries) - Add AbortController/AbortSignal to Serializable type - Add observability revivers for abort types - Add isSystem to postgres schema and web-shared attribute panel All 454 existing tests pass with no regressions. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat: wire up AbortController in workflow VM and process queue on completion - Wire up AbortController/AbortSignal in workflow VM (workflow.ts) - Add abort processing to suspension handler (hook resume + stream write) - Process pending queue items on workflow completion (throw WorkflowSuspension instead of warning for actionable items) - Fix instanceof guards for non-function AbortSignal in VM - Update test to expect WorkflowSuspension for unawaited steps All 454 existing tests pass. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat: implement tests and Request.signal serialization Tests (516 passing, 18 todo for integration tests): - 18 VM behavior tests (abort-controller.test.ts) - 18 step-side behavior tests (abort-controller-step.test.ts) - 4 consistency tests + 14 integration todos (abort-consistency.test.ts) - 14 serialization round-trip tests (serialization.test.ts) - 7 hook integration + 4 integration todos (step.test.ts) Request.signal serialization: - Add signal field to SerializableSpecial Request type - Include signal in Request reducer when present - Pass signal through in external and step Request revivers Fix workflow reviver for AbortController/AbortSignal: - Use plain objects instead of prototype-based stubs Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * test: implement all remaining .todo test stubs Convert all 27 remaining .todo stubs to real implementations: - 14 consistency tests (race conditions, partial failures, queue processing) - 4 hook integration tests (suspension handler, hydration, eventual consistency) - 9 e2e tests (timeout, parallel, step-abort, hook-cancel, replay, external signal) All 558 tests pass, 0 todos remaining. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: address PR review comments + add changelog PR review fixes: - Move cancellation after streaming in foundations nav - Fix AbortSignal reducer to detect WorkflowAbortSignal via symbol - Guard AbortController reducer from matching AbortSignal objects - Add e2e tests: throwIfAborted, reason types, uncaught fetch AbortError Changelog: - Add hidden changelog section (not in sidebar, accessible via URL) - Add draft changelog entry for serializable AbortController/AbortSignal Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat: show changelog in nav for preview deployments only - Add `preview` flag to nav items in geistdocs.tsx - Filter preview items in Navbar (server component) based on VERCEL_ENV - Show "Preview" badge on preview nav items in DesktopMenu - Changelog link visible in preview deployments and local dev only Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat: move preview badge from home page to navbar Move the PreviewBadge (with package tarball install modal) from the fixed bottom-right position on the home page to the navbar, so it appears on every page during preview deployments. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat: consolidate preview tools into single Internal page Replace separate Changelog nav item and PreviewBadge with a single "Internal" page that only appears in preview deployments: - Rename docs/changelog/ to docs/internal/ - Internal page includes preview package install commands and draft changelogs in one place - Nav shows "Internal" with Preview badge in preview/dev only - Remove PreviewBadge from navbar (now on the Internal page) - Add callout that page is preview-only Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat: use real deployment URLs on internal page + exclude from indexing - Add PreviewInstall component with copy-to-clipboard buttons using the actual VERCEL_URL (not placeholders) - Register PreviewInstallServer as MDX component for docs pages - Exclude /internal/ pages from sitemap.xml, sitemap.md, and llms.mdx - Add robots.txt Disallow for /internal/ paths Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: add missing type declarations for docs code sample typechecking Add declare statements and @setup/@skip-typecheck annotations for undeclared functions in code samples (stepA, stepB, fetchData, cancellableStep, splitIntoChunks, processChunk). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: add missing type declarations for all docs code samples Fix docs typecheck CI by adding declare statements and @skip-typecheck annotations for all undeclared function references across cancellation docs, error page, how-it-works page, and internal changelog. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: only suspend on completion for abort items, not all pending items The previous logic threw WorkflowSuspension for any pending queue item on completion (steps, waits, hooks). This broke fire-and-forget patterns like `void sleep('1d').then(...)` which intentionally leave a wait in the queue without awaiting it. Now only abort-related items (hooks with abortRequested) trigger suspension on completion. Other pending items get the original warning behavior — they may be intentional fire-and-forget operations. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: all pending queue items are fire-and-forget on completion Remove special-case suspension for abort items on workflow completion. ALL pending queue items (steps, hooks, waits, abort signals) are now fire-and-forget when the workflow completes — they get warned about but don't block completion. This matches the existing behavior for fire-and-forget patterns like `void sleep('1d').then(...)`. Abort signals propagate through the normal suspension flow during the workflow (not at completion time). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: resolve docs typecheck errors in code samples Move declare statements before imports to avoid TypeScript overload signature conflicts with auto-inferred imports. Add @skip-typecheck for conceptual snippets. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: abort() in workflow updates signal.aborted synchronously abort() must update signal.aborted immediately so that: 1. Subsequent reads in the workflow see the correct state 2. Serialization captures aborted=true when passing signal to steps 3. Event listeners fire synchronously The hook resumption still happens via the suspension handler for durable event log recording. Both local state and durable state are now updated. Fixes e2e failures where steps received aborted=false for signals that were aborted before being passed to the step. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * docs: update how-it-works to reflect synchronous signal.aborted update abort() now updates signal.aborted synchronously in the workflow. Update lifecycle diagram and remove outdated paragraph about signal not being updated synchronously. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: ensure abort listeners fire at deterministic point across replays On replay, hook_received is processed during event consumer subscription (at AbortController construction time), which is BEFORE the abort() call in the workflow code. If listeners fired during event processing, they'd fire at a different point than on first-run — breaking determinism. Solution: split abort into two phases: 1. _markAbortedFromReplay(): Sets signal.aborted=true (for reads/serialization) but does NOT fire listeners. Called by event consumer during replay. 2. abort(): Detects the replay flag and fires listeners at the call site. On first-run, fires listeners immediately as before. This ensures listeners fire at the abort() call site on BOTH first-run and replay, maintaining consistent ordering of side effects. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * test: add replay ordering tests for interleaved hook scenarios Add 3 tests validating that abort listeners fire at the abort() call site on both first-run and replay, even when other hook events are interleaved in the event log. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: signal.aborted stays false until abort() is called for deterministic replay _markAbortedFromReplay no longer sets signal.aborted = true. Both aborted state and listener firing are fully deferred to abort(). This prevents if-checks on signal.aborted from taking different branches on first-run vs replay. Add deterministic branching test (unit + e2e): const controller = new AbortController(); if (controller.signal.aborted) { return 'was aborted'; // never taken } else { controller.abort(); return 'just aborted'; // always taken, both runs } Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * test: add abort+hook ordering matrix e2e tests (4 combinations) Test all combinations of listener registration order and event trigger order to validate deterministic ordering across first-run and replay: 1. addEventListener first, abort() first 2. addEventListener first, resumeHook first 3. hook.then first, abort() first 4. hook.then first, resumeHook first Each test verifies that abort-listener fires synchronously at the abort() call site (immediately before 'after-abort' in the log), regardless of when the hook is resumed or when listeners are registered. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: simplify abort — event consumer calls _setAborted directly Remove the deferred _markAbortedFromReplay approach. The event consumer now calls _setAborted directly when hook_received is processed, which sets signal.aborted = true AND fires listeners at that point. This is correct because: - Cross-execution aborts (step/external): signal.aborted SHOULD be true on replay since the abort is a fact from a previous run. Listeners must fire so the workflow can react to the abort. - Same-execution aborts: abort() fires _setAborted synchronously. On replay, the event consumer fires it first, and abort() is a no-op. - The promiseQueue ensures listeners fire at the deterministic point matching the hook_received event's position in the event log. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * test: skip abort+hook ordering e2e tests pending full integration The 4 ordering matrix tests require the abort controller's internal system hook to be fully wired through the suspension handler. The hook creation timing interacts with the user hook lookup in getHookByToken. Skip until the full integration is complete. All 13 other abort e2e tests pass on CI. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * handle dangling streams * fix postgres world * fix abort serialization bug * refactors * add drizzle migration file * fix tests * fix tests * replace setTimeout probe and any casts with typed abort internals Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * cover post-serialization abort and nested-in-Request reader cleanup Two leak paths the prior fix left uncovered: - External signal aborted after serialization: verifies the listener attached by reduceAbortWithListener actually fires and writes the abort packet once the caller aborts later. - Signal nested inside a Request: exposed a real leak. The Request constructor copies the signal to an internal AbortSignal, so the ABORT_READER_CANCEL symbol set by reviveAbortSignal never reached request.signal, and cancelAbortReaders' walker had no Request case so Object.values(request) returned []. Fixed both sides: - Request reviver copies abort-internal symbols via copyAbortInternals - Walker descends into Request.signal explicitly Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * add v4/v5 docs switcher and pre-release gating - Mark new abort-controller/cancellation pages with preRelease: true (cancellation, how-it-works/cancellation, abort-signal-timeout-in-workflow, serializable-abort-controller). preRelease is a new optional frontmatter field declared in source.config.ts. - lib/geistdocs/versions.ts: declarative version list (v4 Latest, v5 Pre-release) plus getVersionFromPathname and buildVersionUrl helpers used by the switcher. - lib/geistdocs/version-source.ts: filter preRelease pages out of the v4 sidebar tree; rewrite sidebar URLs to /v5/docs/* on v5 so links stay in the pre-release view. - components/geistdocs/version-switcher.tsx: dropdown at the top of the sidebar, styled after the ai-sdk.dev pattern (label + subtitle). - components/geistdocs/pre-release-banner.tsx: banner rendered above the docs layout on all /v5/docs/* routes, linking back to /docs/* (Latest). - app/[lang]/v5/docs: parallel route (layout + page) that reuses the existing docs rendering but keeps preRelease pages visible. - app/[lang]/docs/[[...slug]]: 404 direct access to preRelease pages on v4 so unreleased content is never reachable without the /v5 prefix. - next.config.ts: /v5/docs -> /v5/docs/getting-started mirror of the existing /docs -> /docs/getting-started redirect. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix version switcher URL when default locale is hidden buildVersionUrl assumed segment 0 was the locale, but next.js i18n middleware hides the default locale from the URL so usePathname() returns '/docs/...' rather than '/en/docs/...'. The old logic treated 'docs' as the locale and produced '/docs/v5/getting-started' (404) instead of '/v5/docs/getting-started'. Detect the locale by checking whether segment 0 is a known structural token ('docs' or 'v5') rather than by position, so the function works for both '/docs/...' and '/<locale>/docs/...' inputs. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * match ai-sdk pre-release banner styling Filled sparkles glyph, blue tint on the message text, and a plain underlined "Go to ..." link in the foreground color instead of a bordered pill. Matches the ai-sdk.dev v7 banner reference. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * match ai-sdk switcher icons and banner link color - Switcher: colored rounded icon tile next to each version (orange tint for pre-release, blue for latest), matching the ai-sdk.dev dropdown. Uses a workflow glyph inside a tinted ring. - Banner link: blue text with a softer underline by default, deeper blue on hover. Replaces the foreground-colored link that didn't match ai-sdk's styling. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * use exact ai-sdk icons and darker banner link - Switcher tile: use the T-mark SVG and the bg-orange-100/border-orange-300 (pre-release) / bg-blue-100/border-blue-300 (latest) palette extracted from the ai-sdk.dev live markup, with matching dark-mode variants. - Pre-release banner sparkle: replaced the placeholder with the exact three-path geist sparkle used by ai-sdk. - Banner "Go to Latest" link: foreground color with a muted underline by default (same weight as ai-sdk's near-black link), underline intensifies on hover. The previous blue-600 was too light. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(docs): correct dark-mode colors for pre-release banner and version switcher The geistcn design-system palette inverts brightness semantics in dark mode (low indices = dim, high indices = bright) and remaps `blue-*` but not `orange-*`, so the previous token choices rendered as dim gray-blue text and a mid-bright blue icon inconsistent with the dropdown list. - Banner: use `dark:text-blue-900` for icon + label and switch the "Go to" link from `text-foreground` to the same blue (with a blue underline) so it reads as a single colored banner. - VersionSwitcher: move the text color onto the SVG itself so the `DropdownMenuItem` SVG-color override no longer hijacks the T color, and invert the dark blue palette (dark bg, light border, bright T) so the selected/trigger icon matches the list icon. - Active-row check icon: use green instead of `fd-primary` (which resolves to near-white in dark mode). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix: add signal field to Request serializable type The merge from main moved the Request type into serialization/types.ts without carrying over the signal?: AbortSignal field, causing the abort-related reducers/revivers in serialization.ts to fail typecheck. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * refactor: address review feedback on abort serialization - Dedupe abort listener attach in serialization reducers via marker symbol (prevents N-listener leak when one controller is serialized to N steps, which would double-close the backing stream on abort). - Replace token.replace('abrt_', '') string-surgery in suspension-handler by storing streamName directly on HookInvocationQueueItem at the point where it's already known (workflow/abort-controller.ts construction). - Document the deliberate sync-vs-microtask listener divergence in the workflow VM (replay determinism > spec parity inside the VM). - Add changeset noting the AbortError -> FatalError behavior change. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs: correct cancellation docs against implementation - Remove the contradictory paragraph claiming signal.aborted is not set synchronously when abort() is called in the workflow. The implementation sets it sync via _setAborted; replay re-applies via the events consumer. - Reword the "Stream Succeeds, Hook Fails" recovery — there's no in-process retry loop on the step-side resumeHook call; convergence comes from the next replay re-reading the stream. - Tighten Request.signal handling: plain non-aborted native signals are intentionally dropped to avoid minting stream infra for auto-generated Request signals; only already-aborted or workflow-tagged signals are forwarded. - Replace the wrong "Pending queue items processed on completion" bullet with an accurate fire-and-forget note matching the warn-only behavior. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix: DOMException serialization (replace broken isNativeError guard) DOMException is `instanceof Error` in Node but does NOT pass `types.isNativeError()` — the existing reducer's first guard was `isNativeError(value)`, so DOMException never matched. Devalue then fell through to its arbitrary-POJO failure path. This surfaced as a real bug for AbortController/AbortSignal: when abort() is called with no argument, native AbortController synthesizes a default DOMException as signal.reason. Returning that signal's reason from a step (e.g. `{aborted, reason: signal.reason}`) crashed step return-value serialization. Replace the guard with a constructor-name check (cross-VM safe; same pattern used elsewhere for matching Error subclasses across realms). Also fixes 7 pre-existing DOMException tests in serialization.test.ts that were previously failing on main. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix: drain pending queue items on workflow completion End-of-run now goes through the same suspension handler that processes a real suspension. Previously, items left in the invocations queue when the workflow function returned (or threw) were dropped with an "uncommitted operation" warning — `controller.abort()` called as the last statement of a workflow never actually propagated. Concretely fixes: - Abort hooks now write hook_received + stream packet so in-flight steps on other compute instances see signal.aborted=true and bail out. - Unawaited hooks are created (so external callers can resume them). - Unawaited steps and sleeps are queued (will execute / fire later). Strengthens abortTimeoutWorkflow's test to inspect the event log for the hook_received event — the original assertion only verified the workflow VM's local signal.aborted, which was set synchronously by the abort() call regardless of whether propagation actually happened. The strengthened test fails on main and passes after this commit. Drops the warnPendingQueueItems warning entirely. Drain failures are swallowed so the workflow's own outcome (return value or thrown error) remains the source of truth for the run's terminal state. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test: cover the deserialized AbortSignal listener path with an in-flight fetch The existing abort tests exercised either the polled `signal.aborted` read path (longStep busy-wait) or the already-aborted-before-fetch path. Nothing exercised the live listener path: signal starts non-aborted, step kicks off a fetch against a slow endpoint, abort fires while fetch is awaiting the response, and fetch's internal `signal.addEventListener('abort', …)` listener cancels the in-flight HTTP request. The pre-existing `fetchWithSignal` helper step was orphaned — defined but not referenced by any workflow. Wires it into a new `abortFetchInFlightWorkflow` that races a 30s fetch against a 2s sleep, aborts when the sleep wins, and returns the step's catch-path result. The test asserts both `winner=timeout` and `fetchResult.aborted=true`, which together prove fetch saw the cancellation mid-flight (the natural-completion path would set ok=true,aborted=false). Adds a local /api/delay endpoint to the nextjs-turbopack workbench so the test doesn't depend on an external service. Honors the request's own AbortSignal so cancelled connections close immediately. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test: extend abortFromStepWorkflow to verify in-flight sibling cancellation The original test only asserted that the workflow VM's signal saw aborted=true after a step called controller.abort(). It didn't actually verify that another in-flight step received the cancellation through the backing stream — those two paths are different (workflow VM signal updates via the hook event; sibling-step propagation runs through the live stream packet). Restructure the workflow to run longStep (a 30s polling loop on signal.aborted) in parallel with abortFromStep (now sleeps 1s, then aborts). The new assertion expects longStep.result === 'aborted' — proving it exited via the abort branch within ~1.5s, NOT ran to its 30s natural completion. Returning 'completed' would mean realtime cross-step cancellation is broken. abortFromStep gained an optional delayMs parameter so it can be sequenced against a sibling without an out-of-band sleep. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix: dehydrate abort stream packets via the same machinery as hook events The abort stream packet was being encoded with bare `JSON.stringify({reason})` on the writer and decoded with `JSON.parse(text).reason` on the reader. That codec drops `undefined` (so a reason-less abort wrote literally `{}` and the observability UI showed an empty stream), and doesn't handle DOMException or any other type the rest of the codebase serializes via devalue+reducers. Switch all three sites — suspension-handler workflow-side write, patched abort step-side write, and `setupAbortStreamReader` — to use `dehydrateStepArguments`/`hydrateStepArguments`. Now the `reason` round-trips with full type fidelity (DOMException, custom errors, encrypted payloads), matching what the hook event payload already does. The suspension handler literally reuses the same dehydrated bytes for the event and the stream so they're guaranteed identical. Encryption key threading: - Suspension handler: `encryptionKey` was already in scope. - Patched abort: read from `contextStorage.getStore()?.encryptionKey` (set by the step handler before invoking the deserialize chain). - Reader (`setupAbortStreamReader`): read from `contextStorage.getStore()?.encryptionKey` for the same reason; falls back to `undefined` when called outside step context (the hydrate path is key-tolerant). On-disk verification: - Before: chunk for `controller.abort()` (no reason) was `00 7b 7d` — 3 bytes, the literal JSON `{}`, no reason carried at all. - After: chunk is `00 64 65 76 6c [{"aborted":1,"reason":2},true,"test"]` — 43 bytes, devalue-flat-encoded with the reason intact. Updated the existing stream-reader unit test to encode its mock payload through the same dehydrate path so the reader can decode it. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test: cover addEventListener, mid-flight throwIfAborted, and step-initiated determinism The polled-`signal.aborted` path was the only abort consumption pattern exercised end-to-end. Three new e2e tests fill the gaps: - **abortListenerWorkflow** — `signal.addEventListener('abort', cb)` firing on the deserialized step-side signal. Distinct from abortFetchInFlightWorkflow which only proves it indirectly through fetch's internal listener; this one verifies user-attached listeners directly. Step resolves with via:'listener' if propagation worked, via:'timeout' on a 30s safety timeout if it didn't. - **abortThrowIfAbortedMidFlightWorkflow** — throwIfAborted() in a polling loop, not just at step entry. The existing abortThrowIfAbortedWorkflow only covers the synchronous-throw case on a pre-aborted signal. This one starts the signal non-aborted, polls throwIfAborted every 500ms, and aborts from a sibling step after 1s. Verifies the DOMException propagates as FatalError (no retries) when fired mid-flight. - **abortDeterministicBranchFromStepWorkflow** — counterpart to abortDeterministicBranchWorkflow, but with the abort source being a step (via the patched abort() path / hook event) instead of the workflow body. Both branch-reads MUST take the same path on every replay. Uncovered a real semantic: signal.aborted reflects step-initiated aborts only after the next promise-queue checkpoint (sleep, step await, etc.) since _setAborted is chained on promiseQueue. The test inserts the required sleep('1s') checkpoint and asserts both pre and post values. Helper steps factored: stepWaitingOnAbortListener and stepPollingThrowIfAborted. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test: drop signal.aborted shortcut in stepWaitingOnAbortListener The shortcut would have masked a regression in the addEventListener-on-an- already-aborted-signal contract. Per the AbortSignal spec, calling addEventListener('abort', cb) on an aborted signal fires the callback (on a microtask), so user code that subscribes via the listener path alone — the common pattern — depends on it. Test the contract directly: rely solely on the listener resolving the promise. If addEventListener-on-aborted ever silently breaks, this test now reports via:'timeout' instead of paving over it with a fast-path that reads signal.aborted directly. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix: add DOMException reviver to observabilityRevivers so the o11y UI hydrates abort reasons The observability UI (and CLI) hydrates step IO via `observabilityRevivers`, which had no `DOMException` entry. When a step returned a value containing a DOMException (typically `{aborted, reason: <DOMException>}` — synthesized by native AbortController when abort() is called with no reason), devalue's `parse` would throw on the `["DOMException", ...]` tag, `hydrateStepIO`'s try/catch would swallow it, and the raw devalue-flat string survived to the UI. The user-visible result was step Output showing literal text like: devl[{"aborted":1,"reason":2},true,["DOMException",3]...] instead of a JSON viewer with a proper DOMException card. Add the reviver. Reconstruct as a real DOMException when the global is available (modern browsers + Node 18+, where the o11y consumers run), falling back to a name-tagged Error otherwise. Preserves message/name/ stack/cause for display. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test: cover the external-signal-aborted-in-flight propagation path The existing abortExternalSignalWorkflow only validates a static read of an already-aborted signal — it tells us nothing about whether an abort that fires AFTER serialization actually propagates from the caller process, through the listener attached at workflow-start, into the backing stream, and out into the deserialized signals on the in-flight step compute. Add abortExternalSignalInFlightWorkflow that takes a non-aborted signal and runs two parallel consumption patterns against it: longStep (polling signal.aborted) and stepWaitingOnAbortListener (addEventListener path). The test creates a fresh AbortController, calls start() with its non-aborted signal, and aborts the source controller 1.5s later via setTimeout — well after both steps are mid-flight on their compute instances. Both consumers must see the cancellation: - pollResult === 'aborted' (NOT 'completed' — that would mean longStep ran the full 30s without ever seeing signal.aborted=true) - listenerResult.via === 'listener' (NOT 'timeout' — that would mean the addEventListener callback never fired) This exercises the longest end-to-end abort path in the codebase: caller-process AbortController → serialization-time listener → backing stream → step compute → deserialized signal → (poll OR addEventListener) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(test): use httpbin.org/delay for abortFetchInFlightWorkflow The previous setup added a /api/delay route to workbench/nextjs-turbopack to give the test a slow endpoint to fetch against. That made the workflow fail in CI on every other workbench (nextjs-webpack, astro, sveltekit, …) since the route only existed on one of them — fetch returned 404 and the test failed within 1s instead of taking the expected ~3s. Switch to httpbin.org/delay/30, the same external-service pattern used by other e2e workflows in this file (jsonplaceholder, example.com). Removes the per-workbench dependency. Drops the now-unused deploymentUrl argument from the workflow signature and test call site. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs: fix serialization page — drop duplicate header, move AbortController section Two issues on the serialization foundations page: 1. `## Pass-by-Value Semantics` appeared twice. The second occurrence had no body, which rendered as an orphaned heading just above the AbortController section in the docs preview. 2. `## AbortController & AbortSignal` was at the bottom of the page, after `## Custom Class Serialization`. It belongs above the custom-class section so the standard serializable types are grouped together before the advanced topic. Removes the empty duplicate; relocates the AbortController section to sit between Request & Response and Custom Class Serialization. No content changes inside the section. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs: note that run.cancel() is the same as the observability Cancel button The Run Cancellation section showed the programmatic path but didn't tie it back to the UI. Add a callout: calling run.cancel() is the same action as clicking the Cancel button on a run in the observability UI — both produce identical run_cancelled events. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test: cover AbortSignal.any in both workflow VM and step contexts Two distinct paths: the workflow VM ships its own AbortSignal.any impl in workflow/abort-controller.ts (composes WorkflowAbortSignals via listeners, no stream/hook backing on the composite), while steps use the native Node implementation over deserialized signals. Neither was tested. abortAnyInWorkflowWorkflow exercises the VM impl directly: creates two controllers, composes their signals via AbortSignal.any, aborts one, and asserts the composite reflects the abort synchronously without any stream round-trip. Also asserts the other source signal is unaffected so a mass-abort regression would surface here. abortAnyInStepWorkflow exercises the longest end-to-end path that uses AbortSignal.any: source controller is aborted by a sibling step, abort flows through the workflow's VM, then the backing stream, into the step's deserialized signal, into the AbortSignal.any composite, into the user's listener. Returning via:'timeout' instead of via:'listener' would mean a break anywhere on that chain. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * Update .changeset/fix-dom-exception-serialization.md Co-authored-by: Peter Wielander <mittgfu@gmail.com> Signed-off-by: Pranay Prakash <pranay.gp@gmail.com> * Update .changeset/serializable-abort-controller.md Co-authored-by: Peter Wielander <mittgfu@gmail.com> Signed-off-by: Pranay Prakash <pranay.gp@gmail.com> * Update .changeset/drain-pending-queue-on-completion.md Co-authored-by: Peter Wielander <mittgfu@gmail.com> Signed-off-by: Pranay Prakash <pranay.gp@gmail.com> * docs(errors): match slug-as-title convention + simplify the timeout example Two toolbar-comment fixes on the abort-signal-timeout-in-workflow error page: 1. The page title was Title Case ("AbortSignal.timeout() in Workflow") while every other page in docs/content/docs/errors/ uses the kebab-case slug as the title (e.g. timeout-in-workflow, fetch-in-workflow, workflow-not-registered). Match the convention. 2. The recommended replacement for AbortSignal.timeout() was a Promise.race that wrapped the abort + null sentinel + custom Error throw. Boil it down to the much simpler: const controller = new AbortController(); void sleep("10s").then(() => controller.abort()); return await fetchData(controller.signal); If fetchData finishes within 10s you get the response; if not, the timer fires controller.abort(), fetch rejects with AbortError, and the step's failure propagates to the workflow as a FatalError (no retries). Same observable behavior, no Promise.race scaffolding. Adds abortVoidSleepTimeoutWorkflow + matching e2e test that exercises this exact pattern end-to-end so the doc example is verified runnable (not just pseudocode). Asserts the fetch is cancelled mid-flight by the timer, returning aborted=true,ok=false from the step. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Signed-off-by: Pranay Prakash <pranay.gp@gmail.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> Co-authored-by: Karthik Kalyanaraman <karthik.kalyanaraman@vercel.com> Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: Peter Wielander <mittgfu@gmail.com> |
||
|
Nathan Rajlich
|
5f22832675 |
Serialize run_failed/step_failed errors through serialization pipeline (#1851)
* Serialize run_failed/step_failed errors through serialization pipeline
Switch run_failed, step_failed, and step_retrying events to persist
the full thrown value via the workflow serialization pipeline (as
SerializedData / Uint8Array) instead of a lossy { message, stack, code }
StructuredError shape. Consumers hydrate via hydrateRunError /
hydrateStepError to reconstruct the original thrown value, preserving
Error subclass identity, cause chains, and custom properties.
- WorkflowRun.error and Step.error are now SerializedData
- WorkflowRun gains a top-level errorCode plaintext field
- WorkflowRunFailedError.cause is now the hydrated thrown value
- Adds world-postgres migration 0010_add_error_code.sql
- Legacy pre-pipeline errorJson records surface as undefined on read
* Update Next.js workbenches for new WorkflowRunFailedError.cause type
cause is now `unknown` (the hydrated thrown value) rather than
`Error & { code }`. Defensively extract Error-shaped fields when the
hydrated value is an Error, otherwise round-trip the raw value, and
expose the new `errorCode` classification field.
* Update docs for WorkflowRunFailedError.cause: unknown
The hydrated `cause` is now `unknown` (the original thrown value
through the serialization pipeline) and the error classification has
moved to the top-level `errorCode` property. Update the two affected
docs pages and the `TSDoc` interface to reflect the new shape, and
narrow `cause` with `instanceof Error` before accessing fields.
* Expand test coverage for the run/step error serialization pipeline
Unit tests:
- 19 new dehydrate/hydrate{Step,Run}Error round-trip tests covering
FatalError, plain Error, built-in Error subclasses, non-Error thrown
values (string, plain object), cause chains, encryption round-trip,
the binary format prefix contract, and the unserializable / unknown-
format error paths.
- 5 new tests for Run.returnValue when the run is failed: hydrated
FatalError + cause as cause, plain Error preservation, non-Error
thrown values surfaced verbatim, cross-class cause chains, and the
hydration-failure fallback that still surfaces errorCode.
E2E tests (new, in 99_e2e.ts + e2e.test.ts):
- Step throw → workflow catch round-trips a FatalError with a TypeError
cause chain, asserting class identity, fatal marker, and cause name +
message all survive the step_failed event pipeline.
- Workflow throw → run_failed reaches status with the new
top-level errorCode metadata exposed (cause-shape coverage lives at
the unit level, since the SWC plugin's class registration is not
invoked in the plain-Node e2e runner).
- Workflow throw of a non-Error value round-trips that value verbatim
as WorkflowRunFailedError.cause.
Adjustments to existing assertions:
- error.cause is now ; tests narrow with
and use the new top-level field instead of .
- step.error / run.error from CLI --withData are now hydrated payloads:
unregistered class instances surface as Instance refs whose
carries the original message + stack.
Observability hydration:
- hydrateStepIO / hydrateWorkflowIO in serialization-format.ts now
hydrate the field via hydrateData, so the CLI and web UI
continue to surface readable run/step error messages and stacks.
* Tighten error serialization changeset description
* Trim error serialization changeset to a single sentence
* Resolve FatalError/RetryableError revivers via cross-realm registry
When a workflow runs in a Node `vm` context, its bundled
`@workflow/errors` is a different module instance than the host's
import (separate prototype chains, separate class identity). Calling
`new FatalError(...)` from the host-side reviver produces a
host-realm instance that fails `err instanceof FatalError` checks
in the workflow code — even when the serialized payload was correctly
tagged via the dedicated `FatalError` reducer.
Surfaced by the local-prod e2e "step throw round-trips FatalError"
test on Next.js Turbopack: each route gets its own bundled chunk, so
the flow handler's `@workflow/errors` and the workflow VM bundle's
`@workflow/errors` are two distinct copies of the same module.
Fix:
- Each bundled copy of `@workflow/errors` self-registers its
`FatalError` and `RetryableError` classes on `globalThis` via
`Symbol.for("@workflow/errors//FatalError")` /
`Symbol.for("@workflow/errors//RetryableError")`. First load wins
per realm; the descriptor is non-writable / non-configurable to make
accidental clobbering loud.
- The revivers in `@workflow/core`'s common reducers module read the
consumer's `globalThis` (passed in as `global`) to pick up the
realm-local class, falling back to the host-imported class when no
registration is present (e.g. in the CLI / test runner).
* Use `types.isNativeError` to remap workflow stacks across VM realms
The runtime's run-failure path computes a source-map-remapped stack
and then assigns it back onto the thrown value via `if (err
instanceof Error) err.stack = errorStack`. Workflows run inside a
Node `vm` context, so a workflow-thrown error is an instance of the
VM realm's `Error` — `instanceof` against the host realm's
`Error` returns `false`, the assignment is skipped, and the
serialized `run_failed` event carries the un-remapped (bundled-line-
number) stack instead of the source-mapped one.
Switch the gate to `types.isNativeError`, which uses V8's internal
type tag and works across realms — same approach already in place
for the serialization reducers.
Caught by the local-prod e2e "nested function calls preserve message
and stack trace" and "cross-file imports preserve message and stack
trace" tests, which assert that the persisted run-error stack
contains `99_e2e.ts` / `helpers.ts`.
* Sync CLI revivers with core + add toJSON shim for Error subclasses
Two issues with the CLI's hand-rolled reviver list:
1. It hadn't been updated for the new first-class Error subclass
reducers (`TypeError`, `RangeError`, `FatalError`, `RetryableError`,
etc.). devalue throws "Unknown type X" when it encounters a
reduced value with no matching reviver, and `hydrateResourceIO`
swallows that error and surfaces the raw `Uint8Array` payload —
so `step.error` / `run.error` showed up as raw byte dumps in
`workflow inspect` output.
2. Even with all the right revivers, `Error.prototype`'s `message`
/ `stack` / `cause` are non-enumerable, so `JSON.stringify`
(used by `workflow inspect --json`) drops them — leaving the
subclass-specific enumerable fields (e.g. `FatalError.fatal`)
visible but the actual error data missing.
Fix:
- Build the CLI reviver set on top of `getCommonRevivers()` from
`@workflow/core` so the CLI stays in sync with the runtime's
reducer set automatically. New core reducers/revivers will Just
Work without any CLI-side change.
- Wrap each Error reviver from the common set with a thin shim that
attaches a non-enumerable `toJSON` method to the produced
`Error` instance. `JSON.stringify` calls `toJSON` and gets a
full object (`name` + `message` + `stack` + `cause` + any
enumerable subclass fields like `fatal` / `retryAfter` /
`errors`); `util.inspect` ignores `toJSON` and renders the
canonical `Error: msg\\n at ...` format. Best of both worlds for
CLI output without compromising the runtime hydration path.
Caught by the local-prod e2e "basic step error preserves" and
"cross-file step error preserves" tests, which read
`failedStep.error.message` / `.stack` from the CLI's JSON output.
* Clarify parseErrorJson JSDoc to match its always-null return
The previous JSDoc described preserving legacy values "for best-effort
hydration" which contradicted the implementation, where legacy errors
are intentionally surfaced as absent (the pre-pipeline shapes can't be
hydrated by the new error revivers). Rewrite the comment so the contract
matches behavior. Also rename the now-unused parameter to `_errorJson`
to reflect that the function ignores it.
Caught by a code review on #1851.
* Refine error-handler ergonomics on the step / run hot paths
Three review-driven adjustments that all touch the queue handlers and
their interaction with the error serialization pipeline:
1. Memoize the per-run encryption key fetch. The step handler used to
eagerly fetch + import the key at the top of every step delivery so
the value would be in scope for every potential dehydrateStepError
path. That pessimized step-started early-return cases (the fetch
happens unconditionally even when the step never reaches user code)
and required duplicating the same boilerplate at four call sites in
runtime.ts. Introduce `memoizeEncryptionKey(world, run)` in
runtime/helpers.ts that returns a lazy, single-fetch accessor;
step-handler / runtime call sites use `await getEncryptionKey()`
instead. The first caller pays the fetch cost, subsequent callers
await the cached promise, and steps that fail before any
encryption-aware work happens skip the fetch entirely.
2. Preserve the prior attempt's serialized error as the cause on the
defensive max-retries-exceeded `step_failed` re-invocation guard.
The existing comment explicitly opted out of cause attachment, but
the symmetric post-failure path below already does this and the
reviewer is right that consumers shouldn't have to walk the
step_retrying event history to recover the underlying error. Best-
effort: if hydration of the prior `step.error` throws, fall back
to a FatalError without cause rather than letting the event write
itself fail.
3. Document the intentional `unflatten` throw in
`hydrateStepError` / `hydrateRunError` for non-Uint8Array input.
SDK version is pinned per workflow run via skew protection so the
non-binary branch is dead in production; if a misshapen value
reaches it, surfacing the throw via the surrounding o11y try/catch
is more debuggable than masking it. Add a comment so future
reviewers don't reach for a defensive fallback.
A standalone `falls back to plaintext` suggestion on the run_failed
key fetch was rejected: when encryption is configured we should fail
loudly rather than silently emit plaintext error data. The queue's
redelivery semantics will retry the key fetch; persistent KMS outages
get logged with the existing "persistent error preventing the run from
being terminated" message rather than a security regression.
* Hydrate `event.eventData.error` in event listings
`hydrateEventData` enumerated the per-event fields that need
hydration (`result`, `input`, `output`, `metadata`, `payload`)
but omitted the new `error` field on `step_failed`,
`step_retrying`, and `run_failed` events. Without this branch,
o11y tools that list events (e.g. `workflow inspect events`) surface
the raw `Uint8Array` payload instead of a hydrated
`{ name, message, stack, … }` object even though the entity-level
`Run.error` / `Step.error` paths already hydrate.
Mirrors the existing per-field branches; the `try/catch` leaves the
field un-hydrated on parse failure rather than failing the whole
event view. Adds a unit test.
* Use `.is()` static checks in `classifyRunError` for cross-realm safety
Workflows execute inside a separate `vm` realm: the
`WorkflowRuntimeError` class bundled into the workflow code and the
host-imported one are distinct constructors, so an
`err instanceof WorkflowRuntimeError` check on a VM-thrown error
returns `false` and we'd misclassify genuine runtime errors (corrupted
event log, missing timestamps, workflow/step not registered) as user
errors.
Switch to each subclass's `.is()` static (a name-based duck check that
works across realms). Since `WorkflowRuntimeError.is` only matches its
own concrete name, enumerate every concrete subclass we want to
recognize (`StepNotRegisteredError`, `WorkflowNotRegisteredError`)
in a `RUNTIME_ERROR_CHECKS` table; keep that table in sync with the
class hierarchy in `@workflow/errors`.
Existing `classify-error.test.ts` already covers `WorkflowRuntimeError`
and `WorkflowNotRegisteredError` cases — both still pass.
* Add e2e coverage for step throws of non-Error values
We had `errorWorkflowThrowNonErrorValue` (workflow body throws a plain
object — round-trips verbatim as `WorkflowRunFailedError.cause`) but
no symmetric coverage for the step-throw side. Step-throw goes through
a different code path: non-Error values aren't recognized as
`FatalError` (no `name === 'FatalError'`) nor `RetryableError`,
so they take the transient retry path. After max retries the runtime
wraps the original thrown value as `cause` on a fresh `FatalError`
which the workflow's catch block then sees.
Add a workflow that throws a recognizable plain object from a step
with `maxRetries = 0` (so we exhaust on first attempt and avoid a
long test wait) and a workflow that asserts the wrapped FatalError
shape: `isFatal`, `instanceof FatalError`, message includes the
original object's serialized form, `cause` is the original non-Error
object verbatim with structure preserved.
Documents the current retry-then-wrap behavior so any future change
to "non-Error throws skip retries" semantics has to update the test.
* Note legacy postgres error-data loss in the run/step error changeset
Pre-upgrade failed runs that wrote into world-postgres's deprecated
`error` text column can't be hydrated through the new pipeline (the
shape is incompatible with the new revivers). The new runtime
intentionally surfaces them as `error: undefined` on read; the
original payload is still readable directly from the `errorJson`
column for manual inspection. Add a one-sentence note to the
changeset's migration text so consumers upgrading don't get blindsided
by suddenly-empty error fields on historical runs.
|
||
|
Nathan Rajlich
|
7c45e9e213 |
Enforce per-(run, correlation) uniqueness for entity-creating events in world-postgres (#1878)
Adds a unique partial index on workflow_events(run_id, correlation_id, type) filtered to step_created/hook_created/wait_created, and translates the resulting unique-violation (pg code 23505, surfaced via DrizzleQueryError.cause) into EntityConflictError. The steps table already deduped via onConflictDoNothing, but the event row still inserted, leaving duplicate events in the log. Now both rows are kept consistent and the runtime's existing dedup catch path handles concurrent writers cleanly. |
||
|
Peter Wielander
|
8ea1532e48 | [core] Combine flow+step bundle and process steps eagerly (#1338) | ||
|
Peter Wielander
|
873b4e2bb4 | [core] Refactor getWorld interface to be asynchronous (#942) | ||
|
Peter Wielander
|
66d49c0db6 | [world] Restructure stream interface, require run ID for all step and stream operations (#1293) | ||
|
Peter Wielander
|
a5c90cefba | [core] [world] Fix community world E2E tests broken by specVersion bump (#1658) | ||
|
Peter Wielander
|
7e70d1823a | [core] Add configurable stream flush interval per world (#1533) | ||
|
Peter Wielander
|
c8dce52606 | [core] [world] Lazy run creation on start (#1537) | ||
|
Peter Wielander
|
ef2218ab22 | [world] Use zod/v4 in queue files to match @workflow/world schemas (#1588) | ||
|
Peter Wielander
|
a98f8de53f | [core] Combine initial run fetch, event fetch, and run_started event creation (#1569) | ||
|
Peter Wielander
|
329cdb3e1b | [world] Re-enqueue active runs on world restart (#1534) | ||
|
Nathan Colosimo
|
e045b59dc4 |
[world-postgres] Add maxPoolSize config for graphile (#1527)
Co-authored-by: Peter Wielander <mittgfu@gmail.com> |
||
|
Pranay Prakash
|
d1391e1fd9 |
Fix race condition allowing duplicate hook_disposed events (#1523)
* Fix race condition allowing duplicate hook_disposed events Concurrent workflow invocations could both post hook_disposed for the same hook, corrupting the event log with duplicate events. This mirrors the wait_completed race condition fixed in #1057/#1434. - world-local: Add writeExclusive lock file for hook_disposed (same pattern as wait_completed and step terminal states) - world-postgres: Use DELETE ... RETURNING to atomically detect if another caller already deleted the hook entity - suspension-handler: Improve log messages to distinguish hook-already- disposed (EntityConflictError) from run-already-completed (RunExpiredError) Fixes #1266 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * Accept either EntityConflictError or HookNotFoundError in race test The concurrent hook_disposed race has two possible orderings: the loser may hit the lock file (EntityConflictError) or find the hook entity already deleted by the winner (HookNotFoundError). Both are correct. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * Assert only one hook_disposed event in event log after race Verifies the losing concurrent caller didn't sneak an event in before the guard threw. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> |
||
|
Peter Wielander
|
01bbe66d5a | [world] Add stream pagination and metadata endpoints (#1470) | ||
|
Pranay Prakash
|
2ef33d2828 |
feat: export semantic error types and add API reference docs (#1447)
* feat: export semantic error types and add API reference documentation Add missing error exports (HookNotFoundError, EntityConflictError, RunExpiredError, TooEarlyError, ThrottleError, RunNotSupportedError, WorkflowWorldError) to workflow/internal/errors. Create new error classes for world-level semantics. Tighten TSDoc comments on all error classes. Add API reference docs for all error types. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: use @setup declarations, workflow/errors import, and errors/ doc section - Replace @skip-typecheck with proper `declare` + `// @setup` lines so code samples are typechecked but setup lines hidden from readers - Add `workflow/errors` export to package.json (public API, replaces `workflow/internal/errors` in docs) - Add `workflow/errors` path mapping in docs-typecheck type-checker - Add HookConflictError to re-export list - Move all error docs under api-reference/workflow/errors/ subdirectory - Update all internal cross-references and links Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * refactor: move error docs to top-level workflow-errors section - Move semantic error docs to api-reference/workflow-errors/ (matching the workflow/errors import path, like workflow-api for workflow/api) - Keep FatalError and RetryableError in api-reference/workflow/ since they're imported from workflow, not workflow/errors - Fix all cross-reference links Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * chore: update HTTP debug logger JSDoc to clarify scope Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: make TooEarlyError.retryAfter a number (seconds) matching WorkflowWorldError TooEarlyError.retryAfter is now seconds (number) instead of a Date, consistent with ThrottleError and WorkflowWorldError. The conversion from seconds to Date is done at the consumer site (step-handler) rather than at construction time. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: address review feedback on docs accuracy - WorkflowWorldError docs: add status, code, url, retryAfter properties to TSDoc; clarify that .is() only matches direct instances (not subclasses); use instanceof in catch-all example - TooEarlyError/ThrottleError docs: mark retryAfter as optional (?) to match actual type definitions Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> |
||
|
James Berry
|
5502438bac |
[world-postgres] Migrate client from postgres.js to pg (#1484)
|
||
|
Peter Wielander
|
78f1b0e19f | [core] Support negative startIndex for streaming (#1460) | ||
|
Pranay Prakash
|
aee035f944 |
refactor: replace HTTP status code checks with semantic error types (#1342)
* feat: classify run failure error codes and improve error logging - Add RUN_ERROR_CODES (USER_ERROR, RUNTIME_ERROR) to @workflow/errors - Populate errorCode in run_failed events via classifyRunError() - Update web UI StatusBadge to show amber dot for infrastructure errors - Improve world-local queue error logging (concise, no body dump) - Improve schema validation error messages (concise, verbose behind DEBUG) - Add e2e tests for error code flow and infrastructure error retry Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat: add semantic error types to replace HTTP status code checks in runtime Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat: classify run failure error codes and improve error logging - Add RUN_ERROR_CODES (USER_ERROR, RUNTIME_ERROR) to @workflow/errors - Populate errorCode in run_failed events via classifyRunError() - Update web UI StatusBadge to show amber dot for infrastructure errors - Improve world-local queue error logging (concise, no body dump) - Improve schema validation error messages (concise, verbose behind DEBUG) - Add e2e tests for error code flow and infrastructure error retry Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat: classify run failure error codes and improve error logging - Add RUN_ERROR_CODES (USER_ERROR, RUNTIME_ERROR) to @workflow/errors - Populate errorCode in run_failed events via classifyRunError() - Update web UI StatusBadge to show amber dot for infrastructure errors - Improve world-local queue error logging (concise, no body dump) - Improve schema validation error messages (concise, verbose behind DEBUG) - Add e2e tests for error code flow and infrastructure error retry Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * address PR review comments - Remove dead `meta` option from TooEarlyError constructor (TooTallNate) - Extract `throwWithTrace` helper to deduplicate span recording in world-vercel makeRequest (TooTallNate) - Restore `maxAttempts` const for stable retry count logging (TooTallNate) - Fix behavioral regression: add WorkflowAPIError 404 fallback in suspension-handler hook disposal to handle world-vercel path where makeRequest doesn't map 404 to HookNotFoundError (TooTallNate) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: translate 404 to HookNotFoundError at the world-vercel boundary Move the 404 → HookNotFoundError translation into world-vercel's createWorkflowRunEvent, where we know the event type context. For hook-related events (hook_created, hook_disposed, hook_received, hook_conflict), a 404 from the server means the hook was not found. This removes the WorkflowAPIError 404 fallback from the runtime's suspension-handler, keeping the runtime fully decoupled from HTTP status codes. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: parse Retry-After for 425 responses and narrow hook event set - Parse Retry-After header unconditionally so TooEarlyError gets the server-provided delay instead of always falling back to ~1s - Narrow hookEventsRequiringExistence to only hook_disposed and hook_received (matching world-local's set), since hook_created and hook_conflict don't imply the hook must already exist Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * rename WorkflowAPIError to WorkflowWorldError Breaking change: rename WorkflowAPIError → WorkflowWorldError to better reflect that this error represents world (storage backend) failures, not HTTP API errors specifically. Updated across all packages: errors, core, world-local, world-vercel, world-postgres, workflow, and web. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> |
||
|
Pranay Prakash
|
d428d66441 |
[world-postgres, world-local] Fix TOCTOU races in entity state transitions (#1434)
* [world-postgres] Fix TOCTOU race in step_started that corrupts event log The step_started UPDATE had no conditional guard on step status, allowing a concurrent execution to revert a completed step back to 'running'. This caused duplicate step_completed events, triggering CORRUPTED_EVENT_LOG. Add notInArray guard to match the existing pattern on step_completed and the DynamoDB conditional expression used in the Vercel world. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * [world-postgres] Add atomic terminal-state guards to all entity UPDATEs Add conditional WHERE clauses to match the Vercel world's DynamoDB conditional expressions, preventing TOCTOU races where concurrent requests could bypass pre-validation and write invalid state transitions. Changes: - step_started: add NOT IN (completed, failed, cancelled) guard - step_retrying: add terminal-state guard (was unguarded) - step_completed/step_failed: add cancelled to guard - run_completed/run_failed/run_cancelled: add terminal-state guards - isStepTerminal: include cancelled status Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * [world-local] Add atomic terminal-state guards and concurrent race tests Local world fixes: - step_completed/step_failed: use writeExclusive lock to prevent concurrent duplicate terminal transitions - step_started: check for terminal lock file before allowing start - wait_completed: use writeExclusive lock (port from PR #1388) - isStepTerminal: include cancelled status Tests: - Concurrent step_completed race (exactly one succeeds, one gets 409) - Concurrent step_failed race - step_started rejection after concurrent step_completed - Concurrent wait_completed race Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * Address PR review feedback - Fix lock file extension: use .lock instead of .json via taggedPath to avoid polluting entity directories with empty JSON files that cause SyntaxError during listing/parsing - Make startedAt update atomic using COALESCE in SQL instead of deriving isFirstStart from the TOCTOU validation read - Split changeset into separate entries per package - Remove unnecessary context from changeset text Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * Address review feedback from @TooTallNate - step_started: replace fs.access() with a fresh re-read of the step entity — honest about being best-effort rather than claiming atomicity (local world is dev-only; postgres world has SQL-level atomic guards) - wait_completed: clean up lock file on 404 to avoid leaked lock files Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: move lock files to .locks/ subdirectory off basedir Lock files in entity directories (steps/, waits/) broke tests that expect only tagged .json files. Move all locks to basedir/.locks/ and clean them up in clear(). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: pass ISO string to COALESCE instead of Date object The postgres driver can't serialize a Date object inside a raw sql template literal. Convert to ISO string for proper parameterization. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix: embed tag in lock file names for test isolation Parallel vitest workers with different tags would collide on untagged lock files, causing spurious 409s. Include the tag in the lock filename (e.g. stepId.terminal.vitest-0) matching the pattern from PR #1388. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> Co-authored-by: Peter Wielander <mittgfu@gmail.com> |
||
|
Nathan Colosimo
|
02ea057442 | [world-postgres] Route Graphile queue execution over workflow HTTP endpoints, fix for nextjs discovery (#1417) | ||
|
Karthik Kalyan
|
94c14c746b |
[world] When resolveData='none', only strip eventData related to refs, keep other metadata (#1364)
* add stepName with events * add changeset * add workflowname to run created * add postgres migration * update world-local * update world-local * preserve the fields in the original shape * fix tests * strip only ref/payload fields * stub the helper into world * add test coverage * fix web package * fix web package to not pass withData: true |
||
|
Nathan Colosimo
|
3648109861 |
[world-postgres] [world-local] Execute Graphile jobs directly instead of defering to world-local queue (#1334)
Signed-off-by: nathancolosimo <nathancolosimo@gmail.com> Co-authored-by: Peter Wielander <mittgfu@gmail.com> |
||
|
Peter Wielander
|
d8daa2a9a9 | [world] Add method to get individual events (#1287) | ||
|
Peter Wielander
|
11dcb646d3 | [world-local] [world-postgres] [cli] Validate local run ID and quiet dotenv logs (#1273) | ||
|
Nathan Rajlich
|
02f706fb99 |
fix(world-local, world-postgres): default hooks.list() sort order to ascending (#1275)
hooks.list() in world-local and world-postgres defaulted to descending order, while world-vercel defaults to ascending (creation order). This caused the webhookWorkflow e2e test to deadlock: hooks were returned newest-first, so the test sent the first webhook request to the manual- response hook instead of the default-response hook. - world-local: default sortOrder to 'asc', fix getCreatedAt returning new Date(0) which broke ascending cursor pagination - world-postgres: respect the sortOrder param (was hardcoded to desc) and default to 'asc' |
||
|
Pranay Prakash
|
adfe8b6b11 |
Add isWebhook flag to prevent hooks from being resumed via public webhook endpoint (#1270)
* Add isWebhook flag to prevent hooks from being resumed via public webhook endpoint Hooks created with createHook() are now non-resumable via the public webhook endpoint by default (isWebhook=false). Only hooks created with createWebhook() set isWebhook=true, allowing them to be resumed via the public URL. Also adds HookNotFoundError thrown by all world backends when a webhook token doesn't match any hook, and an e2e test for the new behavior. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * Fix world-local: default isWebhook to false and fix test assertions - Default isWebhook to false at write time in events-storage - Default isWebhook to false at read time in hooks-storage (for old data) - Update test assertions to match HookNotFoundError message Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * Fix: default isWebhook to true for backwards compat, add postgres migration - Revert read-side default to `isWebhook ?? true` in world-local for backwards compatibility with existing hooks that predate the field - Add postgres migration 0009 to add `is_webhook` column with default true - Update drizzle schema to match Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> |
||
|
Peter Wielander
|
ece45a0d7c | Lint after biome update (whitespace changes only) (#1243) | ||
|
Robert Vollmer
|
79a730aa68 |
[world-postgres] Hide also "info" logs from Graphile Worker by default (#1171)
Signed-off-by: Robert Vollmer <rovo89@users.noreply.github.com> |
||
|
Robert Vollmer
|
0735b2a55b |
[world-postgres] Fix race conditions in Postgres streamer (#1002)
Signed-off-by: Robert Vollmer <rovo89@users.noreply.github.com> |
||
|
Peter Wielander
|
dda67421cf | [world-postgres] [cli] Skip graphile logs for CLI json mode, observe DEBUG env (#1167) | ||
|
Peter Wielander
|
c9186f9870 | [world-postgres] Add migrations from pg-boss to graphile-worker queue (#1126) | ||
|
Kevin
|
1f9a67c759 | [world-postgres] Replace pg-boss with graphile-worker (#1124) |