Expand description
Built-in executors and related tools.
All asynchronous computation occurs within an executor, which is capable of spawning futures as tasks. This module provides several built-in executors, as well as tools for building your own.
This module is only available when the executor
feature of this
library is activated.
Using a thread pool (M:N task scheduling)
Most of the time tasks should be executed on a thread pool.
A small set of worker threads can handle a very large set of spawned tasks
(which are much lighter weight than threads). Tasks spawned onto the pool
with the spawn_ok
function will run ambiently on
the created threads.
Spawning additional tasks
Tasks can be spawned onto a spawner by calling its spawn_obj
method
directly. In the case of !Send
futures, spawn_local_obj
can be used
instead.
Single-threaded execution
In addition to thread pools, it’s possible to run a task (and the tasks
it spawns) entirely within a single thread via the LocalPool
executor.
Aside from cutting down on synchronization costs, this executor also makes
it possible to spawn non-Send
tasks, via spawn_local_obj
. The
LocalPool
is best suited for running I/O-bound tasks that do relatively
little work between I/O operations.
There is also a convenience function block_on
for simply running a
future to completion on the current thread.
Structs
enter
if an execution scope has already been
entered.