Timm Wagener added the comment:
Proposed changes are done and test coverage is extended. I had to change one
previous test slightly *(test_one_cancellation())*, which seems in line though,
with the proposed behavior.
--
___
Python tracker
<ht
Timm Wagener added the comment:
Ok, seems reasonable as well. I'll change it as proposed.
--
___
Python tracker
<https://bugs.python.org/issue40894>
___
___
Timm Wagener added the comment:
TLDR;
-
The intention of the PR is to make a future from gather return that cancelled()
is True if, and only if, cancel() has successfully been called on it (explicit
user intent) and it was awaited/has finished. All other finishing is not
considered
Timm Wagener added the comment:
Hello Kyle, thanks for reviewing.
> I'm starting to agree that it makes sense to override the behavior for
> `cancelled()`. However, it might make more sense to replace the
> `self._cancel_requested` check with `isinstance(
Change by Timm Wagener :
--
keywords: +patch
pull_requests: +19898
stage: -> patch review
pull_request: https://github.com/python/cpython/pull/20686
___
Python tracker
<https://bugs.python.org/issu
New submission from Timm Wagener :
It seems like the future subclass returned by asyncio.gather()
(_GatheringFuture) can never return True for future.cancelled() even after it's
cancel() has been invoked successfully (returning True) and an await on it
actually raised a CancelledError.
timm added the comment:
I would find it useful to have the exception classes listed in the Python
documentation rather than having to refer to two documents, but I don't have
strong feelings on this.
Given that nobody has fixed this for 4 years, should we just close the ticket?
I'
