David Faure faure at
Thu Dec 1 21:59:50 GMT 2005

On Thursday 01 December 2005 22:47, Aaron J. Seigo wrote:
> On Thursday 01 December 2005 14:34, David Faure wrote:
> > On Thursday 01 December 2005 22:07, Hamish Rodda wrote:
> > > ==9616==    by 0xE5A8B8F: FileSelectorPart::~FileSelectorPart()
> > > (fileselector_part.cpp:62)
> >
> > Deleting this part properly (i.e. before kapp is deleted) is the usual fix
> > for this; I think kdevelop is leaking this part currently (well, not really
> > since klibloader deletes it, but it's too late then).
> and as i said on irc, i believe that working around these issues in app code 
> is a poor solution. =(
> klibloader says "i will clean up for you if you let me"
> and kdirlister says "i'm sensitive to when i'm cleaned up"
> the latter is an internal  implementation detail that is, for all intents and 
> purposes, hidden from the app developer's point of view.

kdirlister is far from being the only thing that's sensitive to post-qapp deletion.

I understand that you would prefer the libs to be more robust, and I don't disagree,
I'm just saying that up to now the way we fixed such crashes was by ensuring
that we properly delete our objects instead of letting the (often too-late) cleanup of
KLibLoader happen.

David Faure, faure at, sponsored by Trolltech to work on KDE,
Konqueror (, and KOffice (

More information about the kde-core-devel mailing list