worker API

完整接口目录 · 插件运行时用法 · 工具包下载

模块说明

Native managed worker lifecycle. Legacy runtime domains use their own loader.

Worker

class Worker

SDK 说明：

Own a fresh peer and async callbacks, running events serially.

Callbacks receive (peer, payload). Use spawn() for cooperative background
work; detached tasks/threads are not covered by drain. The host must pause and join delivery
before drain, and separately join world actions. No event is auto-acknowledged.
run() owns peer cleanup; cancellation closes IPC and joins cooperative tasks.
Do not consume this peer's queues elsewhere. CPU-blocking/uncooperative code
still requires the host's process termination boundary.

Worker.__init__

def __init__(
    self,
    peer,
    *,
    on_event,
    on_validate,
    on_activate,
    on_drain,
    on_service = None,
    on_bus = None,
    on_maintenance = None,
    views = None,
    max_background_tasks = 32,
)

Worker.session

@property
def session(self)

SDK 说明：

Immutable hello metadata, available in all lifecycle callbacks.

Capabilities are the handshake snapshot, not a live permission lease.

Worker.spawn

def spawn(self, coroutine, *, name=None)

SDK 说明：

Start bounded cooperative work during activation or active delivery.

Child errors fail the worker even when the returned Task is awaited.
Drain cancels and joins these tasks; restart timers on resumed events if
needed. No work may be started during validation, drain or shutdown.

Worker.run

async def run(self)

SDK 说明：

Handshake and serve until cancellation, transport failure or callback error.

Worker.peer

self.peer

公开实例属性，在 __init__ 中设置。

Worker.log

self.log

公开实例属性，在 __init__ 中设置。