asterixdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ian Maxon <>
Subject Re: Regarding git workflows and who must push a patch
Date Tue, 05 May 2015 23:10:02 GMT
I was looking around a little, and wondering how Github pull requests are
done under Apache infrastructure, because I imagine they must have the
exact same problem. Looking at this page from the Mahout project ( seems to explain how this
works. Basically the final push to the ASF git repo is done by hand.
We could do this for Gerrit as well rather easily. The master branch from
Gerrit would be akin to the pull-request branch, and the developer would
just have to fetch from there and then push that to the ASF master git. I
suppose it isn't as seamless as what was done before, but if other folks
are doing it, I suppose this might be a workable solution.

- Ian

On Tue, May 5, 2015 at 1:23 PM, Ian Maxon <> wrote:

> I don't think the lack of commit hooks or equivalent in the ASF git
> repositories is the issue here. The crux of the problem is ASF
> infrastructure has one way of recording final responsibility of a commit
> (who pushed it, in the server logs), and we have another (who +2ed it and
> cleared it for commit). I would say we do, at a high level, have the same
> level of access control on what comes out of Gerrit. Nothing is +2ed and
> committed, unless someone who is a committer has done so). It's just that
> the method of recording that information is different.
> Regarding using gitpubsub or something like commit hooks to do pre-commit
> testing, those events are only fired off on commit, so we would have to
> have a 'staging' branch or similar that Jenkins watches. I suppose we could
> emulate something like we have with Gerrit now that way, but it would be
> more manual.
> - Ian
> On Tue, May 5, 2015 at 5:31 AM, Jochen Wiedmann <
> > wrote:
>> On Tue, May 5, 2015 at 9:04 AM, Till Westmann <> wrote:
>> > Do our mentors have so experience with this?
>> Sorry: New area for me too. I have created

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message