GUI Room example app. - #600
Conversation
![devin-ai-integration[bot]](https://cdn.statically.io/img/avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com>
| if self._test_video_task and not self._test_video_task.cancelled(): | ||
| self._test_video_task.cancel() | ||
| return False |
There was a problem hiding this comment.
🟡 toggle_test_video/toggle_test_audio permanently stuck after task completes due to error
The toggle methods use not self._test_video_task.cancelled() to decide whether the task is still active. asyncio.run_coroutine_threadsafe returns a concurrent.futures.Future. After the coroutine completes due to an error (e.g., publish fails because room disconnected), the future is in FINISHED state — cancelled() returns False and done() returns True. Subsequent calls to toggle_test_video() always enter the "stop" branch at line 169, call cancel() on the already-finished future (which returns False and does nothing), and return False. The user can never start a new test video/audio until reconnecting. The fix is to use .done() instead of .cancelled() to check if the task has already completed.
Was this helpful? React with 👍 or 👎 to provide feedback.
| if self._test_audio_task and not self._test_audio_task.cancelled(): | ||
| self._test_audio_task.cancel() | ||
| return False |
There was a problem hiding this comment.
🟡 toggle_test_audio has same stuck-after-error bug as toggle_test_video
Same issue as the video toggle: toggle_test_audio uses not self._test_audio_task.cancelled() at line 233, which becomes permanently stuck after the audio publishing task completes due to an error. Should use .done() instead.
Was this helpful? React with 👍 or 👎 to provide feedback.
| def _on_local_track_unpublished(self, publication): | ||
| logger.info("Local track unpublished: %s", publication.sid) | ||
| identity = self.room.local_participant.identity | ||
| track_sid = publication.sid | ||
| self._track_to_participant.pop(track_sid, None) | ||
| stream = self._video_streams.pop(identity, None) | ||
| if stream: | ||
| asyncio.ensure_future(stream.aclose()) | ||
| self._post_participants() |
There was a problem hiding this comment.
🔴 _on_local_track_unpublished unconditionally removes video stream regardless of track kind
When any local track is unpublished (including audio), the handler at line 393 unconditionally pops from self._video_streams for the local participant identity. If the user has both test video and test audio publishing, stopping the audio track triggers _on_local_track_unpublished, which incorrectly closes the video stream. Compare with _on_local_track_published at examples/wxpy_room/wxpy_room.py:376 which correctly checks track.kind == rtc.TrackKind.KIND_VIDEO. The unpublish handler should similarly check publication.kind (available via TrackPublication.kind at livekit-rtc/livekit/rtc/track_publication.py:44) before removing the video stream.
| def _on_local_track_unpublished(self, publication): | |
| logger.info("Local track unpublished: %s", publication.sid) | |
| identity = self.room.local_participant.identity | |
| track_sid = publication.sid | |
| self._track_to_participant.pop(track_sid, None) | |
| stream = self._video_streams.pop(identity, None) | |
| if stream: | |
| asyncio.ensure_future(stream.aclose()) | |
| self._post_participants() | |
| def _on_local_track_unpublished(self, publication): | |
| logger.info("Local track unpublished: %s", publication.sid) | |
| identity = self.room.local_participant.identity | |
| track_sid = publication.sid | |
| self._track_to_participant.pop(track_sid, None) | |
| if publication.kind == rtc.TrackKind.KIND_VIDEO: | |
| stream = self._video_streams.pop(identity, None) | |
| if stream: | |
| asyncio.ensure_future(stream.aclose()) | |
| self._post_participants() |
Was this helpful? React with 👍 or 👎 to provide feedback.
| except Exception as e: | ||
| self._post_state("error", str(e)) |
There was a problem hiding this comment.
🟡 Event loop thread leaked forever on connection error
When _async_connect catches a connection error at line 311, it posts an "error" state but the coroutine returns normally. The caller _run_loop at line 276 then calls self.loop.run_forever(), which blocks forever since nothing ever calls loop.stop(). The wx-side error handler at line 1111 creates a new RoomManager, orphaning the old thread and event loop. Each failed connection attempt leaks a daemon thread.
Was this helpful? React with 👍 or 👎 to provide feedback.
1 participant
This is a room example app written in wxPython, used to test the basic functionality of the Python SDK.
usage: