feat(trigger-record-change): record-after-write fires one flow on create OR update (#3427) (#3446)

os-zhuang · claude · web-flow · commit 6f55c6300b9c · 2026-07-24T23:06:49.000+08:00
A record_change flow's start node bound to exactly one lifecycle event via `triggerType`, so a rule meant to run on both insert and update forced authors to duplicate the whole flow into two near-identical definitions that drift. Add `record-after-write` / `record-before-write` — the create-OR-update union (delete excluded: a write persists field data, a delete removes the row). One start node binds both lifecycle hooks (afterInsert + afterUpdate) under the same flow; exactly one fires per mutation (a write is an insert xor an update), so it is not a double run. To branch on which event fired, test `previous` (empty on create, populated on update). - New `triggerTypeToHookEvents` (plural) is the canonical mapper, expanding `write` to both events; `triggerTypeToHookEvent` (singular) is kept for back-compat and returns null for the multi-event write tokens. - The engine already forwards any `record-*` token to this trigger, so no engine/lint/spec change is needed — the trigger owns the vocabulary. - Unit + end-to-end tests (a single write flow fires on both create and update, self-write does not loop); docs under Automation > Flows and the README. Claude-Session: https://claude.ai/code/session_018939yJ413zG3irLzcTtqaa Co-authored-by: Claude <noreply@anthropic.com>
diff --git a/.changeset/record-after-write-trigger.md b/.changeset/record-after-write-trigger.md
@@ -0,0 +1,29 @@
+---
+"@objectstack/trigger-record-change": minor
+---
+
+feat(trigger-record-change): `record-after-write` fires one flow on create OR update (#3427)
+
+A `record_change` flow's `start` node bound to exactly one lifecycle event via
+`triggerType`, so a rule meant to run on both insert and update ("recompute the
+SLA whenever a case is created or its priority changes") forced authors to
+duplicate the whole flow — two near-identical definitions that drift.
+
+Adds `record-after-write` and `record-before-write` as the **create-OR-update
+union** trigger tokens. One `start` node binds both lifecycle hooks
+(`afterInsert` + `afterUpdate`) under the same flow; exactly one fires per
+mutation (a write is an insert *xor* an update), so it is not a double run.
+`delete` is deliberately excluded — a write persists field data, a delete
+removes the row. To branch on which event fired inside the flow, test
+`previous` (empty on create, populated on update).
+
+- `triggerTypeToHookEvents(triggerType)` (new, plural) is the canonical mapper:
+  it returns the list of hook events a token binds, expanding `write` to both.
+  `triggerTypeToHookEvent` (singular) is kept for back-compat and now returns
+  `null` for the multi-event `write` tokens rather than silently dropping a
+  binding.
+- The engine already forwards any `record-*` token through to this trigger, so
+  no engine, lint, or spec change is needed — the trigger owns the vocabulary.
+
+Documented under Automation › Flows (Create-or-update flow) and the trigger's
+README.
diff --git a/content/docs/automation/flows.mdx b/content/docs/automation/flows.mdx
@@ -816,10 +816,55 @@ flows: [
 ],
 ```
 
+### Create-or-update flow (`record-after-write`)
+
+A `record_change` start node binds to lifecycle events through `triggerType`.
+The single-event tokens map one-to-one:
+
+| `triggerType`          | Fires on                        |
+| :--------------------- | :------------------------------ |
+| `record-after-create`  | insert (after)                  |
+| `record-after-update`  | update (after)                  |
+| `record-after-delete`  | delete (after)                  |
+| `record-after-write`   | **insert *or* update (after)**  |
+| `record-before-create` | insert (before)                 |
+| `record-before-update` | update (before)                 |
+| `record-before-delete` | delete (before)                 |
+| `record-before-write`  | **insert *or* update (before)** |
+
+For a rule that must run on **both create and update** — "recompute the SLA
+whenever a case is created or its priority changes" — use `record-after-write`
+instead of duplicating the flow. `write` is the create-OR-update union (delete
+is excluded: a write persists field data, a delete removes the row). One `start`
+node binds both lifecycle events; exactly one fires per mutation, so it is not a
+double run.
+
+```typescript
+{
+  id: 'start',
+  type: 'start',
+  label: 'Start',
+  config: {
+    objectName: 'crm_case',
+    triggerType: 'record-after-write', // created OR updated
+  },
+}
+```
+
+To branch on **which** event fired, test `previous` — it is empty on create
+(there was no prior row) and populated on update:
+
+```typescript
+// Edge conditions are bare CEL (ADR-0032) — no {…} braces.
+{ id: 'e_created', source: 'start', target: 'on_create', condition: 'previous == null' },
+{ id: 'e_updated', source: 'start', target: 'on_update', condition: 'previous != null' },
+```
+
 ### Best practices
 
 ✅ **DO:**
 - Keep flows simple and focused
+- Prefer one `record-after-write` flow over two near-identical create/update copies
 - Document the business logic
 - Test recursion and retry behavior
 - Use scheduled flows for batch updates
diff --git a/content/docs/getting-started/common-patterns.mdx b/content/docs/getting-started/common-patterns.mdx
@@ -242,7 +242,10 @@ export const crm = defineApp({
 Create an automation that fires when a record changes. The **start node's `config`
 binds the trigger**: `objectName` plus one lifecycle event in `triggerType`
 (`record-after-create`, `record-after-update`, `record-after-delete`, or the
-`record-before-*` forms) — without it the flow never fires.
+`record-before-*` forms) — without it the flow never fires. To fire on **create
+or update in one flow**, use `record-after-write` (the create-OR-update union)
+instead of duplicating the definition — see
+[Flows › Create-or-update flow](/docs/automation/flows#create-or-update-flow-record-after-write).
 
 ```typescript
 import { defineFlow } from '@objectstack/spec';
diff --git a/packages/triggers/trigger-record-change/README.md b/packages/triggers/trigger-record-change/README.md
@@ -31,14 +31,43 @@ node gate) before running the flow, with `record` (the new row) and `previous`
 
 ### Trigger event → hook mapping
 
-| `triggerType`            | ObjectQL hook  |
-| ------------------------ | -------------- |
-| `record-after-create`    | `afterInsert`  |
-| `record-after-update`    | `afterUpdate`  |
-| `record-after-delete`    | `afterDelete`  |
-| `record-before-create`   | `beforeInsert` |
-| `record-before-update`   | `beforeUpdate` |
-| `record-before-delete`   | `beforeDelete` |
+| `triggerType`            | ObjectQL hook(s)              |
+| ------------------------ | ----------------------------- |
+| `record-after-create`    | `afterInsert`                 |
+| `record-after-update`    | `afterUpdate`                 |
+| `record-after-delete`    | `afterDelete`                 |
+| `record-after-write`     | `afterInsert` + `afterUpdate` |
+| `record-before-create`   | `beforeInsert`                |
+| `record-before-update`   | `beforeUpdate`                |
+| `record-before-delete`   | `beforeDelete`                |
+| `record-before-write`    | `beforeInsert` + `beforeUpdate` |
+
+`record-after-create` / `record-after-insert` are synonyms (both → `afterInsert`).
+
+### Create **or** update in one flow: `record-*-write`
+
+`record-after-write` (and its before-phase form) is the **create-OR-update
+union** — one `start` node that fires on both insert and update, so a
+"recompute whenever a record is created or changed" rule needs **one** flow, not
+two near-identical copies. It binds both lifecycle hooks under the same flow;
+exactly one fires per mutation (a write is an insert *xor* an update), so it is
+not a double-dispatch. `delete` is deliberately excluded — a write persists
+field data, a delete removes the row.
+
+```ts
+{
+  type: 'start',
+  config: {
+    objectName: 'crm_case',
+    triggerType: 'record-after-write', // created OR updated
+  },
+}
+```
+
+To branch on *which* happened inside the flow, test `previous`: it is empty on
+create (there was no prior row) and populated on update. For example, an edge
+condition `previous == null` selects the create path, `previous != null` the
+update path.
 
 ## Usage
 
diff --git a/packages/triggers/trigger-record-change/src/index.ts b/packages/triggers/trigger-record-change/src/index.ts
@@ -4,6 +4,7 @@ export { RecordChangeTriggerPlugin } from './plugin.js';
 export {
     RecordChangeTrigger,
     triggerTypeToHookEvent,
+    triggerTypeToHookEvents,
 } from './record-change-trigger.js';
 export type {
     FlowTrigger,
diff --git a/packages/triggers/trigger-record-change/src/record-change-integration.test.ts b/packages/triggers/trigger-record-change/src/record-change-integration.test.ts
@@ -109,12 +109,37 @@ function stampFlow(name: string, object: string) {
   };
 }
 
+/**
+ * A `record-after-write` flow (create OR update, #3427) that mirrors the
+ * record's live `status` into `mirror` on every write. Its own update_record
+ * write-back also fires afterUpdate, so this doubles as coverage that the
+ * engine's re-entrancy guard suppresses the self-trigger loop a write flow now
+ * exposes (afterUpdate IS bound, unlike a create-only flow).
+ */
+function mirrorWriteFlow(name: string, object: string) {
+  return {
+    name,
+    label: name,
+    type: 'record_change',
+    nodes: [
+      { id: 'start', type: 'start', label: 'Start', config: { objectName: object, triggerType: 'record-after-write' } },
+      { id: 'mirror', type: 'update_record', label: 'Mirror', config: { objectName: object, filter: { id: '{record.id}' }, fields: { mirror: '{record.status}' } } },
+      { id: 'end', type: 'end', label: 'End' },
+    ],
+    edges: [
+      { id: 'e1', source: 'start', target: 'mirror' },
+      { id: 'e2', source: 'mirror', target: 'end' },
+    ],
+  };
+}
+
 const objectDef = (name: string) => ({
   name,
   label: name,
   fields: {
     status: { name: 'status', label: 'S', type: 'text' },
     stamp: { name: 'stamp', label: 'St', type: 'text' },
+    mirror: { name: 'mirror', label: 'M', type: 'text' },
   },
 });
 
@@ -192,4 +217,37 @@ describe('record-change trigger — end-to-end (#1491)', () => {
     const row = await data.findOne('wid2', { where: { id } });
     expect(row?.stamp).toBe('done');
   }, 15000);
+
+  it('a single record-after-write flow fires on BOTH create and update (#3427)', async () => {
+    const kernel = new ObjectKernel({ logLevel: 'silent' });
+    await kernel.use(new ObjectQLPlugin());
+    await kernel.use(new AutomationServicePlugin());
+    await kernel.use(new RecordChangeTriggerPlugin());
+    await kernel.bootstrap();
+
+    const objectql = kernel.getService('objectql') as any;
+    const data = kernel.getService('data') as any;
+    const automation = kernel.getService<AutomationEngine>('automation');
+
+    objectql.registerDriver(makeMemoryDriver(), true);
+    objectql.registry.registerObject(objectDef('wid3'), 'test', 'test');
+    automation.registerFlow('mirror_write', mirrorWriteFlow('mirror_write', 'wid3') as any);
+
+    expect((automation as any).getActiveTriggerBindings()).toContainEqual({
+      flowName: 'mirror_write',
+      triggerType: 'record_change',
+    });
+
+    // Create — the afterInsert leg fires; the flow mirrors status → mirror.
+    const created = await data.insert('wid3', { status: 'a' });
+    const id = Array.isArray(created) ? created[0]?.id : created?.id ?? created;
+    await sleep(200);
+    expect((await data.findOne('wid3', { where: { id } }))?.mirror).toBe('a');
+
+    // Update — the afterUpdate leg of the SAME flow fires; mirror re-syncs. (The
+    // flow's own write-back does not loop: the re-entrancy guard suppresses it.)
+    await data.update('wid3', { id, status: 'b' });
+    await sleep(200);
+    expect((await data.findOne('wid3', { where: { id } }))?.mirror).toBe('b');
+  }, 15000);
 });
diff --git a/packages/triggers/trigger-record-change/src/record-change-trigger.test.ts b/packages/triggers/trigger-record-change/src/record-change-trigger.test.ts
@@ -6,6 +6,7 @@ import type { HookContext } from '@objectstack/spec/data';
 import {
     RecordChangeTrigger,
     triggerTypeToHookEvent,
+    triggerTypeToHookEvents,
     type FlowTriggerBinding,
     type RecordChangeDataEngine,
     type TriggerLogger,
@@ -90,6 +91,35 @@ describe('triggerTypeToHookEvent', () => {
         expect(triggerTypeToHookEvent('record-after-frobnicate')).toBeNull();
         expect(triggerTypeToHookEvent('on_update')).toBeNull();
     });
+
+    it('returns null for the multi-event write token (use triggerTypeToHookEvents)', () => {
+        // `write` maps to TWO events, which the singular mapper cannot express —
+        // it returns null rather than silently dropping one binding.
+        expect(triggerTypeToHookEvent('record-after-write')).toBeNull();
+        expect(triggerTypeToHookEvent('record-before-write')).toBeNull();
+    });
+});
+
+// ─── triggerTypeToHookEvents ────────────────────────────────────────
+
+describe('triggerTypeToHookEvents', () => {
+    it('maps single-lifecycle tokens to a one-element list', () => {
+        expect(triggerTypeToHookEvents('record-after-create')).toEqual(['afterInsert']);
+        expect(triggerTypeToHookEvents('record-after-update')).toEqual(['afterUpdate']);
+        expect(triggerTypeToHookEvents('record-before-delete')).toEqual(['beforeDelete']);
+        expect(triggerTypeToHookEvents('record-after-insert')).toEqual(['afterInsert']);
+    });
+
+    it('expands `write` into the create-OR-update union (#3427)', () => {
+        expect(triggerTypeToHookEvents('record-after-write')).toEqual(['afterInsert', 'afterUpdate']);
+        expect(triggerTypeToHookEvents('record-before-write')).toEqual(['beforeInsert', 'beforeUpdate']);
+    });
+
+    it('returns an empty list for unsupported / missing tokens', () => {
+        expect(triggerTypeToHookEvents(undefined)).toEqual([]);
+        expect(triggerTypeToHookEvents('schedule')).toEqual([]);
+        expect(triggerTypeToHookEvents('record-after-frobnicate')).toEqual([]);
+    });
 });
 
 // ─── RecordChangeTrigger ────────────────────────────────────────────
@@ -116,6 +146,52 @@ describe('RecordChangeTrigger', () => {
         expect(hooks).toHaveLength(0);
     });
 
+    it('binds BOTH afterInsert and afterUpdate for record-after-write (create OR update, #3427)', () => {
+        const { engine, hooks } = fakeEngine();
+        const trigger = new RecordChangeTrigger(engine, silentLogger());
+
+        trigger.start(binding({ event: 'record-after-write' }), async () => {});
+
+        // One start node → both lifecycle hooks, same object, same packageId
+        // (so a single stop() tears both down).
+        expect(hooks).toHaveLength(2);
+        expect(hooks.map((h) => h.event).sort()).toEqual(['afterInsert', 'afterUpdate']);
+        expect(hooks.every((h) => h.object === 'showcase_task')).toBe(true);
+        expect(new Set(hooks.map((h) => h.packageId)).size).toBe(1);
+        expect(hooks[0].packageId).toBe('com.objectstack.trigger.record-change:task_assigned_notify');
+    });
+
+    it('a record-after-write flow fires on both the insert hook and the update hook', async () => {
+        const { engine, hooks } = fakeEngine();
+        const trigger = new RecordChangeTrigger(engine, silentLogger());
+        let fired = 0;
+
+        trigger.start(binding({ event: 'record-after-write' }), async () => {
+            fired += 1;
+        });
+
+        const insertHook = hooks.find((h) => h.event === 'afterInsert')!;
+        const updateHook = hooks.find((h) => h.event === 'afterUpdate')!;
+
+        // Insert: no previous row.
+        await insertHook.handler(hookCtx({ event: 'afterInsert', previous: undefined }));
+        // Update: previous row present.
+        await updateHook.handler(hookCtx({ event: 'afterUpdate' }));
+
+        expect(fired).toBe(2);
+    });
+
+    it('stop() tears down BOTH hooks of a record-after-write flow', () => {
+        const { engine, hooks } = fakeEngine();
+        const trigger = new RecordChangeTrigger(engine, silentLogger());
+
+        trigger.start(binding({ event: 'record-after-write' }), async () => {});
+        expect(hooks).toHaveLength(2);
+
+        trigger.stop('task_assigned_notify');
+        expect(hooks).toHaveLength(0);
+    });
+
     it('warns when the flow targets an object the engine does not know (silent-miss guard)', () => {
         // 2026-07-17 third-party eval: a flow whose start-node `objectName`
         // does not match any registered object binds a hook that never fires —
diff --git a/packages/triggers/trigger-record-change/src/record-change-trigger.ts b/packages/triggers/trigger-record-change/src/record-change-trigger.ts
