On 9 August 2017 at 22:30, Philippe Mathieu-Daudé <f4...@amsat.org> wrote:
> I wonder if I'm understanding correctly what the MAINTAINERS file is for and
> how to use it.
> From an submitter view I feel a bit confused. I thought ./get_maintainer.pl
> would give me the list of person to email the changes I did on some file
> from the repository. This script seems correctly named, I'm looking for some
> ./get_reviewers.pl instead, to know who I'v to keep updated, apart from the
> currently we have:
> "M: Mail patches to: FullName <address@domain>"
> Does this imply FullName is a maintainer?
Does it matter? As a contributor, what you want to know is
"who cares enough about this file to be worth cc'ing, if anybody",
and get_maintainer.pl does that for you.
Hopefully at least one of those people is also in a position
to shepherd the patch through into git master, but that doesn't
affect what you need to do as a patch submitter.
If anybody cares enough about a particular area of the codebase
to want to be cc'd on all patches so they can review them, then
to my mind that makes them effectively a co-maintainer on those
files, and they can be listed in MAINTAINERS.
> If so is it ok I do this change:
> - M: Mail patches to: FullName <address@domain>
> + M: Maintainer: FullName <address@domain>
> + These maintainers must be mailed on patches.
> R: Designated reviewer: FullName <address@domain>
> These reviewers should be CCed on patches.
> actual default for un-matched: "recent contributors" + qemu-devel@
> $ ./scripts/get_maintainer.pl -f disas.c
> get_maintainer.pl: No maintainers found, printing recent contributors.
> get_maintainer.pl: Do not blindly cc: them on patches! Use common sense.
> Peter Maydell <peter.mayd...@linaro.org> (commit_signer:2/3=67%)
> Richard Henderson <r...@twiddle.net> (commit_signer:1/3=33%)
> Thomas Huth <th...@redhat.com> (commit_signer:1/3=33%)
> Michael Tokarev <m...@tls.msk.ru> (commit_signer:1/3=33%)
> Julian Brown <jul...@codesourcery.com> (commit_signer:1/3=33%)
> email@example.com (open list:All patches CC here)
> I find the un-matched "recent contributors" list often confuse, due to files
> being moved, header updated, checkpatch indented.
Yes, the recent-contributors list is often unhelpful, which
is partly why the script warns about them. We might perhaps
switch the default to --no-git-fallback".
> Anyway I now understand these recent contributors are not maintainers but
> no-designated reviewers, unwilling to be maintainers (else they'd have added
> a section/entry by themselves).
They're just people who touched a file more often, that's all;
the script does a little of the git log mining for you in
case it's useful (often it isn't).
The underlying problem, as Cornelia points out, is that some
parts of QEMU have no maintainer at all, or have a theoretical
maintainer who in practice doesn't these days have enough time
to do code reviews very promptly, or have a email listed that
actually bounces because they changed employer 3 years ago and
stopped working on QEMU, etc. Changing wording in the
MAINTAINERS file is not going to help with this :-(