Performance: pipewire_native.py uses time.sleep polling in find_nodes_by_pid

[m96-chan]

Summary

The PipeWireNodeDiscovery class uses a busy-polling pattern with time.sleep instead of proper event-driven waiting.

Location

src/proctap/backends/pipewire_native.py:1027-1037

Problem

# Run main loop for a short time to discover nodes
# Note: This is a simplified approach; a proper implementation
# would use a timer to stop the loop after timeout
if self._pw._main_loop:
    import time
    start = time.time()
    while (time.time() - start) * 1000 < timeout_ms:
        # Give PipeWire time to process events
        time.sleep(0.01)  # 10ms polling
        # Break early if we found nodes
        if self._found_nodes:
            break

Impact

• Wastes CPU cycles on polling
• 10ms sleep granularity means potentially 10ms wasted latency
• Does not actually process PipeWire events (no pw_loop_iterate call)
• The comment itself acknowledges this is a "simplified approach"

Suggested Fix

Use PipeWire's native event loop:

# Use pw_loop_iterate to properly process events
loop = _pw_lib.pw_main_loop_get_loop(self._pw._main_loop)
while not self._found_nodes and elapsed_ms < timeout_ms:
    # Process pending events with small timeout
    _pw_lib.pw_loop_iterate(loop, 10)  # 10ms timeout

Or use a threading.Event that gets set from the callback:

self._found_event = threading.Event()
# In callback: self._found_event.set()
# In find_nodes_by_pid:
self._found_event.wait(timeout=timeout_ms / 1000)

Labels

performance, enhancement