toga-ai 1.0.108 → 1.0.109

package/knowledge/2.0/apps/worker2/features/netsuite-opportunity-sync.md

@@ -40,10 +40,13 @@ second, independently-gated concern in the same handler.
 `test/@dave/NetSuite/api-message-queue/` — upload all four `.js` to one File Cabinet folder):
 - `lib_amq_queue.js` — shared module: record-type/field ids, `transmit()`, `findSendableIds()`.
 - `ue_api_msg_queue_enqueue.js` — User Event on the synced source records (Opportunity, …):
-  writes one queue row per create/edit/delete.
-- `ue_amq_drain.js` — User Event on the **API Message Queue** record: async trigger — kicks the
-  scheduled drainer via `N/task` (no inline HTTP, so it never blocks the source save).
-- `ss_amq_drain.js` — Scheduled Script: catch-up drainer (retries + anything the UE missed).
+  writes one queue row per create/edit/delete, then **kicks the scheduled drainer via `N/task`**
+  (async, non-blocking). The immediate trigger lives here — see gotchas for why.
+- `ss_amq_drain.js` — Scheduled Script: drains rows (the on-demand kick *and* a cron safety-net).
+  Needs **two deployments**: one `Scheduled` (cron) + one `Not Scheduled` (the on-demand slot
+  `task.create` submits into).
+- `ue_amq_drain.js` — **RETIRED.** A UE on the queue record can't fire on enqueuer-created rows
+  (UE-can't-trigger-UE); undeploy it.
 - The custom record itself: **"API Message Queue"**, internal id **1185**.
 
 **worker2 side:**
@@ -59,12 +62,12 @@ second, independently-gated concern in the same handler.
    (`{recordType, internalId, eventType, action}`), `headers`. It maps `record.Type`
    (lowercase) → the camelCase `recordType` worker2 expects (explicit map; e.g.
    `salesorder`→`salesOrder`), and `context.type` → CRUD verb (`create`→POST, `edit`/`xedit`→PUT,
-   `delete`→DELETE).
-2. **Trigger + drain (NetSuite):** the queue record's `ue_amq_drain` UE fires the scheduled
-   `ss_amq_drain` via `N/task` (async); `ss_amq_drain` also runs on its own schedule.
-   `transmit()` claims a row (`Created`→`Sending`) before the HTTP call, POSTs to the row's
-   endpoint, then **deletes on 2xx** / sets `Retry` (+exponential backoff) / `Failed`
-   (terminal — never auto-deleted; Jeff's data-loss rule).
+   `delete`→DELETE). It then **kicks the drainer**:
+   `task.create({taskType:SCHEDULED_SCRIPT, scriptId:'customscript_ss_amq_drain'})` — scriptId only,
+   no deployment id. If no on-demand slot is free the kick is skipped and the cron backs it up.
+2. **Drain (NetSuite):** `ss_amq_drain` runs (from the kick or its cron). `transmit()` claims a row
+   (`Created`→`Sending`) before the HTTP call, POSTs to the row's endpoint, then **deletes on 2xx** /
+   sets `Retry` (+exponential backoff) / `Failed` (terminal — never auto-deleted; Jeff's data-loss rule).
 3. **Ingest (prod):** `webhook.togahub.com/netsuite` → WebhookIngestion **Python Lambda** →
    INSERT `Core.WorkerJobs` + SQS → returns 200 immediately. Fully async; outcome lives in
    `WorkerJobs.isSuccess`/`failureReason`.
@@ -97,6 +100,16 @@ None — platform-wide Forecast sync.
 
 ## Gotchas / known issues
 
+- **A User Event can't trigger another User Event → the immediate trigger lives in the enqueuer.**
+  The enqueuer (a UE) creating the queue row does NOT fire any UE on the queue record (only the
+  *scheduled* drainer's `submitFields` does). So a drain-UE on the queue record never runs for
+  enqueued rows — the enqueuer kicks the drainer itself via `task.create`. (`ue_amq_drain.js` is
+  retired for exactly this reason. Evidence: its log only ever showed `xedit`/`Sending`, never `create`.)
+- **`task.create(SCHEDULED_SCRIPT)` needs a `Not Scheduled` deployment.** A `Scheduled` deployment is
+  reserved for its recurrence → `NO_DEPLOYMENTS_AVAILABLE`; a `Testing` deployment can't be submitted
+  at all. Give `ss_amq_drain` a dedicated `Not Scheduled` deployment as the on-demand slot (add more
+  for concurrency) and keep a `Scheduled` one for the cron. Submit by **scriptId only** — the
+  deployment id is prefix-doubled (`customdeploycustomdeploy_ss_amq_drain`).
 - **worker2 routes EVERY request to `_Controller_Index::worker()`** — `worker2/_.php` sets
   `DEFAULT_CONTROLLER_METHOD = ['_Controller_Index','worker']`, which short-circuits `_Route`'s
   URI matching. There is no per-URL routing, and `.htaccess` denies direct `.php` files, so you
@@ -124,13 +137,16 @@ None — platform-wide Forecast sync.
   string throws `Object of class stdClass could not be converted to string`. Normalize before
   matching.
 - **Drain dedup** uses a `Sending` claim status: `transmit()` flips `Created`→`Sending` before the
-  HTTP call so a concurrent pass skips it; the immediate UE trigger only fires for `Created`/
-  `Pending`, so the drainer's own `Sending`/`Retry`/`Failed` writes don't re-trigger it (a
-  `skip (status not triggerable)` debug line on those is expected). `findSendableIds` must list
-  every sendable status (`Created`,`Pending`,`Retry`,`Sending`) — omitting `Created` silently
-  matches nothing (`candidates:0`).
+  HTTP call so a concurrent drainer pass skips it (`Sending` ∉ sendable), narrowing the double-send
+  window to the load→claim gap. `findSendableIds` must list every sendable status
+  (`Created`,`Pending`,`Retry`,`Sending`) — omitting `Created` silently matches nothing (`candidates:0`).
 
 ## Change history
+- 2026-06-16 — Trigger correction: the immediate trigger moved INTO the enqueuer (`task.create`
+  the drainer) after discovering a UE can't trigger a UE — `ue_amq_drain` retired. `task.create`
+  needs a `Not Scheduled` deployment of `ss_amq_drain` (else `NO_DEPLOYMENTS_AVAILABLE`); submit by
+  scriptId only (deployment id is prefix-doubled). Fixed an `itemType`-as-object crash in the
+  importer. POST/DELETE coded but only PUT (edit) verified live. (dfranks)
 - 2026-06-16 — Built the queue enqueuer/drainer SuiteScripts, the worker2 `/netsuite` router, and
   the `_Worker_Netsuite_Opportunity` Forecast importer (header + items + status), with ClickUp
   parity. Verified opportunity 5483750 imports end-to-end in dev via ngrok. (dfranks)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "toga-ai",
-  "version": "1.0.108",
+  "version": "1.0.109",
   "description": "TOGA Technology Team Claude Knowledge System — shared AI coding harness with skills, knowledge base CLI, and project installer for Claude Code.",
   "keywords": [
     "claude",