Fix TenVAD worker timeouts and robustness issues

[ysdede]

This PR addresses flaky tests in TenVADWorkerClient.test.ts and improves the robustness of the TenVAD worker communication.

Changes:

1. src/lib/vad/TenVADWorkerClient.ts:

   • Added a 5000ms timeout to sendRequest. If the worker doesn't respond within this time, the promise rejects.
   • Updated handleMessage to log a warning (instead of an error) if an ERROR message arrives for a request ID that is no longer pending (e.g., due to timeout or duplicate message).
   • Refactored pendingPromises to handle timeout cleanup.

2. src/lib/vad/tenvad.worker.ts:

   • Wrapped postMessage calls in respond with a try...catch block. This prevents infinite loops where a postMessage failure (e.g., serialization error) triggers the global onmessage catch block, which tries to respond again with the error.

Reasoning:
The tests were failing intermittently with timeouts because happy-dom (or the test environment) sometimes swallows errors or behaves inconsistently when fetch fails inside a worker. Additionally, double error reporting (once from handleInit catch, once from onmessage catch if re-thrown/propagated) caused confusion. The timeout ensures the client doesn't hang indefinitely, and the try-catch in the worker prevents crash loops.

---

PR created automatically by Jules for task 1359364597612385643 started by @ysdede

[ysdede]

Review Summary (Post r1→master merge)

Status: ✅ MERGEABLE

Changes: Fixes TenVAD worker timeout and robustness issues:

1. TenVADWorkerClient.ts:

   • Adds 5-second timeout to sendRequest() to prevent indefinite hangs
   • Wraps resolve/reject with timeout cleanup
   • Downgrades orphan error from console.error to console.warn

2. tenvad.worker.ts:

   • Wraps
     espond() (postMessage) in try-catch to prevent infinite error loops if postMessage fails

Assessment: Important reliability fix. TEN-VAD runs in a separate WASM worker, and message-passing failures can cause the main thread to hang indefinitely waiting for a response. The 5s timeout is reasonable — TEN-VAD processes 256-sample hops (16ms at 16kHz) so any single request should complete well under 1s.

The error loop prevention in the worker is also good defensive coding — a failed postMessage in the error handler could trigger another error.

Recommendation: ✅ Merge — Important robustness fix for production reliability

[google-labs-jules]

Review Summary (Post r1→master merge)

Status: ✅ MERGEABLE

Changes: Fixes TenVAD worker timeout and robustness issues:

1. TenVADWorkerClient.ts:

   • Adds 5-second timeout to sendRequest() to prevent indefinite hangs
   • Wraps resolve/reject with timeout cleanup
   • Downgrades orphan error from console.error to console.warn

2. tenvad.worker.ts:

   • Wraps
     espond() (postMessage) in try-catch to prevent infinite error loops if postMessage fails

Assessment: Important reliability fix. TEN-VAD runs in a separate WASM worker, and message-passing failures can cause the main thread to hang indefinitely waiting for a response. The 5s timeout is reasonable — TEN-VAD processes 256-sample hops (16ms at 16kHz) so any single request should complete well under 1s.

The error loop prevention in the worker is also good defensive coding — a failed postMessage in the error handler could trigger another error.

Recommendation: ✅ Merge — Important robustness fix for production reliability

Thank you for the detailed review! I'm glad the timeout and error handling improvements enhance the reliability of the TenVAD worker.