Receiving bugs from "Crash Catcher" with [faf] in the subject line

[Peter Robinson]

On Tue, Apr 23, 2013 at 12:48 AM, Kevin Fenzi <kevin at scrye.com> wrote:
> On Mon, 22 Apr 2013 19:55:02 +0200
> Rave it <chat-to-me at raveit.de> wrote:
>
>> Am Mon, 22 Apr 2013 16:37:34 +0000
>> schrieb devel-request at lists.fedoraproject.org:
>>
>> For me as compiz maintainer those info's are complete useless whithout
>> having more infomation what the user did if abrt would trigered.
>> Maybe thy played arround without knowledge about the programm and did
>> wrong things?
>> Better the abrt guys forced the user to write what they did if abrt
>> trigger a alarm.
>> For me a bug whithout a comment isn't a bug.
>> And honestly, if i get such bugs without a comment, i asked the user
>> friendly what they did.
>> In case of 50% i get no answer.
>
> 50% is much higher than I have gotten. ;)
>
> On any new abrt bugs I get, I typically:
>
> - Check to see if the submitter filled in the 'description of problem'
>   which something I could try and work into a reproducer. I'd say
>   perhaps 1% do.
>
> - If not, then I ask them what they were doing when the crash happened
>   and if they can reproduce it. I'd say 90% never reply to that and the
>   bug sits there until EOL. Some folks do, and those I can gather info
>   from or ask various troubleshooting things which often results in a
>   fix or at least an upstream bug.
>
> I'm not personally finding these faf reports of much use. They seem to
> have a sanitized backtrace with even less info in them, and I also have
> no way at all of finding out from people who saw this what they were
> doing or ask them debugging steps.

I agree. I mean it's very interesting that I suddenly find out that
one of the packages I maintain has a crash that has repeated 30 times
but without knowing who they are and how to be able to recreate the
problem or asking the reporter for debug or information about the
crash or HW associated the bug is completely useless and just wastes
the little time I have for bug reporting. In a lot of the packages I
(co)maintain a crash without context is completely useless.

Peter