Re: gtk test programs core dump

[ERDI Gergo <cactus cactus rulez org>]

On Fri, 22 Jun 2001, Brian Cameron wrote:

> Erdi:

oh btw Gergo is the first name (don't feel bad about it, noone gets it
right:)

> > does this `controleld environment' start from scratch each day?
> 
> Yes sir.

This can't be. Your GTK+ is compiled with different headers than your
installed glib. For example, let's say GObject is defined like this:

struct GObject
{
	int old_field;
	int new_field1;
	int new_field2;
}

and there is a derived object, GFoo, in libgfoo:

struct GFoo
{
	GObject parent

	int foo_field;
}

if you compiled libgfoo with the old glib headers,
sizeof(GFoo)==sizeof(GObject)+sizeof(int)==2*sizeof(int). If you compiled
libgfoo with the new glib headers,
sizeof(GFoo)==sizeof(GObject)+sizeof(int)==4*sizeof(int).
So runnign the old libgfoo with the new glib will mean that
sizeof(GObject)==3, yet sizeof(GFoo)=2, and this can't be since GFoo is
derived from GObject so it should include GObject which means it should be
at least as big as GObject.

Hope this explanation helps
	

-- 
   .--= ULLA! =---------------------.   `We are not here to give users what
   \     http://cactus.rulez.org     \   they want'  -- RMS, at GUADEC 2001
    `---= cactus cactus rulez org =---'
Gravity doesn't exist: the earth sucks.