Re: [[gtkmm] Destroying container problem]

["Andrew E. Makeev" <andrew solvo ru>]

Murray Cumming wrote:

> On Tue, 2002-07-02 at 16:40, Andrew E. Makeev wrote:
> >
> > Unfortunatelly, I couldn't reproduce this one with small example...
>
> You need to break it down. I wouldn't do anything that you couldn't do,
> until you discover something that you think is a gtkmm bug.
>

You know, I am tired of getting this bug separated from my whole app...

It's disapperaing once I cut my code off to leave only problem part compiling.

To remove the problem I redesigned interaction btw my app classes several times. I
am using now signal-slot function connection from one class to another instead of
passing class pointer to another class constructor. Also, I am using manage()
mostly to create widgets that gonna be added to containers. But problem is raising
time to time when my app is growing further.

First time: segfault in destructor when removing first packed widget (Gtk::VBox)
from my class based on Gtk::Window.
solution: removed with manage().

Second time: segfault when my app is being started - add first widget (Gtk::VBox)
to my class based on Gtk::Window. I used manage() again, but then same segfault
happens in other class with same conditions (ctor {add(pVBox)...}), and that pVBox
was created with manage().
solution: created signals in my menu classes to set interface properties instead of
passing interface class pointer to ctor.

And now (third time): segfault in same place as in second time.

I think, you should have my example in complex to catch that. Of course, I will
remove as much code as I could to reproduce bug.

One more question: is there any relation with g++ version I am using? (It is 2.96)

Thanks
-Andrew