[trissim]

+26 / −80 across 2 files. Queue in message_pump.py goes from a 50-line hand copy of asyncio.Queue to a 15-line subclass of it.

Before / after

• Before: class Queue(Generic[T]) with its own deque, ready_event, wakeup-recheck loop and a task_done that did nothing.
• After: class Queue(asyncio.Queue[T]) keeps only what asyncio lacks:
  • peek_nowait() reads self._queue[0] without taking the head from a waiting consumer.
  • activity (from The pilot waits only on pumps that owe work; the queue is the pump's and says so #252): OWING is set in asyncio's _put hook, the same subclass protocol PriorityQueue and LifoQueue use. SETTLED is set when get finds the queue empty.

Deleted

• The deque, the Event, the recheck loop and the task_done stub.
• Three tests that pinned the copy's own semantics: one item per waiter, recheck after a drained wakeup, no lost wakeup on a cancelled waiter. CPython's asyncio.Queue owns those guarantees.
• In their place: one test for peek_nowait and one for owing/settled.

Behaviour difference (traced)

The copy re-marked SETTLED every time a woken waiter found the queue drained again. Now that happens once per get call. The only synchronous drain is get_nowait in _discard_pending_messages (message_pump.py:693), which runs after the pump has stopped. So the pilot can see OWING slightly longer, never SETTLED early.

Proof

tests/test_queue.py, test_message_pump.py, test_message_pump_payload_lifetime.py, test_real_app_pump_lifetime.py, test_pilot.py: 169 passed in 21 s in Toad's test env against this checkout.

🤖 Generated with Claude Code

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota. Try requesting a review at a lower effort level.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing.