Advise version incompatibility with Python 3.9.7 #93
Loading…
x
Reference in New Issue
Block a user
No description provided.
Delete Branch "kuanhulio/patch-1"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Summary
Due to a bug in Python 3.9.7, code situations where we call an init function in a subclass will fail with a TypeError.
This is only a bug within the Python language and was patched out with Python 3.10 and this doesn't affect Python 3.9.6.
Here, we are advising the incompatibility.
You can view that bug report here on the Python Bugs site.
Checklist
can you try updating your python 3.9.7 version to see if it fixes the issue or did they fix it in 3.9.8?
Hey there,
3.9.8 is not released yet, but I have not tested 3.9.6 nor 3.10. I'll give these a deployment in a few hours.
Per my tests, 3.9.6 and 3.10.0 seem unaffected. I reinstalled Python 3.9.7 and the same edge case error happened.
I feel like this should be left as a note in the affected places, instead of as a big notice at the top of the readme, on that note, where are the affected places? I have been running 3.9.7 for development since it came out on
pacman
and seem to have no problems.Well, on my hands, I'm working on Windows 10 with the 3.9.7 directly from python.org and it seemingly throws this error.
Affected areas would primarily be wherever a class derives from a
Protocol
and we callsuper().__init__()
. I'm not really familiar with d.py codebase or the additions you guys have made, however, this specific error was called fromdiscord/ext/commands/converter.py:116
Here's the full traceback:
Now, again, I don't know where the specific spots are where this applies, all I know is this is a bug with Python 3.9.7 and from that, that's why I believe it's certainly easier to just block out 3.9.7 instead of making changes to the underlying codebase. This behavior was patched out on 3.10 and 3.9.6 is unaffected in my uses.
A warning at the top of the readme is not appropriate for this kind of bug.
Honestly I don't see a reason to add this warning anywhere, we can simply point people to the python bug on the off chance that they come across this error. However if the others agree that they'd wish to add a notice, it'll have to be in strictly the affected area of the docs, not the top of the readme
This, I can respect. However, I'm unaware if there's more of these cases here. I'm fine with redoing it to the docs, but the question becomes where more instances like this happen with the codebases.
Checkout
From your project repository, check out a new branch and test the changes.