libpurple crashing since 18 months

[Felipe Contreras]

On Thu, Mar 26, 2009 at 6:05 PM, Felipe Contreras
<felipe.contreras at gmail.com> wrote:
> On Thu, Mar 26, 2009 at 5:20 PM, Sascha Vogt <FunkyFish at gmx.net> wrote:
>> Hi,
>>
>> please note: I'm just a Pidgin user, no developer nor anyhow related to
>> the Pidgin project, apart from reading the mailing-lists.
>>
>> Thorsten schrieb:
>>>> and bugs are fixed through bug tracker, not by direct communication
>>>
>>> I filed my bug report  #7621 in November 2008, and the first reply came
>>> in January, but nothing happened since then, so it is 4 months old and
>>> just sits there. Do you call *THIS* fixing bugs?
>> You were asked for a valgrind log. This question has never be answered.
>> Do YOU call this communicating?
>
> That was requested 23 hours ago.

That backtrace already gives pretty good information, if this bug was
reported on msn-pecan first I would ask for a debug log, because it's
clear the slplink has been freed by the time the switchboard is freed,
a debug log *might* give a hint as to why the slplink was freed, and
when.

But in fact, if you understand the code you would see that the whole
freeing of structures is *very* error-prone, it should be rewritten to
use recounts, and reorganized to avoid these problems as much as
possible.

I already started doing that in msn-pecan, but there are still issues
that are hard to reproduce, that's why I need to implement a test
system. Not only to track and fix the problems, but also to avoid
regressions in the future.

Even if the valgrind log was requested long time ago, that's not an
excuse for not fixing the software. The bug is still there, it's up to
the developers how to chase it, if at all.

-- 
Felipe Contreras