@tiens.nguyen/gonext-local-worker 1.0.22 → 1.0.23

package/gonext-local-worker.mjs

@@ -162,6 +162,10 @@ async function runChatJob(job) {
   /** Chains debounced chunk POSTs so we never PATCH `completed` while a chunk POST is still in flight. */
   let flushTail = Promise.resolve();
 
+  /** Batch streamed text: fewer HTTPS round-trips to the API than a 12ms debounce per flush. */
+  const CHUNK_DEBOUNCE_MS = 80;
+  const CHUNK_MAX_BUF = 6144;
+
   const flushChunks = async () => {
     const t = buf;
     buf = "";
@@ -199,13 +203,23 @@ async function runChatJob(job) {
     if (!s) return;
     fullText += s;
     buf += s;
+    if (buf.length >= CHUNK_MAX_BUF) {
+      if (flushTimer) {
+        clearTimeout(flushTimer);
+        flushTimer = null;
+      }
+      flushTail = flushTail.then(() => flushChunks()).catch((err) => {
+        console.error("[gonext-worker] chunk flush error:", err);
+      });
+      return;
+    }
     if (!flushTimer) {
       flushTimer = setTimeout(() => {
         flushTimer = null;
         flushTail = flushTail.then(() => flushChunks()).catch((err) => {
           console.error("[gonext-worker] chunk flush error:", err);
         });
-      }, 12);
+      }, CHUNK_DEBOUNCE_MS);
     }
   };
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tiens.nguyen/gonext-local-worker",
-  "version": "1.0.22",
+  "version": "1.0.23",
   "description": "Polls GoNext cloud API for async local LLM jobs and runs them against Ollama/OpenAI-compatible servers on this Mac",
   "type": "module",
   "license": "MIT",