The main problem was that an unluckily timed task cancellation could cause
the semaphore to be stuck. There were also doubts about strict FIFO ordering
of tasks allowed to pass.
The Semaphore implementation was rewritten to be more similar to Lock.
Many tests for edge cases (including cancellation) were added.
This reverts commit 0587810698.
Reason: This broke buildbots (some warnings added by that commit are turned to errors in the SSL buildbot).
Repro: ./python Lib/test/ssltests.py
Warn on loop initialization, when setting the wakeup fd disturbs a previously set wakeup fd, and on loop closing, when upon resetting the wakeup fd, we find it has been changed by someone else.
When a task catches CancelledError and raises some other error,
the other error should not silently be suppressed.
Any scenario where a task crashes in cleanup upon cancellation
will now result in an ExceptionGroup wrapping the crash(es)
instead of propagating CancelledError and ignoring the side errors.
NOTE: This represents a change in behavior (hence the need to
change several tests). But it is only an edge case.
Co-authored-by: Thomas Grainger <tagrain@gmail.com>
Once the task group is shutting down, it should not be possible to create a new task.
Here "shutting down" means `self._aborting` is set, indicating that at least one task
has failed and we have cancelled all others.
Co-authored-by: Łukasz Langa <lukasz@langa.pl>
Fix#91487
When transferring a small file, e.g. 256 KiB, the speed of this PR is comparable. However, if a large file, e.g. 65536 KiB, is transferred, asyncio UDP will be over 100 times faster than the original. The speed is presumably significantly faster if a larger file is transferred, e.g. 1048576 KiB.
Automerge-Triggered-By: GH:gpshead
The existing event loop `start_tls()` method is not sufficient for
connections using the streams API. The existing StreamReader works
because the new transport passes received data to the original protocol.
The StreamWriter must then write data to the new transport, and the
StreamReaderProtocol must be updated to close the new transport
correctly.
The new StreamWriter `start_tls()` updates itself and the reader
protocol to the new SSL transport.
Co-authored-by: Ian Good <icgood@gmail.com>
After a long deliberation we ended up feeling that the message argument for Future.cancel(), added in 3.9, was a bad idea, so we're deprecating it in 3.11 and plan to remove it in 3.13.
Example:
async with asyncio.timeout(5):
await some_task()
Will interrupt the await and raise TimeoutError if some_task() takes longer than 5 seconds.
Co-authored-by: Guido van Rossum <guido@python.org>
Also from the _asyncio C accelerator module,
and adjust one test that the change caused to fail.
For more discussion see the discussion starting here:
https://github.com/python/cpython/pull/31394#issuecomment-1053545331
(Basically, @asvetlov proposed to return False from cancel()
when there is already a pending cancellation, and I went along,
even though it wasn't necessary for the task group implementation,
and @agronholm has come up with a counterexample that fails
because of this change. So now I'm changing it back to the old
semantics (but still bumping the counter) until we can have a
proper discussion about this.)
This changes cancelling() and uncancel() to return the count of pending cancellations.
This can be used to avoid bugs in certain edge cases (e.g. two timeouts going off at the same time).
* Remove task group names (for now)
We're not sure that they are needed, and once in the code
we would never be able to get rid of them.
Yury wrote:
> Ideally, there should be a way for someone to build a "trace"
> of taskgroups/task leading to the current running task.
> We could do that using contextvars, but I'm not sure we should
> do that in 3.11.
* Pass optional name on to task in create_task()
* Remove a bunch of unused stuff
asyncio/taskgroups.py is an adaptation of taskgroup.py from EdgeDb, with the following key changes:
- Allow creating new tasks as long as the last task hasn't finished
- Raise [Base]ExceptionGroup (directly) rather than TaskGroupError deriving from MultiError
- Instead of monkey-patching the parent task's cancel() method,
add a new public API to Task
The Task class has a new internal flag, `_cancel_requested`, which is set when `.cancel()` is called successfully. The `.cancelling()` method returns the value of this flag. Further `.cancel()` calls while this flag is set return False. To reset this flag, call `.uncancel()`.
Thus, a Task that catches and ignores `CancelledError` should call `.uncancel()` if it wants to be cancellable again; until it does so, it is deemed to be busy with uninterruptible cleanup.
This new Task API helps solve the problem where TaskGroup needs to distinguish between whether the parent task being cancelled "from the outside" vs. "from inside".
Co-authored-by: Yury Selivanov <yury@edgedb.com>
Co-authored-by: Andrew Svetlov <andrew.svetlov@gmail.com>
This parameter was accidentally removed when fixing
https://bugs.python.org/issue45129, this reverts the unnecessary changes
there.
Co-authored-by: Andrew Svetlov <andrew.svetlov@gmail.com>
Due to significant security concerns, the reuse_address parameter of
asyncio.loop.create_datagram_endpoint, deprecated in Python 3.9, is
now removed. This is because of the behavior of the socket option
SO_REUSEADDR in UDP.
Co-authored-by: Łukasz Langa <lukasz@langa.pl>
Remove the @asyncio.coroutine decorator
enabling legacy generator-based coroutines to be compatible with async/await
code; remove asyncio.coroutines.CoroWrapper used for wrapping
legacy coroutine objects in the debug mode.
The decorator has been deprecated
since Python 3.8 and the removal was initially scheduled for Python 3.10.
* [Enum] reduce scope of new format behavior
Instead of treating all Enums the same for format(), only user mixed-in
enums will be affected. In other words, IntEnum and IntFlag will not be
changing the format() behavior, due to the requirement that they be
drop-in replacements of existing integer constants.
If a user creates their own integer-based enum, then the new behavior
will apply:
class Grades(int, Enum):
A = 5
B = 4
C = 3
D = 2
F = 0
Now: format(Grades.B) -> DeprecationWarning and '4'
3.12: -> no warning, and 'B'
asyncio.get_event_loop() emits now a deprecation warning when it creates a new event loop.
In future releases it will became an alias of asyncio.get_running_loop().
test_unix_events.py no longer checks if waitstatus_to_exitcode() mock
has been called or not to make the test more functional, rather than
checking the exact implementation.
This is my first issue!
So, if there's anything wrong, please tell me!
Also, thank you always for all the contributors!
Automerge-Triggered-By: GH:asvetlov
* Update code after merge review from 1st1
* Use a sentinel approach for loop parameter
Remove unnecessary _get_running_loop patching
* Use more clear function name (_verify_parameter_is_marker -> _verify_no_loop)
* Add init method to _LoopBoundMixin to check that loop param wasn't used
# Improve asyncio.wait function
The original code creates the futures set two times.
We can create this set before, avoiding the second creation.
This new behaviour [breaks the aiokafka library](https://github.com/aio-libs/aiokafka/pull/672), because it gives an iterator to that function, so the second iteration become empty.
Automerge-Triggered-By: GH:1st1
* bpo-41696: Fix handling of debug mode in asyncio.run
This allows PYTHONASYNCIODEBUG or -X dev to enable asyncio debug mode
when using asyncio.run
* 📜🤖 Added by blurb_it.
Co-authored-by: hauntsaninja <>
Co-authored-by: blurb-it[bot] <43283697+blurb-it[bot]@users.noreply.github.com>
Stopping and restarting a proactor event loop on windows can lead to
spurious errors logged (ConnectionResetError while reading from the
self pipe). This fixes the issue by ensuring that we don't attempt
to start multiple copies of the self-pipe reading loop.
Currently, if `asyncio.wait_for()` itself is cancelled it will always
raise `CancelledError` regardless if the underlying task is still
running. This is similar to a race with the timeout, which is handled
already.
When I was fixing bpo-32751 back in GH-7216 I missed the case when
*timeout* is zero or negative. This takes care of that.
Props to @aaliddell for noticing the inconsistency.
asyncio.AbstractEventLoop.run_in_executor should be a method that returns an asyncio Future, not an async method.
This matches the concrete implementations, and the documentation better.
* bpo-41273: Proactor transport read loop to use recv_into
By using recv_into instead of recv we do not allocate a new buffer each
time _loop_reading calls recv.
This betters performance for any stream using proactor (basically any
asyncio stream on windows).
* bpo-41273: Double proactor read transport buffer size
By doubling the read buffer size we get better performance.
When I wrote the documentation for `asyncio.to_thread()`, I mistakenly assumed that `return await loop.run_in_executor(...)` within an async def function would return a Future. In reality, it returns a coroutine.
This likely won't affect typical usage of `asyncio.to_thread()`, but it's important for the documentation to be correct here. In general, we also tend to avoid returning futures from high-level APIs in asyncio.
Implements `asyncio.to_thread`, a coroutine for asynchronously running IO-bound functions in a separate thread without blocking the event loop. See the discussion starting from [here](https://github.com/python/cpython/pull/18410#issuecomment-628930973) in GH-18410 for context.
Automerge-Triggered-By: @aeros
When an asyncio.Task is cancelled, the exception traceback now
starts with where the task was first interrupted. Previously,
the traceback only had "depth one."
Currently, if asyncio.wait_for() timeout expires, it cancels
inner future and then always raises TimeoutError. In case
those future is task, it can handle cancelation mannually,
and those process can lead to some other exception. Current
implementation silently loses thoses exception.
To resolve this, wait_for will check was the cancelation
successfull or not. In case there was exception, wait_for
will reraise it.
Co-authored-by: Roman Skurikhin <roman.skurikhin@cruxlab.com>
Motivation for this PR (comment from @vstinner in bpo issue):
```
Warning seen o AMD64 Ubuntu Shared 3.x buildbot:
https://buildbot.python.org/all/#/builders/141/builds/2593
test_devnull_output (test.test_a=syncio.test_subprocess.SubprocessThreadedWatcherTests) ...
Warning -- threading_cleanup() failed to cleanup 1 threads (count: 1, dangling: 2)
```
The following implementation details for the new method are TBD:
1) Public vs private
2) Inclusion in `close()`
3) Name
4) Coroutine vs subroutine method
5) *timeout* parameter
If it's a private method, 3, 4, and 5 are significantly less important.
I started with the most minimal implementation that fixes the dangling threads without modifying the regression tests, which I think is particularly important. I typically try to avoid directly modifying existing tests as much as possible unless it's necessary to do so. However, I am open to changing any part of this.
https://bugs.python.org/issue38356
If waitpid() is called elsewhere, waitpid() call fails with
ChildProcessError: use return code 255 in this case, and log a
warning. It ensure that the pidfd file descriptor is closed if this
error occurs.
Capturing exceptions into names can lead to reference cycles though the __traceback__ attribute of the exceptions in some obscure cases that have been reported previously and fixed individually. As these variables are not used anyway, we can remove the binding to reduce the chances of creating reference cycles.
See for example GH-13135
* This just copies the docs from `StreamWriter` and `StreamReader`.
* Add docstring for asyncio functions.
https://bugs.python.org/issue36889
Automerge-Triggered-By: @asvetlov
This PR deprecate explicit loop parameters in all public asyncio APIs
This issues is split to be easier to review.
fourth step: queue.py
https://bugs.python.org/issue36373
This PR deprecate explicit loop parameters in all public asyncio APIs
This issues is split to be easier to review.
Third step: locks.py
https://bugs.python.org/issue36373
This PR deprecate explicit loop parameters in all public asyncio APIs
This issues is split to be easier to review.
Second step: streams.py
https://bugs.python.org/issue36373
Add BaseEventLoop.wait_executor_on_close attribute: true by default.
loop.close() now waits for the default executor to finish by default.
Set loop.wait_executor_on_close attribute to False to not wait for
the executor.
This will address the common mistake many asyncio users make:
an "except Exception" clause breaking Tasks cancellation.
In addition to this change, we stop inheriting asyncio.TimeoutError
and asyncio.InvalidStateError from their concurrent.futures.*
counterparts. There's no point for these exceptions to share the
inheritance chain.
In 3.9 we'll focus on implementing supervisors and cancel scopes,
which should allow better handling of all exceptions, including
SystemExit and KeyboardInterrupt
This PR proposes a solution to [bpo-35545](https://bugs.python.org/issue35545) by adding an optional `flowinfo` and `scopeid` to `asyncio.base_events._ipaddr_info` to carry the full address information into `_ipaddr_info` and avoid discarding IPv6 specific information.
Changelog entry & regression tests to come.
https://bugs.python.org/issue35545
When the future returned by shield is cancelled, its completion callback of the
inner future is not removed. This makes the callback list of inner inner future
grow each time a shield is created and cancelled.
This change unregisters the callback from the inner future when the outer
future is cancelled.
https://bugs.python.org/issue35125
*Moved from python/asyncio#493.*
This PR fixes issue python/asyncio#480, as explained in [this comment](https://github.com/python/asyncio/issues/480#issuecomment-278703828).
The `_SelectorDatagramTransport.sendto` method has to be modified ~~so `_sock.sendto` is used in all cases (because it is tricky to reliably tell if the socket is connected or not). Could that be an issue for connected sockets?~~ *EDIT* ... so `_sock.send` is used only if `_sock` is connected.
It also protects `socket.getsockname` against `OSError` in `_SelectorTransport`. This might happen on Windows if the socket is not connected (e.g. for UDP broadcasting).
https://bugs.python.org/issue31922
Added two keyword arguments, `delay` and `interleave`, to
`BaseEventLoop.create_connection`. Happy eyeballs is activated if
`delay` is specified.
We now have documentation for the new arguments. `staggered_race()` is in its own module, but not exported to the main asyncio package.
https://bugs.python.org/issue33530
Cyker Way
Kumar Aditya
Guido van Rossum
Michel Hidalgo
Hendrik Makait
Pamela Fox
Kirill
Yury Selivanov
Łukasz Langa
Thomas Grainger
Victor Stinner
Oleg Iarygin
David Hewitt
Alexander Shadchin
msoxzw
Andrew Svetlov
Vincent Bernat
Duprat
Maximilian Hils
Alex Grönholm
Stefan Zabka
Serhiy Storchaka
Tin Tvrtković
Nikita Sobolev
Alex Waygood
Emiya
Jim Crist-Harif
Kumar Aditya
Sam Bull
Gregory P. Smith
Joongi Kim
Christian Clauss
Hugo van Kemenade
Jesús Cea
Mariusz Felisiak
Illia Volochii
Ethan Furman
Binbin
Pablo Galindo
Christian Heimes
Roberto Hueso
Richard Kojedzinszky
Chris Jerdonek
Zackery Spytz
Wansoo Kim
Yurii Karabas
Diogo Dutra
Jakub Stasiak
Shantanu
Ben Darnell
Elvis Pranskevichus
Cleber Rosa
James Weaver
Vinay Sharma
Tony Solomonik
Rémi Lapeyre
Kyle Stanley
Fantix King
Kunal Bhalla
romasku
Galden
Lidi Zheng
Markus Mohrhard
idomic
Batuhan Taşkaya
Hill Ma
Benjamin Peterson
Emmanuel Arias
Xtreak
Min ho Kim
Victor Stinner
依云
sbstp
Yury Selivanov
Matthias Bussonnier
Niklas Fiekas
Erwan Le Pape
Romain Picard
Vincent Michel
twisteroid ambassador
gescheit
Quentin Dawans