1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
|
.. currentmodule:: asyncio
.. _asyncio-policies:
========
Policies
========
An event loop policy is a global object
used to get and set the current :ref:`event loop <asyncio-event-loop>`,
as well as create new event loops.
The default policy can be :ref:`replaced <asyncio-policy-get-set>` with
:ref:`built-in alternatives <asyncio-policy-builtin>`
to use different event loop implementations,
or substituted by a :ref:`custom policy <asyncio-custom-policies>`
that can override these behaviors.
The :ref:`policy object <asyncio-policy-objects>`
gets and sets a separate event loop per *context*.
This is per-thread by default,
though custom policies could define *context* differently.
Custom event loop policies can control the behavior of
:func:`get_event_loop`, :func:`set_event_loop`, and :func:`new_event_loop`.
Policy objects should implement the APIs defined
in the :class:`AbstractEventLoopPolicy` abstract base class.
.. _asyncio-policy-get-set:
Getting and Setting the Policy
==============================
The following functions can be used to get and set the policy
for the current process:
.. function:: get_event_loop_policy()
Return the current process-wide policy.
.. function:: set_event_loop_policy(policy)
Set the current process-wide policy to *policy*.
If *policy* is set to ``None``, the default policy is restored.
.. _asyncio-policy-objects:
Policy Objects
==============
The abstract event loop policy base class is defined as follows:
.. class:: AbstractEventLoopPolicy
An abstract base class for asyncio policies.
.. method:: get_event_loop()
Get the event loop for the current context.
Return an event loop object implementing the
:class:`AbstractEventLoop` interface.
This method should never return ``None``.
.. versionchanged:: 3.6
.. method:: set_event_loop(loop)
Set the event loop for the current context to *loop*.
.. method:: new_event_loop()
Create and return a new event loop object.
This method should never return ``None``.
.. method:: get_child_watcher()
Get a child process watcher object.
Return a watcher object implementing the
:class:`AbstractChildWatcher` interface.
This function is Unix specific.
.. deprecated:: 3.12
.. method:: set_child_watcher(watcher)
Set the current child process watcher to *watcher*.
This function is Unix specific.
.. deprecated:: 3.12
.. _asyncio-policy-builtin:
asyncio ships with the following built-in policies:
.. class:: DefaultEventLoopPolicy
The default asyncio policy. Uses :class:`SelectorEventLoop`
on Unix and :class:`ProactorEventLoop` on Windows.
There is no need to install the default policy manually. asyncio
is configured to use the default policy automatically.
.. versionchanged:: 3.8
On Windows, :class:`ProactorEventLoop` is now used by default.
.. deprecated:: 3.12
The :meth:`get_event_loop` method of the default asyncio policy now emits
a :exc:`DeprecationWarning` if there is no current event loop set and it
decides to create one.
In some future Python release this will become an error.
.. class:: WindowsSelectorEventLoopPolicy
An alternative event loop policy that uses the
:class:`SelectorEventLoop` event loop implementation.
.. availability:: Windows.
.. class:: WindowsProactorEventLoopPolicy
An alternative event loop policy that uses the
:class:`ProactorEventLoop` event loop implementation.
.. availability:: Windows.
.. _asyncio-watchers:
Process Watchers
================
A process watcher allows customization of how an event loop monitors
child processes on Unix. Specifically, the event loop needs to know
when a child process has exited.
In asyncio, child processes are created with
:func:`create_subprocess_exec` and :meth:`loop.subprocess_exec`
functions.
asyncio defines the :class:`AbstractChildWatcher` abstract base class, which child
watchers should implement, and has four different implementations:
:class:`ThreadedChildWatcher` (configured to be used by default),
:class:`MultiLoopChildWatcher`, :class:`SafeChildWatcher`, and
:class:`FastChildWatcher`.
See also the :ref:`Subprocess and Threads <asyncio-subprocess-threads>`
section.
The following two functions can be used to customize the child process watcher
implementation used by the asyncio event loop:
.. function:: get_child_watcher()
Return the current child watcher for the current policy.
.. deprecated:: 3.12
.. function:: set_child_watcher(watcher)
Set the current child watcher to *watcher* for the current
policy. *watcher* must implement methods defined in the
:class:`AbstractChildWatcher` base class.
.. deprecated:: 3.12
.. note::
Third-party event loops implementations might not support
custom child watchers. For such event loops, using
:func:`set_child_watcher` might be prohibited or have no effect.
.. class:: AbstractChildWatcher
.. method:: add_child_handler(pid, callback, *args)
Register a new child handler.
Arrange for ``callback(pid, returncode, *args)`` to be called
when a process with PID equal to *pid* terminates. Specifying
another callback for the same process replaces the previous
handler.
The *callback* callable must be thread-safe.
.. method:: remove_child_handler(pid)
Removes the handler for process with PID equal to *pid*.
The function returns ``True`` if the handler was successfully
removed, ``False`` if there was nothing to remove.
.. method:: attach_loop(loop)
Attach the watcher to an event loop.
If the watcher was previously attached to an event loop, then
it is first detached before attaching to the new loop.
Note: loop may be ``None``.
.. method:: is_active()
Return ``True`` if the watcher is ready to use.
Spawning a subprocess with *inactive* current child watcher raises
:exc:`RuntimeError`.
.. versionadded:: 3.8
.. method:: close()
Close the watcher.
This method has to be called to ensure that underlying
resources are cleaned-up.
.. deprecated:: 3.12
.. class:: ThreadedChildWatcher
This implementation starts a new waiting thread for every subprocess spawn.
It works reliably even when the asyncio event loop is run in a non-main OS thread.
There is no noticeable overhead when handling a big number of children (*O(1)* each
time a child terminates), but starting a thread per process requires extra memory.
This watcher is used by default.
.. versionadded:: 3.8
.. class:: MultiLoopChildWatcher
This implementation registers a :py:data:`SIGCHLD` signal handler on
instantiation. That can break third-party code that installs a custom handler for
:py:data:`SIGCHLD` signal.
The watcher avoids disrupting other code spawning processes
by polling every process explicitly on a :py:data:`SIGCHLD` signal.
There is no limitation for running subprocesses from different threads once the
watcher is installed.
The solution is safe but it has a significant overhead when
handling a big number of processes (*O(n)* each time a
:py:data:`SIGCHLD` is received).
.. versionadded:: 3.8
.. deprecated:: 3.12
.. class:: SafeChildWatcher
This implementation uses active event loop from the main thread to handle
:py:data:`SIGCHLD` signal. If the main thread has no running event loop another
thread cannot spawn a subprocess (:exc:`RuntimeError` is raised).
The watcher avoids disrupting other code spawning processes
by polling every process explicitly on a :py:data:`SIGCHLD` signal.
This solution is as safe as :class:`MultiLoopChildWatcher` and has the same *O(N)*
complexity but requires a running event loop in the main thread to work.
.. deprecated:: 3.12
.. class:: FastChildWatcher
This implementation reaps every terminated processes by calling
``os.waitpid(-1)`` directly, possibly breaking other code spawning
processes and waiting for their termination.
There is no noticeable overhead when handling a big number of
children (*O(1)* each time a child terminates).
This solution requires a running event loop in the main thread to work, as
:class:`SafeChildWatcher`.
.. deprecated:: 3.12
.. class:: PidfdChildWatcher
This implementation polls process file descriptors (pidfds) to await child
process termination. In some respects, :class:`PidfdChildWatcher` is a
"Goldilocks" child watcher implementation. It doesn't require signals or
threads, doesn't interfere with any processes launched outside the event
loop, and scales linearly with the number of subprocesses launched by the
event loop. The main disadvantage is that pidfds are specific to Linux, and
only work on recent (5.3+) kernels.
.. versionadded:: 3.9
.. _asyncio-custom-policies:
Custom Policies
===============
To implement a new event loop policy, it is recommended to subclass
:class:`DefaultEventLoopPolicy` and override the methods for which
custom behavior is wanted, e.g.::
class MyEventLoopPolicy(asyncio.DefaultEventLoopPolicy):
def get_event_loop(self):
"""Get the event loop.
This may be None or an instance of EventLoop.
"""
loop = super().get_event_loop()
# Do something with loop ...
return loop
asyncio.set_event_loop_policy(MyEventLoopPolicy())
|