arts bugs

Stefan Westerfeld stefan at space.twc.de
Fri Jul 4 20:08:57 BST 2003


   Hi!

On Thu, Jun 19, 2003 at 05:00:05PM -0400, George Staikos wrote:
> There's a problem in arts right now where Arts::Object gets a pool that is:
> creator = 0
> created = true
> count = 1
> base = 0x0
> 
> This results in _method_call returning 0, and therefore crashing all the IDL 
> generated code since it assumes either _cache != 0 or _method_call() != 0.  
> In this case, both are 0.  It's very easy to see this crash by running 
> artsbuilder.  What is the proper fix here?  Is this pool no good, or is the 
> IDL broken?  Or is artsbuilder not using the objects correctly in the first 
> place?

I think you're seeing the representation of a "null pointer" here. Calling a
method on a "null pointer" is expected to segfault (as would calling a method
on an ordinary C++ object which is NULL). Code which gets object references
from somewhere should call foo.isNull() prior to calling any methods on foo,
to ensure that no segfaults occur.

I can not reproduce a crash here, though. I can properly load an example and
execute it. I know that there is a problem if you have compiled artsbuilder
with some gcc versions (http://bugs.kde.org/show_bug.cgi?id=51667)... maybe
you can make a backtrace of artsbuilder and see if artsd crashes, too (then
make a backtrace of artsd as well)?

   Cu... Stefan
-- 
  -* Stefan Westerfeld, stefan at space.twc.de (PGP!), Hamburg/Germany
     KDE Developer, project infos at http://space.twc.de/~stefan/kde *-         



More information about the kde-multimedia mailing list