[issue41273] asyncio: proactor read transport: use recv_into instead of recv

STINNER Victor added the comment: Tony: "asyncio: proactor read transport: use recv_into instead of recv" seems to be implemented, but PR 21446 "bpo-41279: Add StreamReaderBufferedProtocol" is a new feature. I suggest to open a new issue for this feature, and close this one. --

[issue41273] asyncio: proactor read transport: use recv_into instead of recv

David Bolen added the comment: It looks like there was an underlying asyncio.recv_into bug that was the likely root issue here. It's recently been fixed in bpo-41467, and test_asyncio is passing on at least the Win10 buildbot. -- ___ Python

[issue41273] asyncio: proactor read transport: use recv_into instead of recv

Tony added the comment: If the error is not resolved yet, I would prefer if we revert this change then. The new PR is kinda big I don't know when it will be merged. -- ___ Python tracker

[issue41273] asyncio: proactor read transport: use recv_into instead of recv

David Bolen added the comment: First, I also no longer see the error with a local PR 21446 build on the Win10 buildbot. As for timing, I believe policy is to revert, but in my view it can probably depend on how short "a bit" is for the fix. We've been in this state for 4-5 days already,

[issue41273] asyncio: proactor read transport: use recv_into instead of recv

Tony added the comment: Ok so I checked and the PR I am currently having a CR on fixes this issue: https://github.com/python/cpython/pull/21446 Do you want me to make a different PR tomorrow that fixes this specific issue to get it faster to master or is it ok to wait a bit? --

[issue41273] asyncio: proactor read transport: use recv_into instead of recv

Tony added the comment: I see, I'll start working on a fix soon -- ___ Python tracker ___ ___ Python-bugs-list mailing list

[issue41273] asyncio: proactor read transport: use recv_into instead of recv

David Bolen added the comment: I can at least confirm that this commit is the source of the issue on the Windows 10 buildbot. Interactively, it fails every time with it in place (unlike the buildbot which has had the occasional success) and passes reliably with it reverted. The error

[issue41273] asyncio: proactor read transport: use recv_into instead of recv

Stefan Behnel added the comment: There have been sporadic buildbot failures in "test_asyncio" since this change, message being "1 test altered the execution environment", e.g. https://buildbot.python.org/all/#/builders/129/builds/1443 Could someone please check if they are related?

[issue41273] asyncio: proactor read transport: use recv_into instead of recv

Yury Selivanov added the comment: New changeset 568fb0ff4aa641329261cdde20795b0aa9278175 by Tony Solomonik in branch 'master': bpo-41273: asyncio's proactor read transport's better performance by using recv_into instead of recv (#21442)

[issue41273] asyncio: proactor read transport: use recv_into instead of recv

Change by Tony : -- pull_requests: +20593 pull_request: https://github.com/python/cpython/pull/21446 ___ Python tracker ___ ___

[issue41273] asyncio: proactor read transport: use recv_into instead of recv

Change by Tony : -- pull_requests: +20589 pull_request: https://github.com/python/cpython/pull/21442 ___ Python tracker ___ ___

[issue41273] asyncio: proactor read transport: use recv_into instead of recv

Change by Tony : -- keywords: +patch pull_requests: +20588 stage: -> patch review pull_request: https://github.com/python/cpython/pull/21439 ___ Python tracker ___

[issue41273] asyncio: proactor read transport: use recv_into instead of recv

New submission from Tony : Using recv_into instead of recv in the transport _loop_reading will speed up the process. >From what I checked it's about 120% performance increase. This is only because there should not be a new buffer allocated each time we call recv, it's really wasteful.