[openstack-dev] Bad review patterns

Sean Dague sean at dague.net
Sun Nov 10 07:50:56 UTC 2013

Not that I know of. I've considered writing my own gerrit front end
mail service to do just that, because I agree, the current mail volume
and granularity is not very good. If I manage to carve time on it,
I'll do it on stackforge. Joe Gordon took a different approach and
wrote a front end client to mark review threads read that are past.

On Thu, Nov 7, 2013 at 8:40 PM, David Ripton <dripton at redhat.com> wrote:
> On 11/07/2013 07:54 PM, Sean Dague wrote:
>> On 11/08/2013 01:37 AM, Pedro Roque Marques wrote:
>>> Radomir,
>>> An extra issue that i don't believe you've covered so far is about
>>> comment ownership. I've just read an email on the list that follows a
>>> pattern that i've heard many complaints about:
>>>         -1 with a reasonable comment, submitter addresses the comment,
>>> reviewer never comes back.
>>> Reviewers do need to allocate time to come back and follow up on the
>>> answers to their comments.
>>> Perhaps there is an issue with the incentive system. You can earn karma
>>> by doing a code review... certainly you want to incentivise developers that
>>> help the project by improving the code quality. But if the incentive system
>>> allows for "drive by shooting" code reviews that can be a problem.
>> It's not really an incentive system problem, this is some place where
>> there are some gerrit limitations (especially when your list of reviewed
>> code is long). Hopefully once we get a gerrit upgrade we can dashboard
>> out some new items like that via the new rest API.
>> I agree that reviewers could be doing better. But definitely also
>> realize that part of this is just that there is *so* much code to review.
>> Realize that most core reviewers aren't ignoring or failing to come back
>> on patches intentionally. There is just *so* much of it. I feel guilty
>> all the time by how big a review queue I have, but I also need a few
>> hours a day not doing OpenStack (incredible to believe). This is where
>> non core reviewers can really help in addressing the first couple of
>> rounds of review to prune and improve the easy stuff.
>> We're all in this together,
> Is there a way for Gerrit to only send email when action is required, rather
> than on any change to any review you've touched?  If Gerrit sent less mail,
> it would be easier to treat its mails as a critical call to action to
> re-review.  (There's probably a way to use fancy message filtering to
> accomplish this, but that would only work for people willing/able to set up
> such filtering.)
> --
> David Ripton   Red Hat   dripton at redhat.com
> _______________________________________________
> OpenStack-dev mailing list
> OpenStack-dev at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Sean Dague

More information about the OpenStack-dev mailing list