Crash telemetry is essential to maintain quality, but Red Hat is not assigning developers to work on ABRT anymore. Our repeated calls for help on Fedora devel@ have not resulted in any interested developers stepping up to maintain ABRT. So, I propose that we completely remove all of ABRT in Fedora 45. It requires significant modernization work, but it is entirely abandoned.
If anybody is willing to maintain ABRT, please let us know. Iâd be happy to withdraw this proposal if a maintainer were to emerge.
Preferably we would make a lot of improvements to ABRT. It shouldnât be spamming maintainers. At the very least, if it doesnât work well for a package, a simple off switch ought to exist. Ideally we would have some sort of scheme to control how bug reports are handled for each package. If package version is new enough, report to configurable upstream issue tracker rather than downstream. A denylist of known symbols that should result in the bug report not being created if present; e.g. WebKit always crashes in the same place if you update it while an application is using it, and I donât want to see those crash reports. It also needs more aggressive dropping of duplicate issue reports.
Suffice to say, maintaining ABRT should not be considered just a matter of keeping the lights on. It needs some serious work.
@catanzaro, having it removed from the default package selection is a shame. However, what I meant to ask was whether it has been deprecated:
I ask because I frequently utilise it to report (SELinux) errata:
âŚand would be unable to otherwise, I believe. Unless you know of a replacement that can be utilised on a desktop computer. Apologies if this tangent has diverged too far.
Thanks, but that is not very helpful. You might notice that the reaction on the comment you linked was done my me - so I was already following this ticket.
However, this thread is about âLast call to save ABRTâ. The comment you linked exclusively lists problems with FAF, not the abrt tooling itself.
At risk of responding to your assumption with another assumption: probably?
To keep the ABRT service, I would expect somebody to fix the retrace server, and monitor (at least read) incoming bug reports (bug reports against ABRT itself).
(To bring back the GUI would require further discussion, because the GUI has many warts.)
The fact that you donât know the answer to this question doesnât exactly inspire confidence for the future of this project even if somebody shows up to help.
Fair. The GitHub project has two maintainers: @msuchy and @msrb. Can you please answer Fabioâs question: would access to the relevant upstream ABRT projects be granted to a new maintainer? (I guess it would probably be the usual answer: yes, if the contributor builds a little trust first?)
It is quite unrealistic to expect this level of ongoing effort on a
strictly volunteer basis. Especially with the apparent expectation of
being on call to babysit the service rather than just maintaining the software.
Without a âserious budgetâ to attach to this âserious workâ your
feature/requirement list might as well end with âand another ponyâ.
I am there for historic reasons only. I handed over the project to @msrb ages ago. I personally do not have a problem handing it over to any maintainer willing to work on it. But is is really up to @msrb
FYI - The current situation does not make me happy, as my name is still somehow stuck with ABRT. In recent days, I asked several people at Red Hat to do something about ABRT. Either properly kill it (or offer it to any other maintainer) or give it some attention. They incline to the former, but even that may take some time.
Would access to the relevant upstream projects be granted / transferred?
Absolutely.
However⌠as other people already pointed out elsewhere, ABRT requires a lot of work. The project has been in a maintenance mode for the last decade (if not longer). The project tries to covers a lot of use cases, but none of them very well (imo).
Also, ABRT will stop working once Fedora moves away from Bugzilla.
If there are any volunteers interested in problem/crash reporting in Fedora, Iâd suggest exploring alternatives.
Thanks for the clarifications! Sounds like if somebody would be interested in crash reporting / analytics, it would be better to start a new project with much more limited scope.