Suggestion: Changes to release process, release from branch

[DJ Delorie]

Siddhesh Poyarekar <siddhesh@gotplt.org> writes:
> are any of these issues going to be serious enough to be problematic?

Probably not, which is why I don't feel strongly about them.  I'm just
bringing them up to make sure they're considered.

> There is one caveat when comparing with gcc, binutils, etc. though, in
> that the release branches get actively maintained, which isn't
> necessarily the case for glibc.

glibc release branches are maintained too... but not as much attention
is given to them as the master branch, which is true to some extent in
all projects.  And we don't re-release from a branch, which makes it
more important to get it right the first time.  If we never produce a
2.44.1 tarball, maintenance of the branch is not going to be picked up
be all who should.

> The branch-and-continue-keeping-master-open approach keeps a decent
> momentum going IMO.

Perhaps, but that's the momentum that causes problems - those developers
should be focused on testing the release.  I fear that the "branch, wait
two weeks, release" will just mean 2 weeks of nothing happening, because
everyone has shifted focus to new work for master.