Re: [paragui-users] Help with menus...

[Atani]

On Monday 24 June 2002 07:47, Alexander Pipelka wrote:
> Yes. It definitely depends on "when" the app object will be destroyed.
> If it's in the main loop everythings ok because the destructor of
> PG_Application is able to cleanup all objects.
>
> BTW, there's a new static function PG_Application::GetApp()

Speaking of which, in the GetApp() method, there should be a check to ensure 
that pGlobalApp is non-NULL.  When it is NULL, go ahead and initialize it.

Mike

>
> :)
>
> Alex
>
> Am Mon, 2002-06-24 um 16.09 schrieb Atani:
> > Interestingly enough, if i declare it (globally) as:
> >
> > PG_Application *app;
> >
> > and then in main do: app = new PG_Application(); it doesnt fail out
> > either. So something is perhaps dependent on when it executes... do you
> > use atexit to trigger cleanups?
> >
> > Mike
> >
> > On Sunday 23 June 2002 13:18, Alexander Pipelka wrote:
> > > Hi Mike!
> > >
> > > That's a really weird one .. :))
> > >
> > > Ok. Creating the PG_Application object inside main() solves the
> > > problem. I haven't fully debugged it till now but it seems there is
> > > some confusion in the shutdown/cleanup code.
> > >
> > > Alex
> > >
> > > Am Son, 2002-06-23 um 19.41 schrieb Atani:
> > > > On Sunday 23 June 2002 10:31, Eric Ross wrote:
> > > > > Mensaje citado por Atani <address@hidden>:
> > > > > > Hi all,
> > > > > >     Using the attached code I generate a SIGSEGV everytime i close
> > > > > > the app..
> > > > >
> > > > > Can you attach some _working_ code ? I mean something we could
> > > > > compile, execute and test ?
> > > >
> > > > Actually, the only thing that is missing from that zip (globals.h,
> > > > includes headers, etc... and the main method..)
> > > >
> > > > I have attached the full source for the client app... its actually a
> > > > very basic app thus far that I am doing... The final project will be
> > > > an online RPG of sorts :)
> > > >
> > > > Note, to compile SDL_net is required...
> > > >
> > > > > I've got what you're describing (SEGV when closing the app)
> > > > > sometimes when there's memory corruption(buffer overrun and so).
> > > > > Maybe you could try linking with some memory tracer or
> > > > > something(libefence?) to check it's really a bug ?
> > > >
> > > > Thats the really strange part... I believe that I have done something
> > > > very basic as to expose the bug it just requires running the app and
> > > > then close the app... do nothing in the app... From what I have found
> > > > so far, it appears that calling app->Quit() still allows some events
> > > > to be processed and when they are processed some parts of PG are
> > > > already cleaned up and so the SIGSEGVs are appropriate as an NPE type
> > > > case.
> > > >
> > > > Mike
> > >
> > > _______________________________________________
> > > paragui-users mailing list
> > > address@hidden
> > > http://mail.freesoftware.fsf.org/mailman/listinfo/paragui-users
> >
> > _______________________________________________
> > paragui-users mailing list
> > address@hidden
> > http://mail.freesoftware.fsf.org/mailman/listinfo/paragui-users
>
> _______________________________________________
> paragui-users mailing list
> address@hidden
> http://mail.freesoftware.fsf.org/mailman/listinfo/paragui-users