[Numpy-discussion] Could we simplify backporting?

[Marten van Kerkwijk]

Hi Ralf,

Yes, good to think about other policies. For astropy, we do the
decision by labelling with the bug-fix branch (with a policy that it
really should fix a bug), and inserting text in that release's bug-fix
notes (we really should automate that part...). Then, backporting is
done shortly before the bug-fix release and, as far as I can tell (not
having done it myself), outside of github. In rare cases with
hard-to-resolve merge conflicts, the original PR author gets a note
asking for help.

As for a travis test: here I was mostly thinking of an allowed-to-fail
test that would at least alert one if backporting was going to be an
issue.  I think travis runs again once one merges, correct? If so, on
that merge it could, in principle, do the backport too (if given
enough permission, of course; I'm not sure at all I'd want that, just
pointing out the possibility! E.g., it might trigger on a message in
the merge commit.).

All the best,

Marten