Yury Selivanov added the comment:
New changeset 3e9751819ad13a965e8be13c1e5bc5a6811fe6b8 by Yury Selivanov in
branch 'master':
bpo-32273: Move asyncio.test_utils to test.test_asyncio (#4785)
https://github.com/python/cpython/commit/3e9751819ad13a965e8be13c1e5bc5
Change by Yury Selivanov :
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.or
Yury Selivanov added the comment:
> `asyncio._get_running_loop()` will be deprecated and removed in Python 3.9.
_get_running_loop() was left as-is.
--
___
Python tracker
<https://bugs.python.org/issu
Yury Selivanov added the comment:
> See https://eklitzke.org/goroutines-nonblocking-io-and-memory-usage for an
> interesting discussion of the drawbacks of some buffer handling idioms.
Thanks for the link!
It does make sense to use a pool of buffers for the proposed BufferedProtocol
Yury Selivanov added the comment:
>> Looks nice. Can it speed up aiohttp too?
> Yes.
> aiohttp uses own streams but public API and internal implementation are
> pretty close to asyncio streams.
> Moreover C accelerated HTTP parser should work with proposed BufferedProtocol
New submission from Yury Selivanov :
asyncio.get_event_loop(), and, subsequently asyncio._get_running_loop() are one
of the most frequently executed functions in asyncio. They also can't be sped
up by third-party event loops like uvloop.
When implemented in C they become 4x f
Change by Yury Selivanov :
--
keywords: +patch
pull_requests: +4719
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32296>
___
___
Py
Yury Selivanov added the comment:
First, I've no questions about the proposed implementation. It shouldn't have
performance impact when unawaited coroutine tracking is off, which is the
default. It will cause minimal overhead when the tracking is on, which is
fine. Adding
Yury Selivanov added the comment:
> send_ping() # don't care about result, forgot await
> # get collected
> await something_that_will_trigger_check_coros_weakreaf() # oh no !
I don't understand what you are trying to say with thi
Yury Selivanov added the comment:
New changeset a70232f28882d2fecb3ebe06643867701016070f by Yury Selivanov in
branch 'master':
bpo-32296: Implement asyncio.get_event_loop and _get_running_loop in C. (#4827)
https://github.com/python/cpython/commit/a70232f28882d2fecb3ebe06643867
Change by Yury Selivanov :
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
type: -> performance
___
Python tracker
<https://bugs.python
Yury Selivanov added the comment:
I don't like the low-level API of run_in_executor. "executor" being the first
argument, the inability to pass **kwargs, etc.
I'd expect to see a more high-level API, perhaps the one that supports 'async
with':
async wit
Change by Yury Selivanov :
--
pull_requests: +4739
stage: resolved -> patch review
___
Python tracker
<https://bugs.python.org/issue32296>
___
___
Python-
Yury Selivanov added the comment:
Thanks Victor. I've made a PR to fix this.
--
___
Python tracker
<https://bugs.python.org/issue32296>
___
___
Pytho
New submission from Yury Selivanov :
There's a fairly extensive discussion here:
https://github.com/python/asyncio/pull/465
In short, asyncio.run() is a pretty straightforward addition, so let's add it.
The discussion was more focused on the asyncio.run_forever() proposal. I now
Change by Yury Selivanov :
--
keywords: +patch
pull_requests: +4741
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32314>
___
___
Py
Yury Selivanov added the comment:
I'll open a separate issue to improve Server's API. AFAIK it's the main reason
for having run_forever().
--
___
Python tracker
<https://bugs.pyt
Yury Selivanov added the comment:
New changeset bfbf04ef18c93ca8cab0453f76aeea1d8fc23fb1 by Yury Selivanov in
branch 'master':
bpo-32296: Unbreak tests on Windows (#4850)
https://github.com/python/cpython/commit/bfbf04ef18c93ca8cab0453f76aeea
Yury Selivanov added the comment:
Matthias,
Thanks a lot for such a detailed write-up! I now better understand how you
want to use the proposed API in Trio, and why the weakref approach isn't going
to work (at least not in a straightforward way).
>> 2. What if another fra
Yury Selivanov added the comment:
New changeset 176baa326be4ec2dc70ca0c054b7e2ab7ca6a9cf by Yury Selivanov (Jelle
Zijlstra) in branch 'master':
bpo-30241: implement contextlib.AbstractAsyncContextManager (#1412)
https://github.com/python/cpython/commit/176baa326be4ec2dc70ca0c054b7e2
Change by Yury Selivanov :
--
resolution: -> fixed
stage: -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.org/issue30241>
___
___
Change by Yury Selivanov :
--
nosy: yselivanov
priority: normal
severity: normal
status: open
title: Remove "globals()" call from "socket.accept()"
___
Python tracker
<https://bug
New submission from Yury Selivanov :
socket.accept currently has this code:
type = self.type & ~globals().get("SOCK_NONBLOCK", 0)
which I believe is (a) bad Python style; (b) slows things down.
--
components: +Library (Lib)
nosy: +asvetlov, vstinner
type: -> enh
Change by Yury Selivanov :
--
keywords: +patch
pull_requests: +4746
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32318>
___
___
Py
Yury Selivanov added the comment:
> So first... that's not how nursery.start_soon works :-). It actually takes an
> async function, not a coroutine object. [..]
Interesting, and I think I like it. I definitely understand the motivation to
not tell users the difference betwee
Yury Selivanov added the comment:
New changeset 02a0a19206da6902c3855a1fa09e60b208474cfa by Yury Selivanov in
branch 'master':
bpo-32314: Implement asyncio.run() (#4852)
https://github.com/python/cpython/commit/02a0a19206da6902c3855a1fa09e60
Change by Yury Selivanov :
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.or
Yury Selivanov added the comment:
> Wow, I really love your new function. Thanks Yury! asyncio examples in the
> doc look much better now!
This year I'll stay for the sprints at PyCon, and will focus on improving the
docs further. I needed asyncio.r
New submission from Yury Selivanov :
There's a lengthy discussion of this proposal over here:
https://github.com/python/asyncio/issues/477
Guido there likes the idea of adding asyncio.create_task(), so let's add it.
--
___
Python track
Change by Yury Selivanov :
--
nosy: +asvetlov
___
Python tracker
<https://bugs.python.org/issue32327>
___
___
Python-bugs-list mailing list
Unsubscribe:
New submission from Yury Selivanov :
There's a handful of event loop methods that are currently documented as
coroutines, but in reality they return asyncio.Future:
* loop.sock_sendall,
* loop.sock_accept,
* loop.sock_recv, and
* loop.run_in_executor.
These methods need to become p
Change by Yury Selivanov :
--
keywords: +patch
pull_requests: +4765
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32327>
___
___
Py
Yury Selivanov added the comment:
New changeset 19a44f63c738388ef3c8515348b4ffc061dfd627 by Yury Selivanov in
branch 'master':
bpo-32327: Convert asyncio functions documented as coroutines to coroutines.
(#4872)
https://github.com/python/cpyt
Change by Yury Selivanov :
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.or
New submission from Yury Selivanov :
On Linux, socket type is both a socket type and a bit mask (of SOCK_CLOEXEC and
SOCK_NONBLOCK). Therefore, anyone who write code like 'if sock.type ==
SOCK_STREAM' writes non-portable code, that occasionally breaks on Linux.
This caused so
Change by Yury Selivanov :
--
keywords: +patch
pull_requests: +4771
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32331>
___
___
Py
New submission from Yury Selivanov :
I propose to add type slots for magic methods introduced by PEP 560. In the
brief discussion on the mailing list Guido OK'd the idea:
https://mail.python.org/pipermail/python-dev/2017-December/151262.html
I'll submit a PR that implements this p
Change by Yury Selivanov :
--
keywords: +patch
pull_requests: +4772
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32332>
___
___
Py
Yury Selivanov added the comment:
Thank you, Andrew!
--
___
Python tracker
<https://bugs.python.org/issue32311>
___
___
Python-bugs-list mailing list
Unsub
Yury Selivanov added the comment:
Serhiy, I've tested METH_STATIC approach and it works just fine. Thanks for
pointing that out!
I agree that adding slots for something that's already possible is an overkill,
so let's close this issue.
--
resolution: -> rej
Yury Selivanov added the comment:
> You can easily work around your problem by replacing "sock.type ==
> SOCK_STREAM" with "sock.type & SOCK_STREAM == SOCK_STREAM".
Heh :) No, that would be a buggy code. Try this on your Linux box:
(socket.SOCK_
Yury Selivanov added the comment:
On GH (https://github.com/python/cpython/pull/4877), Antoine wrote:
> I agree with Victor:
> it can't go into 3.6
> making the change in 3.7 is contentious
> Can there be an other way to solve the issue? Can we for example keep
> socket.t
Change by Yury Selivanov :
--
title: apply SOCK_TYPE_MASK to socket.type on Linux -> Add socket.truetype
property
___
Python tracker
<https://bugs.python.org/issu
Yury Selivanov added the comment:
Update:
I've rewritten the PR from scratch.
1. SOCK_TYPE_MASK isn't exported on Linux. Therefore we will not export
socket.SOCK_TYPE_MASK too.
2. I've added the new socket.truetype property.
3. When a socket is created without an FD, e.g.
Yury Selivanov added the comment:
Let's quickly iterate over what's possible first:
* It's possible to check the type of the FD using "getsockopt(SOL_SOCKET,
SO_TYPE)" on all platforms.
* It's possible to check family/proto of the FD using "getsockopt(SOL_
Yury Selivanov added the comment:
New changeset e796b2fe26f220107ac50667de6cc86c82b465e3 by Yury Selivanov in
branch 'master':
bpo-27456: Ensure TCP_NODELAY is set on linux (#4231)
https://github.com/python/cpython/commit/e796b2fe26f220107ac50667de6cc8
Change by Yury Selivanov :
--
pull_requests: +4791
___
Python tracker
<https://bugs.python.org/issue27456>
___
___
Python-bugs-list mailing list
Unsubscribe:
Change by Yury Selivanov :
--
pull_requests: +4792
___
Python tracker
<https://bugs.python.org/issue27456>
___
___
Python-bugs-list mailing list
Unsubscribe:
Yury Selivanov added the comment:
Another name option: socket.realtype
--
___
Python tracker
<https://bugs.python.org/issue32331>
___
___
Python-bugs-list mailin
Yury Selivanov added the comment:
New changeset 572636d42566da8eb6e85d5b8164e9ed8860a255 by Yury Selivanov in
branch '3.6':
bpo-27456: Ensure TCP_NODELAY is set on linux (#4231) (#4898)
https://github.com/python/cpython/commit/572636d42566da8eb6e85d5b8164e9
Change by Yury Selivanov :
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.or
Yury Selivanov added the comment:
I've been thinking a lot about this problem, and I'm really tempted to fix
sock.type property:
1. The problem first appeared in Python 3.2.
2. Python 2.7 doesn't have this problem at all, and doesn't even export
socket.SOCK_NONBLOCK. I
Yury Selivanov added the comment:
There's also a precedent of us breaking socket API. In Python < 3.6:
sock = socket.socket()
fd = sock.fileno()
os.close(fd)
sock.close()
Everything works fine. In 3.6+:
sock.close() # Will raise OSError
uvloop broke on this during
Change by Yury Selivanov :
--
title: Add socket.truetype property -> Fix socket.type on Linux
___
Python tracker
<https://bugs.python.org/issue32331>
___
___
Py
Yury Selivanov added the comment:
Nathaniel, thanks a lot for the comprehensive analysis.
It's now obvious that this weird Linux-specific socket.type quirk is of our own
making and specific only to Python.
I've updated the PR:
1. *type* argument of 'socket.socket()' i
New submission from Yury Selivanov :
Key observation: 99.9% of the time Futures and Tasks have only one callback.
Currently we have a list of callbacks in each Future/Task. We can avoid list
object allocation if we add a field to store the first callback. This way
we'll only need to
Change by Yury Selivanov :
--
keywords: +patch
pull_requests: +4801
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32348>
___
___
Py
Yury Selivanov added the comment:
New changeset 1b7c11ff0ee3efafbf5b38c3c6f37de5d63efb81 by Yury Selivanov in
branch 'master':
bpo-32348: Optimize asyncio.Future schedule/add/remove callback. (#4907)
https://github.com/python/cpython/commit/1b7c11ff0ee3efafbf5b38c3c6f37d
Yury Selivanov added the comment:
The attached benchmark from https://bugs.python.org/issue32204 shows that the
updated Task is 15% faster.
--
Added file: https://bugs.python.org/file47335/t.py
___
Python tracker
<https://bugs.python.
Change by Yury Selivanov :
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.or
Yury Selivanov added the comment:
Liran, Task (C implementation) was optimized in issue 32348; your benchmark
now runs 15% faster.
--
___
Python tracker
<https://bugs.python.org/issue32
New submission from Yury Selivanov :
asyncio.gather can be made faster if:
1. we don't use functools.partial
2. create less intermittent collections
3. drop unnecessary code (e.g. gather has some code that's duplicated in
ensure_future that it uses etc)
The proposed PR makes asyncio
Change by Yury Selivanov :
--
keywords: +patch
pull_requests: +4807
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32355>
___
___
Py
Change by Yury Selivanov :
--
nosy: +asvetlov
___
Python tracker
<https://bugs.python.org/issue32356>
___
___
Python-bugs-list mailing list
Unsubscribe:
New submission from Yury Selivanov :
As briefly discussed on https://github.com/python/asyncio/issues/488 and
https://github.com/python/cpython/pull/528
--
messages: 308509
nosy: yselivanov
priority: normal
severity: normal
status: open
title: asyncio: Make transport.pause_reading
Change by Yury Selivanov :
--
keywords: +patch
pull_requests: +4808
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32356>
___
___
Py
Yury Selivanov added the comment:
New changeset 902ab80b590e474bb2077b1fae8aac497b856d66 by Yury Selivanov
(Nathaniel J. Smith) in branch 'master':
bpo-30050: Allow disabling full buffer warnings in signal.set_wakeup_fd (#4792)
https://github.com/python/cpyt
Yury Selivanov added the comment:
I think that what Nathaniel proposes is very reasonable. The PR is LGTM and
I've just merged it. Closing the issue.
--
components: +Interpreter Core
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
type:
Yury Selivanov added the comment:
NP. I have another PR in the pipeline:
https://github.com/python/cpython/pull/4913
Both optimizations make your benchmark run 30% faster on 3.7. If you compile
asyncio.gather() with Cython you will get it another 5-15% faster. If you use
uvloop - another
New submission from Yury Selivanov :
asyncio.Task now uses asyncio.iscoroutine() to give a comprehensible error if a
user creates a Task for a non-awaitable type.
The problem is that iscoroutine() is quite expensive for non-native coroutines
(like the ones compiled with Cython), as it uses
Change by Yury Selivanov :
--
title: Optimize asyncio.iscoroutine() for non-native coroutines -> Optimize
asyncio.iscoroutine() and loop.create_task() for non-native coroutines
___
Python tracker
<https://bugs.python.org/issu
Change by Yury Selivanov :
--
keywords: +patch
pull_requests: +4809
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32357>
___
___
Py
Change by Yury Selivanov :
Added file: https://bugs.python.org/file47337/bench.py
___
Python tracker
<https://bugs.python.org/issue32357>
___
___
Python-bugs-list mailin
Yury Selivanov added the comment:
+1, let's do it. I'd add a new module - 'asyncio.abc' and put them both there
(as well as asyncio.isfuture() function).
--
___
Python tracker
<https://bug
Yury Selivanov added the comment:
+1.
--
___
Python tracker
<https://bugs.python.org/issue32363>
___
___
Python-bugs-list mailing list
Unsubscribe:
Yury Selivanov added the comment:
New changeset d757aaf9dd767d13205bf9917e520ebf43e7f6e5 by Yury Selivanov in
branch 'master':
bpo-32356: idempotent pause_/resume_reading; new is_reading method. (#4914)
https://github.com/python/cpython/commit/d757aaf9dd767d13205bf9917e520e
Change by Yury Selivanov :
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.or
Yury Selivanov added the comment:
> This issue is not specific to Linux. FreeBSD is also affected:
So instead of '#ifdef __linux__' we should use `#ifdef SOCK_NONBLOCK` etc.
I'll update the PR.
--
___
Python tracker
<ht
Yury Selivanov added the comment:
> But I would prefer to avoid a syscall just to clear flags :-/
Yes, that's what I want to do.
We control what constants the socket module exports. Let's just clear them if
we export them -- that's a good working solution (and also
Yury Selivanov added the comment:
New changeset 9818142b1bd20243733a953fb8aa2c7be314c47c by Yury Selivanov in
branch 'master':
bpo-32331: Fix socket.type when SOCK_NONBLOCK is available (#4877)
https://github.com/python/cpython/commit/9818142b1bd20243733a953fb8aa2c
Yury Selivanov added the comment:
I've merged the PR.
Summary of the final change:
1. socket.socket(family, type, proto) constructor clears SOCK_NONBLOCK and
SOCK_CLOEXEC from 'type' before assigning it to 'sock.type'.
2. socket.socket(family, SOCK_STREAM | SOCK_NO
Change by Yury Selivanov :
--
title: Fix socket.type on Linux -> Fix socket.type on OSes with SOCK_NONBLOCK
___
Python tracker
<https://bugs.python.org/issu
Yury Selivanov added the comment:
> In this case, bpo-27456 should be reopened or a new issue should be opened to
> fix asyncio in Python 3.6, no?
I'll make a PR to fix it in 3.7. Code in 3.6 is good.
--
___
Python trac
Change by Yury Selivanov :
--
pull_requests: +4817
___
Python tracker
<https://bugs.python.org/issue27456>
___
___
Python-bugs-list mailing list
Unsubscribe:
Yury Selivanov added the comment:
Actually, I don't think we need a deprecation period. Both methods are
completely unusable now.
Given:
async def foo():
await asyncio.sleep(1)
print('A')
return 42
1. Let's try to use Task.set_result():
asy
Yury Selivanov added the comment:
About the only use case I can come up with is subclassing asyncio.Task and
overriding set_result and set_exception implementations. This use case has
been broken (unintentionally) in Python 3.6, when we first implemented Task in
_asynciomodule.c. C Task
Change by Yury Selivanov :
--
keywords: +patch
pull_requests: +4818
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32363>
___
___
Py
Yury Selivanov added the comment:
New changeset a7bd64c0c01379e9b82e86ad41a301329a0775d9 by Yury Selivanov in
branch 'master':
bpo-27456: Simplify sock type checks (#4922)
https://github.com/python/cpython/commit/a7bd64c0c01379e9b82e86ad41a301
Change by Yury Selivanov :
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
type: -> performance
___
Python tracker
<https://bugs.python
Yury Selivanov added the comment:
New changeset a9d7e552c72b6e9515e76a1dd4b247da86da23de by Yury Selivanov in
branch 'master':
bpo-32357: Optimize asyncio.iscoroutine() for non-native coroutines (#4915)
https://github.com/python/cpython/commit/a9d7e552c72b6e9515e76a1dd4b247
Yury Selivanov added the comment:
New changeset 36c2c044782997520df7fc5604742a615ccf6b17 by Yury Selivanov in
branch 'master':
bpo-32355: Optimize asyncio.gather() (#4913)
https://github.com/python/cpython/commit/36c2c044782997520df7fc5604742a
Change by Yury Selivanov :
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.or
Yury Selivanov added the comment:
New changeset 319c0345cdd8fddb49d235462e71883f1dd51b99 by Yury Selivanov
(Julien Duponchelle) in branch 'master':
bpo-29711: Fix stop_serving in proactor loop kill all listening servers (#431)
https://github.com/python/cpyt
Change by Yury Selivanov :
--
resolution: -> fixed
stage: -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.org/issue29711>
___
___
New submission from Yury Selivanov :
Currently we have the following methods:
* socket.settimeout(t) -- can set the socket in blocking mode, when t==0.
* socket.setblocking(flag) -- sets in blocking or non-blocking mode.
* socket.gettimeout() -- returns 0 when socket is in non-blocking mode
Change by Yury Selivanov :
--
nosy: +asvetlov, vstinner
___
Python tracker
<https://bugs.python.org/issue32373>
___
___
Python-bugs-list mailing list
Unsubscribe:
Change by Yury Selivanov :
--
keywords: +patch
pull_requests: +4820
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32373>
___
___
Py
Yury Selivanov added the comment:
It looks like we have a bug with 'sock.settimeout()' and non-blocking/blocking
modes (or maybe this is a feature?)
Currently:
* to make a socket non-blocking, we call 'sock.settimeout(0)'.
* to make a socket blocking, we call 'sock.
Yury Selivanov added the comment:
It appears the the timeouts situation is a bit more complex. It's perfectly
normal for a Python socket object to be in a "blocking" mode and for its FD to
be in a non-blocking mode. Read more about this in the latest docs update to
Yury Selivanov added the comment:
Unfortunately they will not be backported, that's against our release policies,
and I can't do anything about it. You can backport them yourself and build
your own CPython 3.6. That's what bigger users (e.g. Facebook/Google) of
Py
901 - 1000 of 3129 matches
Mail list logo
