Closed ubaumann closed 1 year ago
I am completely fine removing 3.7 support as AFAIK it is EOL. @ktbyers thoughts?
When I remove the list of outdated looks like this:
urs@DESKTOP-BM35OBL:~/projects/nornir$ poetry show --outdated
docutils 0.18.1 0.20.1 Docutils -- Python Documentation Utilities
importlib-metadata 4.13.0 6.8.0 Read metadata from Python packages
ipython 8.12.2 8.14.0 IPython: Productive Interactive Computing
sphinx 6.2.1 7.2.2 Python documentation generator
sphinxcontrib-applehelp 1.0.4 1.0.7 sphinxcontrib-applehelp is a Sphinx extension which outputs Apple help books
sphinxcontrib-devhelp 1.0.2 1.0.5 sphinxcontrib-devhelp is a sphinx extension which outputs Devhelp document.
sphinxcontrib-htmlhelp 2.0.1 2.0.4 sphinxcontrib-htmlhelp is a sphinx extension which renders HTML help files
sphinxcontrib-qthelp 1.0.3 1.0.6 sphinxcontrib-qthelp is a sphinx extension which outputs QtHelp document.
sphinxcontrib-serializinghtml 1.1.5 1.1.8 sphinxcontrib-serializinghtml is a sphinx extension which outputs "serialized" HTML files (json and pickle).
ipython
and sphinx
also dropped Python 3.8 support. In my opinion, dropping Python 3.7 support should be okay. On the other side, the dependencies are only development and documentation dependencies.
Yep sounds good on removing Python3.7 (since it is EOL).
I'd be fine dropping 3.8 too but I don't have strong opinions here.
Python 3.8 will be EOL on October 24. We could add to each development dependency a restriction to only install it when a newer version is unused but feels a little hacky.
If someone would want to install the development and documentation dependencies on Python 3.8 it wouldn't install them without an error.
I'd avoid hacks, specially if 3.8 will be EOL in October. @ktbyers do you have anything against dropping 3.8 too or would you rather wait until October?
Thanks @ubaumann for the work :)
@dbarrosop It will be October next year, but I agree on avoiding hacky stuff.
I should have written October 2024
Ah, that's not as near :P. I'd still be fine dropping support for it but I guess enterprise users may still be using it. Let's see if Kirk has any opinions here but I am fine with the PR as it is.
Yeah, I think we should keep PY3.8.
Everything looks good on the PR.
I also updated main branch protections
to not require PY3.7 any longer (and added PY3.11 tests).
Because of Python 3.7 support, I could not update some packages (black, mypy, sphinx, ...) to the newest version.
mypy
after update:List of outdated packages on my system: