On Mon, Apr 25, 2011 at 09:16:50AM -0600, Tim Flink wrote:
I was talking to some of the other Fedora QA people about how to
encourage more testing of F(n-1) so that we don't have quite so many
updates languishing in updates-testing and one of the things that came
up was more data on the testing of updates.
I know that I could write some code to pull data from bodhi but since
there are already some metrics there, I figured that I would ask you all
about it first to see if it might make sense to have them in bodhi
instead of writing my own standalone metrics..
My first thoughts on what we would like to see are:
- Most active proventesters for a release in the last X days
- Possibly in the form of karma given, by tester in the last X days
- Most active proventesters for a given component
- Most active non-proventesters karma givers
Do you have any thoughts on whether these kinds of metrics would be
appropriate to include in bodhi? If not, do you have a suggestion for
gathering the data I would need other than the bodhi client? Pulling
down data through the bodhi client and populating my own database just
seems like a rather long and obtuse way of going about things.
Definitely do-able but not necessarily ideal.
I think these metrics would be great inside of bodhi.
Mo and I did some brainstorming at FUDCon for what we want Bodhi v2.0 to
look like, and this was the initial mockup:
http://lewk.org/img/bodhi-20-frontpage.png
As you can see we really want to shine the spotlight on testers, and
make it obvious as to what needs testing the most.
Bodhi currently has a basic metrics API in it that is currently used to
generate some basic graphs. We could potentially generate the
proventesters metrics in there as well.
https://fedorahosted.org/bodhi/browser/bodhi/metrics.py
https://admin.fedoraproject.org/updates/metrics/?release=F13
luke