[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [cdt-dev] Not using Gerrit for minor fixes?

I think there are really two choices for committers.

1) If you want people to see your changes, put them on Gerrit and +2 the
change request yourself. Don't use bugzilla anymore. I'd like to use
Gerrit for this since it's much easier to see what changes are in the
queue and what has gone in.

2) Just push the change yourself directly to the main git repo. This is
what most of the committers are doing at the moment. (and I know because I
keep an eye on the cdt-commits list).

Right now, we really only have 5 or so committers regularly committing. I
think it's too much to demand code reviews with this few developers, and
especially since we don't have a lot of overlap in expertise. Reviews are
recommended, and we saw a good exchange between Mikhail and Marc that
resulted from that, so they do have a lot of value. But we need to balance
that with keeping the wheels going. For that, I trust the committers to
use their best judgement there and "do the right thing".


On 12-04-20 3:37 PM, "Marc Khouzam" <marc.khouzam@xxxxxxxxxxxx> wrote:

>last year we had an email discussion about committers reviewing
>other committers' patches.  We said that it was ok for a committer
>to commit changes, and then ask for a voluntary review in bugzilla.
>This was because:
>1- it is sometimes hard to find another committer to review changes
>2- blocking a commit until the review is done, puts pressure on the other
>committer who may not have time currently for the review.
>How do we handle this with Gerrit?
>I currently have a minor commit to remove warnings.  It makes my
>life much simpler to commit it right away.  But I still feel I should
>ask for a (optional) review.
>Should I commit the code from git, and also push it to Gerrit?
>Or should I push it to Gerrit and commit it myself from Gerrit,
>marking the code as reviewed?
>I don't like the first option because it would not be clear in
>Gerrit that the code has already been committed.
>But I'm not a big fan of the second option because no one will
>review the code if I've already marked it as reviewed.
>Any opinions?
>What I'm going to do now, is stick to bugzilla and not use
>Gerrit in this case, which sadly removes access to the nice
>code-reviewing features of Gerrit.
>cdt-dev mailing list