@pylonsync/react 0.3.179 → 0.3.181

package/package.json

@@ -3,7 +3,7 @@
   "publishConfig": {
     "access": "public"
   },
-  "version": "0.3.179",
+  "version": "0.3.181",
   "type": "module",
   "main": "src/index.ts",
   "types": "src/index.ts",
@@ -12,8 +12,8 @@
     "check": "tsc -p tsconfig.json --noEmit"
   },
   "dependencies": {
-    "@pylonsync/sdk": "0.3.179",
-    "@pylonsync/sync": "0.3.179"
+    "@pylonsync/sdk": "0.3.181",
+    "@pylonsync/sync": "0.3.181"
   },
   "peerDependencies": {
     "react": ">=19.0.0"

package/src/hooks.ts

@@ -493,11 +493,23 @@ export function useMutation<TArgs = Record<string, unknown>, TResult = unknown>(
       }
 
       try {
-        const result = await callFn<TResult>(
-          fnName,
-          serverArgs as Record<string, unknown>,
-          { token: tokenRef.current }
-        );
+        // Route through SyncEngine.fn when one is wired so the response's
+        // X-Pylon-Change-Seq triggers a fallback pull if the WS broadcast
+        // hasn't landed yet. Falling back to the free callFn for hooks
+        // wired without a sync engine (rare — only legacy non-React-init
+        // callers) keeps backwards compatibility, but apps using
+        // useMutation in the normal `init()` flow now match db.fn's
+        // change-seq behavior.
+        const result = sync
+          ? await sync.fn<TResult>(
+              fnName,
+              serverArgs as Record<string, unknown>,
+            )
+          : await callFn<TResult>(
+              fnName,
+              serverArgs as Record<string, unknown>,
+              { token: tokenRef.current },
+            );
         if (mounted.current) setData(result);
         return result;
       } catch (e) {