Re: Panel segfaults

[Manish Vachharajani <mvachhar vger rutgers edu>]

On Mon, 20 Jul 1998, George wrote:

> On Mon, Jul 20, 1998 at 03:12:46PM -0400, Manish Vachharajani wrote:
> > Ok, well I am not touching the code on this one :).  It seems that there
> > are many interdependencies that I am not aware of.  The function that sets
> 
> the panel code needs to be cleaned up still ... there is quite a lot of
> dirty code here and there, that just needs to be squashed ...
> 
> otherwise it is slightly ugly but I need to document the initial
> "conversation" of applets somewhere ... this was the cleanest solution
> for the problems I found at that moment, but it means there are several
> calls ... and it's not straight forward as to what is happening inisde the
> panel ... :)
> 
> > the tooltips called from apply_global_config definately tries to set the
> > ior, but this maybe, as you said, because of the ORB not being
> > initialized.  At any rate, we can't just move apply_global config into an
> > idle because init_user_panels function will then suffer from the same
> > problem(it too sends info to applets I think).  I think a better bet would
> > be just to initialize the ORB in main right after gnome_config_pop_prefix,
> > provided that this won't break anything else.  The other option would be
> > to put init_user_panels(), _applets(), and apply_global_config(), into an
> > idle function that removes itself, but this is not clean IMHO.  If we want
> > to maintain ORB independence in the main code we could add a
> > panel_corba_orb_init() call instead of the CORBA_ORB_init calls.  What do
> > you think about this?

Looks like Elliot made the necessary changes.  But is there this
mico-glue.h file around?  It seems that it declares stuff for orbit-glue.c
and friends.  Very confusing :)

Manish Vachharajani
<mvachhar@vger.rutgers.edu>