On Mon, Dec 22, 2025 at 10:25:14PM -0500, Eli Schwartz wrote: > On 12/22/25 9:22 PM, Nuno Silva wrote: > > On 2025-12-23, Eli Schwartz wrote: > > > >> Do the nightlies also work with python 3.13 / 3.14? > > > > Not yet. There is work ongoing to update these python parts. I'm not > > currently aware of the details. Might be just that it's a longer task, > > or it might be that somebody is hitting a roadblock. I'll try to find > > out more. > > > > > > How urgent is the situation, will it not be possible to have 2.53.23 on > > the portage tree if it requires older versions of python3 to build? > > > Please see this and the followup: > > https://public-inbox.gentoo.org/gentoo-dev/[email protected]/T/#u > > tl;dr several key packages, upstream dropped python 3.11 support and > Gentoo must follow. The consequence is that many other packages which > either USE-depend or test-depend on them, have to drop support too. > > This is a juggling act which requires a lot of maintenance, and the > simplest solution is to fully disable 3.11 inside the eclasses (i.e. a > repository-level ban on requiring 3.11 inside another package). > > > We are not there yet but the handwriting is on the wall. If there's a > serious expectation that the status will change "soon", then I think we > can reconsider this last-rite. Most of the packages being last-rited are > either abandoned upstream or effectively abandoned in Gentoo, else they > would have been updated to support python 3.12 some time in the last 2.5 > years!!! And preferably 3.13 (some time in the last 1.5 years). > > If packages are realistically never going to be fixed, then proactively > removing them is prep work we need to do in order to finally drop 3.11 > targets in regular packages. As you've noticed, it also attracts > attention from users of the package :) and does so in advance of finally > pulling the plug on 3.11 entirely. > > ... > > Please consider the situation roughly as urgent as a "final call to > action". And check with upstream about the status -- Gentoo isn't the > only distro for which failure to support current python versions is > singlehandedly a steep challenge to packaging it at all.
fwiw it should be pretty easy to keep using 3.11 without blocking anything here, seamonkey *only* needs the interpreter and afaik we're not dropping that from the tree just because of sphinx or so (we still have 3.8, 3.9, 3.10, and even 2.7 still). If want to drop eclass support, it'd just need to be adjusted to find the right python without using python-any-r1, and then add it to BDEPEND. Long term that would still need proper fixing, but it'd be a way to gain time without being a burden on the python team. Not that I plan to work on or test seamonkey myself (never used it). -- ionen
signature.asc
Description: PGP signature
