buildgrid.server.threading module

class buildgrid.server.threading.ContextThreadPoolExecutor(max_workers: int | None = None, thread_name_prefix: str = '', initializer: Callable[[], Any] | None = None, initargs: tuple[Any, ...] = (), immediate_copy: bool = False)

Bases: ThreadPoolExecutor

submit(fn: ~collections.abc.Callable[[~_P], ~buildgrid.server.threading._T], /, *args: ~typing.~_P, **kwargs: ~typing.~_P) → Future[_T]

Submits a callable to be executed with the given arguments.

Schedules the callable to be executed as fn(*args, **kwargs) and returns a Future instance representing the execution of the callable.

Returns:

A Future representing the given call.

class buildgrid.server.threading.ContextThread(target: Callable[[_P], _T], name: str | None = None, args: Iterable[Any] = (), kwargs: dict[str, Any] | None = None, *, daemon: bool | None = None)

Bases: Thread

class buildgrid.server.threading.ContextWorker(target: Callable[[Event], None], name: str | None = None, *, on_shutdown_requested: Callable[[], None] | None = None)

Bases: object

is_alive() → bool
start() → None
stop() → None
wait(timeout: float | None = None) → None
buildgrid.server.threading.run_concurrently(fn: Callable[[T], R], items: Sequence, executor: Executor | None, max_helpers: int | None = None) → list[R]

Apply fn to every item and return the results in input order.

The calling thread always processes items itself. Up to max_helpers (default len(items) - 1) helper tasks are submitted to executor and process items too if a worker picks them up before the work runs out; helpers that haven’t started by then are cancelled. A saturated (or absent) executor therefore degrades to running everything sequentially on the calling thread, never to waiting in a queue.

Each call to fn runs in its own copy of the caller’s context, wherever it runs, so context variables set by fn don’t leak into the caller or other items.

If any call raises, no further items are started, in-flight calls are awaited, and the exception from the lowest-index failing item is re-raised, as Executor.map would. Items are started in input order, so every item before a failing one has also run.