-
-
Notifications
You must be signed in to change notification settings - Fork 153
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
Remove PySide2 from APIs on Python 3.11+ #400
Conversation
Hi @StSav012 thank you checking this! Although I'm a little bit confused, it is possible to install PySide2 with pip on Python 3.11? I think you can't since there are no compatible wheels available, right? Also, seems like you can install PySide2 5.15.8 with conda (conda-forge) and after testing locally seems like it works with Python 3.11 🤔 Also, just in case, what do you think about this @CAM-Gerlach @ccordoba12 ? |
I'm actually surprised. First, in the tests, there is Python 3.10.8 (see conda info) despite the tests are titled Python 3.11. Then, by the fact that the developers are unaware of As a side note, how old Python, PySide2, and PyQt5 do you support? In the code, I see some lines about PyQt5 5.9 and 5.12, and I can PR patches for PyQt5 5.6 if someone needs them. Edit. I can use PySide2 on Python 3.11 almost right, except for a crash when an app quits. So, for the end user, there is a mere difference. |
Checking the CI, we install the respective Python version for the test at the
PySide2 5.12+ and PyQt5 5.9+. For the validations we have defined those minimum version here: Lines 146 to 149 in aa3fe4b
Oh then I think we should create an issue in the PySide2 feedstock repo: https://github.com/conda-forge/pyside2-feedstock or maybe the issue is related with something we are doing over the QtPy side? |
If PySide2 is working fine with Python 3.11 with Conda-forge packages, then we certainly don't need to do this.
That's no surprise for us because there's an important disconnect between developers outside of the Scientific Python ecosystem and those inside it. |
Right now, we actually aren't testing PySide2 on Python 3.11 on Conda, as I hadn't gotten it to work in PR #392 , but I tested it just now and it works just fine, so I suspect that was a result of #397 . I've opened PR #401 to unskip it. In addition, if we did want to do it, this wouldn't be the way to go about it. This PR removes it from the lookup table of API names, which ensures that a somewhat ambiguous error message will be immediately presented if the user manually specifies it with the
Yeah, there are no wheels past 3.10, and no sidsts at all, so the only way a user would install it would be either through conda (which works) or through manually building their own source checkout (which they could patch to fix). Therefore, it's not clear to me there's any user benefit to restricting it.
Yup, that's right. |
Your replies make perfect sense. Thank you. I drop the PR. |
Thank you @StSav012 for raising up this possible source of issues and thinking about a possible solution! If you have further concerns in the future about possible things rasing issues let us know! |
And it was also thanks to this PR that we opened #401 to unskip PySide2 Py3.11 testing in CI on Conda, as it is now working but was previously skipped. |
As far as I can judge from the
PySide*
chat message by Florian Bruhin, PySide2 is not going to run on Python 3.11+. Currently, it crashes Python. To avoid further issues, I suggest removing it from the APIs for Python 3.11 and up.