On Fri, Apr 24, 2015 at 10:31 AM, Doug Hellmann <[email protected]> wrote:
> Excerpts from Amrith Kumar's message of 2015-04-24 15:02:01 +0000: > > There have been many replies on this thread, I'll just reply to this one > rather than trying to reply piecemeal. > > > > Doug, there's asking a question because something is unclear (implying > that the code is needlessly complex, missing a comment, is unintuitive, > ...). I believe that this most definitely warrants a -1 as you describe > because it is indicative that the code, once submitted, would be hard for a > future reader to follow. > > > > In my mind, the yardstick has been, and continues to be this; would a > reasonable person believe that the patch set as presented should be allowed > to merge. > > > > - If I can answer that question unambiguously, and unequivocally with a > NO, then I will score the patch with a negative score. > > > > - If I can answer that question unambiguously, and unequivocally with a > YES, then I will score the patch with a positive score. > > > > - For anything else, I'll use a 0. > > I've run into too many cases where a "trivial" change has an > unintended consequence, so I suppose I'm more conservative with my > code reviews. I don't use 0 very often at all, not because of any > stats counting, but because I don't assume that conveys any information > to the author or other reviewers. I vote the way I mean for my > comments to be taken, so I use -1 to indicate that more work is > needed, even if that work is just explaining something better or > demonstrating that an edge case is going to be handled. > > When I get a -1 on one of my patches with a question, I personally treat it as a short coming of the commit message. To often in the past I have looked at a file, and in trying to figure out why that line is there I do a git blame only to see a useless commit message with me as the author. > > > > If there was a patch to make reviewstats count +0, I'd support that. But > I would also like to understand what the differences are between > reviewstats and stackalytics. Ideally I would like stackalytics to count > 0's as well. If you can take the time to review a patch, you should get > credit for it (no matter what tool is used to count). > > > > I would support changes to both reviewstats and stackalytics to do the > following. > > I'm not sure where all of the interest in stats counting comes from, but > it's definitely not a motivation of my own behavior. > > Doug > > __________________________________________________________________________ > OpenStack Development Mailing List (not for usage questions) > Unsubscribe: [email protected]?subject:unsubscribe > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev >
__________________________________________________________________________ OpenStack Development Mailing List (not for usage questions) Unsubscribe: [email protected]?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
