On 1/14/20 12:01 PM, Frédéric Chapoton wrote:
>
> [A] It seems to me that the conclusion of the vote is rather clear, and
> that we should not feel obliged to make 9.1 compilable with python2.
>
Since I'm not on that list yet, one thing is for sure: there are going
to be two consecutive versions of sage, one where python-2.x works, and
one where it doesn't. Users won't care too much which versions those are
unless there's a feature or bugfix that they need in the new version --
they can just use the old one.
Most projects solve this with multiple release branches. When you break
backwards-compatibility in a major way, you keep the previous version
(before the break) in a separate branch, and backport critical fixes to
it for a limited time. It's not how the sage release process works, but
I don't see why we couldn't do it this once, to keep everyone happy:
1. Branch off the 9.0 release
2. Drop compatibility in 9.1
3. Proceed as normal
4. If there are any major bugs found in 9.1, backport the fix to
the 9.0 branch, and release a 9.0.1 that supports python-2.x.
Mathematical bug fixes that merge clean and pass the tests could also
make the cut, but ideally step (4) would require little work. This would
buy users a few more months of python2 compatibility, and lets the rest
of us move on.
The main downside is that this relies solely on the release manager to
merge bugfix tickets twice, into both release branches. So if he thinks
this is a waste of time (understandable), then it's a non-starter.
--
You received this message because you are subscribed to the Google Groups
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to sage-devel+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/sage-devel/ab573d83-ab79-4bdc-1e23-36f411e26a44%40orlitzky.com.
