|Summary:||Bugzilla only notifies port maintainers when issue is created in Individual Port(s) component|
|Product:||Services||Reporter:||Yuri Victorovich <yuri>|
|Component:||Bug Tracker||Assignee:||Bugmeister <bugmeister>|
|Severity:||Affects Only Me||CC:||bugmeister|
Description Yuri Victorovich 2018-03-31 20:43:14 UTC
Please see https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=227145 devel/nlohmann-json has a maintainer, yet he wasn't assigned.
Comment 1 Kubilay Kocak 2018-04-01 02:17:40 UTC
The (ports) auto-assigner only processes issues created within the "Individual Port(s)" component, and bug 227145 was created within the "Ports Framework" component (since re-classified), so is accordingly 'Works As Intended'. However: It may be possible to improve the behaviour to include those created within the Ports Framework (and/or other) ports component, though any potential changes should consider: 1) In the majority of cases (historically), Issues matching a summary "category/portname" within the Ports Framework component are misclassifications, and should instead have been created be in "Individual Port(s)". 2) That issues specific to a port (with category/portname in the summary) could be created *correctly* within the Ports Framework component. In this case portmgr@ should still be the resulting Assignee (correct area of responsibility, via correct component), but the maintainer could still be added to (potentially 'additionally') respective maintainer-* flags, (which AA does not currently do) 3) That there is a special/specific case where the maintainer is a committer, in which case the auto-assigner 'assigns' that person, not just adding their email to maintainer-* flags. This conflict of who should be the assignee (committer maintainer or portmgr) in that case must have a clear, obvious and unambiguously 'correct' resolution, which on the surface is not possible. Having said that and generally speaking, AA's purpose is two-fold, 'assignment' for those cases where responsibility can be derived, and notification, for obvious reasons. If we consider only the 'notification' case within this issues context, notably EXCLUDING dealing with or changing Assignee in any way, then I think 'adding people to maintainer-* flags and/or CC lists (where multiple ports are identified in summary), in all (component) cases should be OK, but we'd want to verify that that holds in all other (currently implemented in code) AA cases.