So, it turns out that there is no tensorflow for Python 3.12, which is what is shipped out with F39.
I tried to downgrade my python using
sudo dnf downgrade python --releasever=38
that downgrades a huge bunch of things, including important components like rpm.
However, dnf install/update pretty much anything gives a segmentation fault after that.
$ sudo dnf update --exclude=python* Last metadata expiration check: 2:37:29 ago on Fri 17 Nov 2023 05:50:32 PM CST. Segmentation fault
Staying within the dnf system, is there a better way to keep the older version of python around?
Many thanks, Ranjan
On Fri, Nov 17, 2023 at 9:30 PM Ranjan Maitra via users users@lists.fedoraproject.org wrote:
So, it turns out that there is no tensorflow for Python 3.12, which is what is shipped out with F39.
I tried to downgrade my python using
sudo dnf downgrade python --releasever=38
that downgrades a huge bunch of things, including important components like rpm.
However, dnf install/update pretty much anything gives a segmentation fault after that.
$ sudo dnf update --exclude=python* Last metadata expiration check: 2:37:29 ago on Fri 17 Nov 2023 05:50:32 PM CST. Segmentation fault
Staying within the dnf system, is there a better way to keep the older version of python around?
The Python folks recommend building from sources, but using a different $prefix, and installing with `make altinstall`. See https://docs.python.org/3.9/using/unix.html#building-python.
Debian has a nice explanation of how to run different versions of Python. It follows Python's recommendations. Once you download the version you are interested in, upack it, run `./configure --prefix=/opt && make && make test`. Follow it with a `make altinstall`. Also see https://wiki.debian.org/Python.
If you try Python's method, do not install into /usr/local. Pick a location like /opt/python3.11. Unexpected versions of software in /usr/local can cause quite a bit of pain since /usr/local is often on-path.
Jeff
On 18/11/23 03:29, Ranjan Maitra via users wrote:
So, it turns out that there is no tensorflow for Python 3.12, which is what is shipped out with F39.
I tried to downgrade my python using
sudo dnf downgrade python --releasever=38
that downgrades a huge bunch of things, including important components like rpm.
However, dnf install/update pretty much anything gives a segmentation fault after that.
$ sudo dnf update --exclude=python* Last metadata expiration check: 2:37:29 ago on Fri 17 Nov 2023 05:50:32 PM CST. Segmentation fault
Staying within the dnf system, is there a better way to keep the older version of python around?
You should try something like Pyenv [1] or miniconda [2] to manage different Python versions instead of messing around with the system packages.
[1]: https://github.com/pyenv/pyenv
[2]: https://docs.conda.io/projects/miniconda/en/latest/
On Fri, Nov 17, 2023 at 10:29 PM Ranjan Maitra via users < users@lists.fedoraproject.org> wrote:
So, it turns out that there is no tensorflow for Python 3.12, which is what is shipped out with F39.
I tried to downgrade my python using
sudo dnf downgrade python --releasever=38
that downgrades a huge bunch of things, including important components like rpm.
Many system tools use python -- as you have discovered, downgrading it will cause problems.
However, dnf install/update pretty much anything gives a segmentation fault after that.
$ sudo dnf update --exclude=python* Last metadata expiration check: 2:37:29 ago on Fri 17 Nov 2023 05:50:32 PM CST. Segmentation fault
Staying within the dnf system, is there a better way to keep the older version of python around?
You broke dnf. Fix it and you can install the F39 python3.11 package. There are ways (environments, tweaks to the hashbang line in python scripts). With F38 I used 3rd party python tools that needed python-3.10. I haven't had time to see what is provided with F39, but you can look for conda or mamba packages.
Anaconda Python is a widely used platform independent system that supports use of multiple python versions via environments.
Many thanks, Ranjan -- _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
My thanks again for this!
On Sat Nov18'23 08:26:10AM, George N. White III wrote:
From: "George N. White III" gnwiii@gmail.com Date: Sat, 18 Nov 2023 08:26:10 -0400 To: Community support for Fedora users users@lists.fedoraproject.org Reply-To: Community support for Fedora users users@lists.fedoraproject.org Subject: Re: recommended practice for python 3.11 on F39
On Fri, Nov 17, 2023 at 10:29 PM Ranjan Maitra via users < users@lists.fedoraproject.org> wrote:
So, it turns out that there is no tensorflow for Python 3.12, which is what is shipped out with F39.
I tried to downgrade my python using
sudo dnf downgrade python --releasever=38
that downgrades a huge bunch of things, including important components like rpm.
Many system tools use python -- as you have discovered, downgrading it will cause problems.
However, dnf install/update pretty much anything gives a segmentation fault after that.
$ sudo dnf update --exclude=python* Last metadata expiration check: 2:37:29 ago on Fri 17 Nov 2023 05:50:32 PM CST. Segmentation fault
Staying within the dnf system, is there a better way to keep the older version of python around?
You broke dnf. Fix it and you can install the F39 python3.11 package. There are ways (environments, tweaks to the hashbang line in python scripts). With F38 I used 3rd party python tools that needed python-3.10. I haven't had time to see what is provided with F39, but you can look for conda or mamba packages.
I was not aware that it was possibe to install python3.11 using dnf in F39, so thanks for this. Wow, amazing! The old tensorflow installation just works out of the box. Did not expect that.
Thanks for this!
Anaconda Python is a widely used platform independent system that supports use of multiple python versions via environments.
Thanks also for all the other information, it is very helpful and I will keep it with me for later use.
Best wishes, Ranjan