Re: 0 bot for Git

From: Fengguang Wu <lkp@intel.com>
Date: Apr 13, 2016, 13:44 UTC
Message-ID: <20160413134427.GA21158@wfg-t540p.sh.intel.com>
In-Reply-To: <xmqqmvoypn7g.fsf@gitster.mtv.corp.google.com>

Show 11 quoted lines

> > I don't know how 0 bot solves this, but the obvious issue with this
> > approach is to allow dealing with someone sending a patch like
> >
> > +++ Makefile
> > --- Makefile
> > +all:
> > +	rm -fr $(HOME); sudo rm -fr /
> >
> > to the list. One thing that Travis gives us for free is isolation:
> > malicious code in the build cannot break the bot, only the build
> > itself.

Sure, isolation is a must have for public test services like Travis or 0day. We optimize the 0day infrastructure for good behaviors and also have ways to isolate malicious ones.

Show 10 quoted lines

> True, presumably the Travis integration already solves that part, so
> I suspect it is just the matter of setting up:
> 
>  - a fork of git.git and have Travis monitor any and all new
>    branches;
> 
>  - a bot that scans the list traffic, applies each series it sees to
>    a branch dedicated for that series and pushes to the above fork.
> 
> isn't it?

Right. 0day bot could auto maintain a patch-representing git tree for Travis to monitor and test. As how we already did for the linux kernel project, creating one git branch per patchset posted to the lists:

        https://github.com/0day-ci/linux/branches

In principle the git project should have more simple rules to decide "which base should the robot apply a patch to". But we do need some hints about the git community's rules in order to start the work. If without such hints from the community, we may start with dumb rules like "apply to latest origin/master" or "apply to latest release tag".

Thanks, Fengguang