This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: How to keep Reviewed-by lines in git commits with gerrit.
- From: Joseph Myers <joseph at codesourcery dot com>
- To: Florian Weimer <fw at deneb dot enyo dot de>
- Cc: Carlos O'Donell <carlos at redhat dot com>, libc-alpha <libc-alpha at sourceware dot org>, Florian Weimer <fweimer at redhat dot com>
- Date: Tue, 12 Nov 2019 22:47:28 +0000
- Subject: Re: How to keep Reviewed-by lines in git commits with gerrit.
- Ironport-sdr: p/Igs4W5j5p8dif1DiSWD/Bj2kKCKuv7bSixlVtFLBSJfr223gnJa4AHJgS9SlVd/zU498frd2 7lvRyRH5Y5hr28mQcU3jcBO1qsh3iCBp9q4QqHyS/N4ILOpHi3KczkenGrfdHz9HsSYhCd/3Og qrd+OApwJ5EbpXFPt0x+oGyU3ie/bxpfiTObKchQjRId8eq837kRc2olmfnXzk+ZSt1SvyA3gU CMhQlySSbJr6mCBDo7o/Azu/5KhC6J5iVaT2LnArATtOkL35SOsE5l7EHibtsQt2fQS/HrXJy1 XbM=
- Ironport-sdr: bNhPln24+y/NH5N8jYyg4+qlUWf97jLQi4r01dPgTvVu4xVJajqhExA6S9H8mB52D2gVUZ6aLZ 026n6JkQkbH+webcgXRasuRTYJGs1859GIx2IEJz4sKlfDDLuTpNeAPKMmxk6N9u72XcVS6BkA 1udc2R2Ar9GBb9JrXleQ67T7rADSoNkBpPIHIwMdDTEATZpaINk0cvdS2HPF0mfgt1JsdeLI3+ 1ZRjBfiKU4tcs/xZx4ySL5+1ZQcEnDUtd0RJ1GauahiaFDRCjg3+Zy47cEaVJOQomfH0QIFw18 r3M=
- References: <7b4aa5d2-14e7-c0aa-a258-bbd60455fae5@redhat.com> <alpine.DEB.2.21.1911121737580.10440@digraph.polyomino.org.uk> <28848c99-55e0-cbda-9230-3a0c3c8257ae@redhat.com> <alpine.DEB.2.21.1911122219580.22324@digraph.polyomino.org.uk> <87eeycbu46.fsf@mid.deneb.enyo.de>
On Tue, 12 Nov 2019, Florian Weimer wrote:
> > This shouldn't just be about Reviewed-by. If someone says a patch is OK
> > with a specific change made to it (whether to the code or to the commit
> > message), it should be enough to make that change and retest and push to
> > master, without needing to go through extra administration in a review
> > system just because of that change.
>
> If you want to use Gerrit as some sort of audit trail (and we will
> eventually face attacks on the GNU toolchain sources; maybe it has
> already happened and we just haven't noticed yet), then of course it's
> necessary that any tiny change gets re-reviewed.
We say maintainers can commit changes in their areas without needing
review, and that certain kinds of changes are considered obvious for
anyone with commit access. So I don't think "OK with change X" (without
needing re-review) is fundamentally different from that.
git provides the audit trail of exactly what changes went into the
repository. (And the glibc-cvs list shows which user account did the
push, which is key information if it turns out a malicious change was
pushed. If we enable pushing from gerrit (with some set of gerrit users
who also have write access being able to cause gerrit to push a change),
we should consider how to track which gerrit account it was that caused a
change to be pushed, in a similarly distributed way.)
> If on the other hand it is just an optional tool to help a contributor
> to produce their commit in a collaborative fashion, then it's indeed
> silly to ask for a re-review once the contributor is satisfied with
> what they've got.
I'm thinking of it as an optional tool to help the community review
changes and track the state of changes being proposed (by providing the
database of those that have yet to be pushed to master or abandoned, and
possibly in future by providing other features such as automatic CI to
help reviewers).
--
Joseph S. Myers
joseph@codesourcery.com