-
-
Notifications
You must be signed in to change notification settings - Fork 31.6k
Deprecate-remove passing coroutine objects to asyncio.wait() #78971
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
asyncio.wait() accepts coroutines, wraps them into Tasks, and later returns those implicitly created Tasks in (done, pending) sets. This is very confusing to new asyncio users and it's almost impossible to figure out what is going on. See the first PR to the docs for more info. Andrew, I think we should deprecate passing coroutines to asyncio.wait() in 3.8, and disallow that in 4.0. |
4.0 is too far. Why not disallow them in 3.10? |
What's the current plan? I thought it's going to be 3.8, 3.9, 4.0. Is there a PEP detailing this? |
Actually, since Andrew also agrees that we need to deprecate passing coroutines to wait(), I'll keep this issue open until we add an actual DeprecationWarning in 3.8. |
PendingDeprecationWarning |
AFAIK there is no a plan for 4.0. But seems many core developers (including me) think that changing the major number should reflect some major changes like removing GIL. And Guido preferred 3.10 after 3.9. |
Since 3.8 has been released and the deprecation notice is in the 3.8 whatsnew document, should we implement the warning in Lib/asyncio/tasks.py? If so, I can open a PR.
Also, it's not clear to me if this should be a DeprecationWarning or PendingDeprecationWarning. The most recent message from Yury in the issue suggests a PendingDeprecationWarning, but the actual entry in the documentation (https://docs.python.org/3/library/asyncio-task.html?highlight=asyncio%20wait#asyncio.wait) seems like it might imply that it would be a DeprecationWarning:
|
I think we can add Since it is a) just a warning b) already marked as deprecated in docs -- the harm is minimal. Honestly, we just missed the issue when were prepared for 3.8 |
If we add the deprecation warning just for 3.9, would the removal release also be pushed forward?
Yeah that's definitely understandable, there were quite a number of major changes made in 3.8. It was the first time I saw a deprecation clearly documented in a final release that didn't have an associated deprecation warning. As for 3.8, I would like to update the whatsnew entry to add a link to this bpo issue if that's okay, as it does not at the moment. I recently included the entry in a PR that added entries for multiple significant changes for asyncio in the 3.8, but I forgot to include the bpo link (mostly because I found it in the documentation, rather than from the bpo issue). |
I want to clarify that this may be a more common occurrence than I'm realizing, I'm not entirely certain. Also it's not intended at all as a criticism, it just surprised me since it was the first time I saw it happen. |
Yes, deprecating in 3.9 with removal in 3.11 is fine. Regarding 3.8 release notes update -- not sure if it is needed flr docs-only change. In the current situation, we have so-called *soft deprecation*: bare coroutines are deprecated in docs without any code change. This is perfectly fine, we give our users extra time to prepare for changes. |
I'm not certain if the entry is necessary; my main concern is just that it's already present in the 3.8 release notes/whatsnew without anywhere to look for further information.
So is something considered to be a "soft deprecation" if there is no code change? If so, I may have made a mistake by specifying a release in the entry I recently added for it (https://docs.python.org/3/whatsnew/3.8.html#deprecated): "The explicit passing of coroutine objects to asyncio.wait() has been deprecated and will be removed in version 3.10." I would propose to adjust this to: "The explicit passing of coroutine objects to asyncio.wait() has been deprecated. (Contributed by Yury Selivanov in :issue:`34790`.)" Then for 3.9 whatsnew, we could specify the release: "The explicit passing of coroutine objects to asyncio.wait() has been deprecated and will be removed in 3.11." Thoughts? |
I'd add |
lgtm |
Sounds good, I'll work on opening a PR. |
#61179 Is a simple fix for the asyncio.wait() whatsnew entry for 3.8. I'll implement the deprecation warning and add a 3.9 whatsnew entry in a separate PR, since those changes won't be backported. |
#61181 Implements the deprecation warning, adds tests, and adds the 3.9 whatsnew entry. Once this PR is finalized, I think this issue can be closed. |
Is this whatsnew/3.8.rst entry correct and complete? ''' |
It looks to be fully consistent with the asyncio docs: https://docs.python.org/3/library/asyncio-task.html#generator-based-coroutines. From the blame of the documentation, it seems like Yury added the deprecation notice (GH-9314) to that section with the listed release for removal as 4.0, and then changed it to 3.10 (GH-9579). At a later date, Andrew implemented the actual deprecation (GH-13346). So as far as I can tell, yes. |
asyncio.wait
's documentation #20008Note: 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:
bugs.python.org fields:
The text was updated successfully, but these errors were encountered: