With most of the new sticking points being debian 12 (EOL in June) or
FreeBSD (timeline uncertain for when they'll re-modernize their python
stack)
One edge case here, however, is Sphinx, which does not actually have a
modern package in centOS stream 9.
CentOS currently ships 3.4.3, but it would not be usable if using the
optional 3.12 upgrade. If we ignore this, both FreeBSD and Debian 12
have sphinx 5.3.0 to offer us, which is a reasonable version jump, but
it means that CentOS would not be able to build docs unless you
enabled `--online` for configuring QEMU or sideloaded Sphinx
otherwise.
CentOS 9 reaches end of life on May 31 next year. I do not think
keeping its ability to build bleeding edge docs offline out of the box
for its final year (give or take) is worth your time, or anybody's.
That's exactly the reason why we (John and I) introduced mkvenv.py,
except it was CentOS 8 and SLES 15 at the time
(https://www.qemu.org/2023/03/24/python/):
"Some of these tools are run through the python3 executable, while
others are invoked directly as sphinx-build or meson, and this can
create inconsistencies. For example, QEMU’s configure script checks for
a minimum version of Python and rejects too-old interpreters. However,
what would happen if code run by Sphinx used a different version?
This situation has been largely hypothetical until recently; QEMU’s
Python code is already tested with a wide range of versions of the
interpreter, and it would not be a huge issue if Sphinx used a different
version of Python as long as both of them were supported. This will
change in version 8.1 of QEMU, which will bump the minimum supported
version of Python from 3.6 to 3.8. While all the distros that QEMU
supports have a recent-enough interpreter, the default on RHEL8 and
SLES15 is still version 3.6, and that is what all binaries in /usr/bin
use unconditionally."
So yeah, there's precedent for doing this, and it's the right thing to
do anyway.
My plan:
- Wait and see what happens after the current QEMU release
- Gleefully drop Ubuntu 22.04 in April (Allows Python 3.11+)
- Gleefully drop Debian 12 in June
- Hold my breath and see if FreeBSD happens to modernize by this point
(They have until October 2027 to do so, so it's a dice roll if they
will or not)
- Do a big round of Python modernization all at once; either to 3.11
or 3.12 depending, and either to Sphinx 5.x or 7.x depending, pray for
forgiveness that I am removing CentOS stream 9's ability to build docs
offline so I don't have to wait another six months and then do another
big cleanup round
The main nice features of 3.12 are improved f-strings, which are nice
but not really a necessity.
If I read it correctly, https://www.freshports.org/lang/python says it
has 3.9 on PPC and 3.11 elsewhere, but
https://www.freshports.org/lang/python311 says 3.11 exists on PPC as well.
I'd go for 3.11 and Sphinx 7. For FreeBSD we can require Python from Ports.
Paolo
