Leak in asyncio.selector_events._SelectorSocketTransport

[begnac]

BPO	43579
Nosy	@asvetlov, @1st1, @begnac

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = None
created_at = <Date 2021-03-21.07:11:10.990>
labels = ['expert-asyncio', '3.9', '3.10', 'performance']
title = 'Leak in asyncio.selector_events._SelectorSocketTransport'
updated_at = <Date 2021-03-21.07:11:10.990>
user = 'https://github.com/begnac'

bugs.python.org fields:

activity = <Date 2021-03-21.07:11:10.990>
actor = 'begnac'
assignee = 'none'
closed = False
closed_date = None
closer = None
components = ['asyncio']
creation = <Date 2021-03-21.07:11:10.990>
creator = 'begnac'
dependencies = []
files = []
hgrepos = []
issue_num = 43579
keywords = []
message_count = 1.0
messages = ['389219']
nosy_count = 3.0
nosy_names = ['asvetlov', 'yselivanov', 'begnac']
pr_nums = []
priority = 'normal'
resolution = None
stage = None
status = 'open'
superseder = None
type = 'performance'
url = 'https://bugs.python.org/issue43579'
versions = ['Python 3.9', 'Python 3.10']

[begnac]

Hello,

Even after close()ing, asyncio.selector_events._SelectorSocketTransport keeps a reference to itself via self._read_ready_cb.

should probably add

def close(self):
    super().close()
    self._read_ready_cb = None

Cheers !
Itaï.

[kumaraditya303]

The gc would break the cycle anyways, is this an issue in practice? @begnac

[begnac]

In my experience, the gc doesn't always break all the cycles.
Personally I like to check for cycles in my code and clean them up, just in case. Sometimes I stumble over cycles in code I use. Specifically about this one, it's been such a long time I do not remember... but if I found it, it's because something didn't get cleaned up :)

[gvanrossum]

I think it's a reasonable suggestion and the fix should be simple.

[kumaraditya303]

Duplicate of #100348