This feature is more about improving the stats counting rather than counting if a user is active or not, because the user activity is already accounted in the pagure/src.fp.org activity. To give an example, the most recent report was
### Found 1429 users in the packager group. ### ### Found 558 users with no builds in Koji over the last year. ### ### Found 277 users with no activity in pagure/src.fp.org over the last year. ### ### Found 253 users which didn't post any message in Fedora Discussion over the last year. ### ### Found 227 users which also show no activity in Bodhi over the last year. ### ### Found 204 users which also show no activity in mailing lists over the last year. ### ### Found 134 users which also show no activity in Bugzilla over the last year. ### ### 134 packagers out of 1429 (9.4%) found inactive. ###
The Koji inactivity seem rather high considering the src.fp.org count, which could be because of packit workflow taking over the koji activity. Similarly these could overlap in Bodhi and bugzilla since it can take over some of those aspects as well.
packit
The feature request here is to try and pick up activity in src.fp.org, specifically PR activity, and if the project has activity due to packit user then attribute the koji/bodhi activity to the user that did the PR merge (or has PR activity in general on that project).
Some of the mismatch you see is due to people like me who have not made a Koji build in the last year (or at least I don't think I have), but I have had activity on Pagure. Note that the check is for both pagure.io and src.fedoraproject.org.
I don't see the value in trying to go back and adjust the Koji count after the check has been run.
I think the stats as I reported might be misunderstood. You should read: "out of 1.429 users in packager group, 558 did not fire a build in Koji. Of these 558 users, 277 had no activity in src.fp.o/pagure.io neither". So, there were ~300 users which had activity in src/pagure, but did not build anything, perhaps due to packit.
Anyway, I think merging PRs is counted in pagure activity, so if a user just merges a PR from packit it should be already marked as active. I need to check Pagure APIs.
Oh, this makes more sense reading those numbers now. It didn't feel right when counting Bodhi as well, since I would assume it would have similar numbers to Koji.
Then it wouldn't make much sense to do that filtering.
Metadata Update from @lecris: - Issue close_status updated to: Not planned - Issue status updated to: Closed (was: Open)