On Fri, Feb 22, 2013 at 12:05 PM, Dave Cahill <dcah...@midokura.com> wrote: > Hi, > > Since it appears there is consensus in this thread for at least setting up > Gerrit (if not for mandating a certain number of +1 reviews), what is the > next step? > >> This is also a _dramatic_ change in culture - moving from a commit >> then review to a review then commit - this will dramatically slow down >> project development. > > IMO, the chaos in master the past few weeks (still not working end-to-end > from a fresh checkout after 2-3 weeks) has slowed down project development > more than code review enforcement could have. I've certainly lost several > days' development time to debugging. >
+1 I completely understand and I think we should have a code reviewing process for branch merges or feature merges at least. For folks who think it can slow us down, think what dcahill suggests. Committers can still go ahead and commit into the branches, but at least they will know something like gerrit or any other code reviewing platform exists. Regards. > From discussions in cloudstack-dev IRC, it seems I'm not the only one > frustrated with the breakages. Maybe slowing down commits a little in order > to get reviews done would atually speed overall development. > > Thanks, > Dave. > > > On Wed, Feb 20, 2013 at 1:20 AM, Chip Childers > <chip.child...@sungard.com>wrote: > >> On Thu, Feb 14, 2013 at 03:13:11PM +0530, Prasanna Santhanam wrote: >> > One thing I've seen is that almost all >> > reviews in OS come in with tests and without passing the devstack >> > tests (also integrated in gerrit) the patch isn't even picked up by >> > committers for code review. >> >> This is a practice that we should be following *now* for commits as well >> as patches submitted for review. >> >> -chip >>
