@odla-ai/chapter 0.58.0 → 0.58.1

package/README.md

@@ -1850,6 +1850,15 @@ These bite silently — a smoke test won't catch them:
   named and whether that person joined without folding the seat status machine
   itself; `NamedSeatRecipientLine` renders it.
 
+  **Availability saves land on the seeded group (0.58.1).** A group's `id` is
+  its public key, not its entity id: the CLI seeds the row under a fresh entity
+  id and writes `id` as an attribute, which is how every live chapter's group is
+  stored. `PUT /api/admin/scheduling` keyed its write by the `id` string, so
+  odla-db targeted a phantom entity and the strict schema refused it
+  (`required_attribute`) — a 500 on 0.57.0, a 503 with that `cause` on 0.58.0.
+  The write now goes through the `groups.id` lookup like every other group
+  write, and the test double seeds the group the way the CLI does.
+
   **Availability saves keep what the editor does not show (0.58.0).**
   `PUT /api/admin/scheduling` now merges the fields it receives onto the
   stored config before validating, so an editor that sends only hours, days,

package/dist/worker/index.cjs

@@ -11149,7 +11149,7 @@ var handleAdminScheduling = async (req, url, env, ctx) => {
   const checked = validateScheduling({ ...stored, ...body });
   if (!checked.ok) return json({ error: "invalid scheduling config", errors: checked.errors }, 400);
   try {
-    await db.transact([{ t: "update", ns: "groups", id: String(group.id), attrs: { schedulingJson: checked.value } }]);
+    await db.transact([{ t: "update", ns: "groups", id: { ns: "groups", attr: "id", value: String(group.id) }, attrs: { schedulingJson: checked.value } }]);
   } catch (error) {
     const code = error.code;
     console.error("admin_scheduling_write_failed", { code: typeof code === "string" ? code : "unknown" });