Encourage users to get backtraces using apport

[Emmet Hikory]

On Dec 13, 2007 11:50 PM, Sebastien Bacher <seb128 at ubuntu.com> wrote:
> Did you run into bugs where apport was not able to upload the report?

    Only a couple where the reporter indicated there was some issue
with the relation between apport and complexities of a local proxy
environment.

> That looks like a corner case or an apport bug. Retracing requires
> having the exact same versions of the packages installed and we usually
> don't deal with attached .crash very quickly, which means it's hard to
> retrace those.

    For crashes in stable releases, I feel we have a much better
chance of being able to retrace a report successfully, as the package
set remains basically constant.  For crashes in the development
release, I agree wholeheartedly.

>  The number of bugs is already higher than what we can
> deal with and important bugs often get reported again quickly we have no
> real need to make special efforts to deal with non-apport generated
> crashed

    Maybe I don't understand then.  I do want the apport generated
crash report, but if apport itself runs into some problem uploading
(maybe the crash happened while the reporter did not have internet
access (e.g. on a plane)), I'd rather see the apport crash bundle in
the bug for analysis than either a gdb generated backtrace or no bug
at all (assuming that frequently reported bugs are properly marked as
duplicate).  For some of universe there is a sufficiently low number
of users that a given bug may go undetected for 8 months, which is
less than ideal.  By encouraging the attachment of the apport report
where apport failed due to proxies or interrupted internet access or
other similar reasons, we have a better chance of addressing these in
a timely manner.

    To put it another way, I'd be much happier if we had 500,000 open
well-triaged bugs than 0 open bugs.  In the former case, there's a
clear indication of what needs to be done (although we may not be able
to finish in any reasonable time), whereas in the latter case, we have
no idea what is wrong with the system.

-- 
Emmet HIKORY