Re: CONTRIBUTE and where to commit bug fixes

[Stefan Monnier <monnier@iro.umontreal.ca>]

> but that's obviously not always true: depending on the release cycle,
> only fixes for bugs that where introduced in the last release are
> accepted, or even less.

Right.  Looks like you've followed this list long enough to know how it
works ;-)

> And if a fix is not ok for the release branch in a moment, should it
> go to master, or should it just wait until later?

Install it on master.  If the patch should apply to the next bug-fix
release, but can't be installed because we're in "deep freeze", we
should (but don't AFAIK) have some other branch for that patch, indeed.

> How can I find out which fixes are acceptable for the release branch at
> any moment without asking every time?

Follow the list with enough dedication to know when we're in deep freeze
and when we're not.

Maybe rather than document the way we work in more detail (which won't
really help solve the problem, obviously), we could change the way we
work to make it more clear:

- when we start the "new major release" process, we create a new branch
  (e.g. emacs-27).  Nothing new here.
- when we enter the "deep freeze" process, rather than announce it and
  hope people will know about it, we create yet another branch
  (e.g. emacs-27.1).

Then the rule can be documented more simply in the doc:
- master is for new features.
- emacs-NN is for any bug and doc fix.
- emacs-NN.MM only takes bug fixes for regressions and requires explicit
  agreement from the maintainers.

This also solves the problem of where to install non-regression bug
fixes when we're in deep freeze.


        Stefan