On Tue, 14 Jun 2011 10:03:41 -0500, Zach Pfeffer <zach.pfef...@linaro.org> wrote: > On 14 June 2011 07:52, James Westby <james.wes...@linaro.org> wrote: > > On Tue, 14 Jun 2011 06:22:23 -0500, Zach Pfeffer <zach.pfef...@linaro.org> > > wrote: > >> No, we shouldn't have to adjust our hosting. Take a look at > >> http://android.git.kernel.org/ where all the Android trees are hosted. > > > > What does this mean for the hosting of Gerrit though. Does this mean > > that Gerrit has to be on a server with enough space to host all the > > trees that will be put through review? Is it one tree per-user? > > > >> > This would take approved changes and attempt to merge them in to a > >> > staging tree. From there it would do a build test, and then ideally push > >> > the results through LAVA. If all of that is successful then it would > >> > replace the current LEB tip with the staging tree and start again? > >> > >> Yup. > > > > Ok, that sounds like the biggest chunk of work. Are there any standard > > extensions to Gerrit to drive some of this? > > I'm not sure. This would be something to ask on > repo-disc...@googlegroups.com or to Nasser at QC who's gone through > this before.
I searched that group and it seems that most people are using something like the Jenkins Gerrit-trigger plugin to test each proposed change, rather than the tested merge that we are talking about. http://code.google.com/p/gerrit/issues/detail?id=618 While we use Jenkins under the hood for the Android build service I don't think that this plugin is very useful to us at this stage given that we want to start with testing the merges before they happen. I'll write up a spec on doing the tested merge involves. Thanks, James _______________________________________________ linaro-dev mailing list linaro-dev@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-dev
