vllm_omni.engine.duplex.session.append_task ¶

One append in flight, and what its failure owes the session.

An append is not a call: it is a task queued behind the previous append, which may be cancelled, may find the session closed before it starts, and on any of those paths has to give back what it reserved --- the PCM bytes it took from the input budget, the committed audio it was going to consume, and the response it precreated.

That compensation is the reason this is an object. The five steps of an append all need the same eleven values, so as closures over _start_append they read as one function with five entry points and no way to test any of them. Named fields make the captured state explicit and the compensation a method.

logger = init_logger(__name__)

before_append: Callable[[], bool] | None = None

ctx: DuplexSessionContext

epoch: int

fail_session: Callable[[str], None]

final: bool

model: ModelChannel

on_append_accepted: Callable[[float], None] | None = None

operation_id: str | None

out: SessionEmitter

payload: dict[str, object]

pcm_reservation: PcmAppendReservation | None

precreated_response_id: str | None

request_id: str

retained_committed_payload: dict[str, object] | None

abandon() -> None

clear_pending_silence(done: Task[bool]) -> None

discard_retained_audio() -> None

release_on_failure(done: Task[bool]) -> None

run_in_wire_order(predecessor: Task[bool] | None) -> bool