[openstack-dev] [Stackalytics] 0.1 release [metrics]
stefano at openstack.org
Fri Jul 26 00:10:07 UTC 2013
On 07/23/2013 07:25 AM, Roman Prykhodchenko wrote:
> I still think counting lines of code is evil because it might encourage
> some developers to write longer code just for statistics.
Data becomes evil when you decide to use them for evil purposes :) I
don't think that lines of code is a bad metric per se: like any other
metric it becomes bad when used in an evile context. I'm getting more
and more convinced that it's a mistake to show ranks and classifications
in the dashboard and I'll be deleting all the ones that we may have on
Counting anything in OpenStack, from commits to number of reviews is not
a race, we don't need to *rank* top contributors. What we need is to
identify trends. Practical example: in the report for Grizzly, most
metrics put Red Hat and IBM visibly on top of many charts, while in
Folsom their contributions were much lower. The story of those numbers
was that IBM and Red Hat changed gear since Folsom and from 'involved'
became visibly and concretely 'committed'. The story of those metrics
was not that Red Hat was first or second in some sort of race.
We should keep in mind that commits or bug resolutions to different
projects are not directly comparable, that line charts can damage the
appearance of some companies/people (loose face). Other charts need to
be explored (punch cards?) and avoid direct comparisons, maybe?
Ask and answer questions on https://ask.openstack.org
More information about the OpenStack-dev