fix(tests): stop LCM thread leak in cross-wall planning tests

[jeff-hykin]

Problem

pytest-slow failing

Solution

async instead of threads

• I have read and approved the CLA.

[greptile-apps]

Greptile Summary

This PR fixes LCM thread leaks in cross-wall planning tests by replacing a daemon thread polling loop with asyncio.add_reader, and wires LCMResourceLogger.stop() into StatsMonitor.stop() to prevent the transport thread from surviving after the coordinator shuts down.

• conftest.py: The _lcm_loop thread is replaced with loop.add_reader(lcm_fd, _on_lcm_readable), removing the need for a lock on shared odometry state; run_cross_wall_test drives the async function via asyncio.run(), and --numprocesses=auto is dropped from bin/pytest-slow so tests no longer interfere with each other over shared LCM multicast.
• logger.py / monitor.py: stop() is added to the ResourceLogger Protocol, implemented on both StructlogResourceLogger (no-op) and LCMResourceLogger (delegates to pLCMTransport.stop()), and then called from StatsMonitor.stop() to close the transport thread.
• bridge.py: visual_override type is corrected to dict[Glob | str, Callable[[Any], Archetype] | None], matching the existing None-means-suppress logic already in _visual_override_for_entity_path.

Confidence Score: 5/5

Safe to merge — the asyncio reader pattern is correct for single-process sequential tests, the transport thread leak is properly closed, and removing parallel process execution eliminates the LCM multicast collision root cause.

All shared mutable state in conftest.py is now accessed exclusively on the asyncio event loop thread, so dropping the lock is correct. The stop() wiring through the ResourceLogger protocol is complete across all implementations.

conftest.py — worth verifying that no future pytest-asyncio configuration introduces a running event loop that would conflict with asyncio.run() in run_cross_wall_test.

Important Files Changed

Filename	Overview
dimos/navigation/nav_stack/tests/conftest.py	Thread + lock replaced with asyncio event loop reader; locks removed since all odom state is now accessed on the single asyncio thread; cleanup moved to finally block with remove_reader. Logic is sound for sequential tests.
dimos/core/resource_monitor/monitor.py	StatsMonitor.stop() now calls self._logger.stop() after joining the collection thread, closing the previously leaking pLCMTransport background thread.
dimos/core/resource_monitor/logger.py	stop() added to ResourceLogger Protocol, StructlogResourceLogger (no-op), and LCMResourceLogger (delegates to pLCMTransport.stop()), fixing the transport thread leak.
bin/pytest-slow	Removes --numprocesses=auto so slow tests run sequentially, preventing multiple test workers from colliding over shared LCM multicast channels.
dimos/visualization/rerun/bridge.py	Type annotation for visual_override corrected to allow None values, matching the existing suppression logic already present in _visual_override_for_entity_path.

Sequence Diagram

sequenceDiagram
    participant Test as run_cross_wall_test
    participant AR as asyncio.run
    participant EL as event loop
    participant LCM as lcm fd reader
    participant Coord as ModuleCoordinator

    Test->>AR: call asyncio.run(_run_cross_wall_test)
    AR->>EL: create and run event loop
    EL->>Coord: ModuleCoordinator.build(blueprint)
    EL->>EL: add_reader(lcm_fd, _on_lcm_readable)
    EL->>EL: await asyncio.sleep (yields control)
    EL->>LCM: fd readable - calls _on_lcm_readable
    LCM->>LCM: lcm.handle - dispatches _odom_handler
    LCM-->>EL: odom_count / robot_x / robot_y updated
    EL->>EL: assert odom received and goals reached
    EL->>EL: finally: remove_reader, lcm.unsubscribe
    EL->>Coord: coordinator.stop
    AR-->>Test: return or propagate AssertionError

Loading

_{Reviews (6): Last reviewed commit: "-" | Re-trigger Greptile}

[codecov]

Codecov Report

❌ Patch coverage is 21.05263% with 30 lines in your changes missing coverage. Please review.
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
dimos/navigation/nav_stack/tests/conftest.py	9.67%	28 Missing ⚠️
...on/nav_stack/tests/test_cross_wall_planning_far.py	0.00%	1 Missing ⚠️
...nav_stack/tests/test_cross_wall_planning_simple.py	0.00%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!