Apport Retracing Service

[Shane Fagan]

On Mon, 2010-04-12 at 22:09 -0700, Bruce Miller wrote:
> If anyone checks out any of the bugs which I wrote up in an earlier message, they will come away with the impression that I have a bee in my bonnet about the functionality of the Apport retracing service.
> 
> This message attempts to move the problem from editorial comments in bug reports to a broader forum.
> 
> Let us take the following scenario:
> A year ago, user A was bitten by a configuration bug in a development version of Jaunty Jackalope 9.04. He dutifully filed bug 101. Ubuntu developers dutifully studied, reproduced and fixed the bug. The fix was propagated later in the development process, but before the new release went out as "final."
> 
> A year later, user X is bitten by a configuration bug in a development version of Lucid Lynx 10.04. He dutiflly files bugs 201. The symptoms of the bug are remarkably similar to old (and fix released) bug 101. Along comes the Apport Retracing Service and marks bug 201 as a duplicate of bug 101. Bug 201 is thus assigned to oblivion, never to receive a developer's TLC.
> 
> Nonsense, say I.
> 
> It is an interesting but awkward coincidence that the symptoms of Bug 201 are so similar to Bug 101. But if the problem of Bug 101 has been genuinely fixed, then by definition is not a duplicate. It is a new bug.
> 
> At a stretch, one can concoct a scenario where some old bad code from the days of Bug 101 got inadvertently re-used. But if so, that's a big problem and marking Bug 201 as a duplicate is exactly the *_wrong_* thing to do. Developer TLC is definitely called for.
> 
> Am I raising a known and much-discussed problem with the Apport Retracing Service? I have only just joined this list. If not, is this the right place to raise such concerns? Should I be filing a bug against Apport? or against Launchpad itself? If so, where?
> 
Actually the apport retracing service is very very good. Maybe your
problem is with launchpad not merging any of the info you have put in
the bug report. I think if the bug reporters made sure their bug wasnt
already reported there wouldnt be an issue here. Launchpad does a great
job of offering a list of bugs with similar titles so the user needs to
just look at that list and check "is this what happened to me?" (then
they mark affects me) and "could I offer anything useful to this bug
report?"(then comment the bug). The duplicate bugs should be marked for
oblivion because if the bug reporter checked in the first place they
wouldnt have made the duplicate.

-fagan