Re: Solaris seg fault

["Mark R. Bowyer" <Moredhel earthling net>]

>Date: Tue, 12 Jan 1999 00:57:18 -0800
>From: Brian Kurotsuchi <bkurotsu@toocool.itslab.calpoly.edu>

>#0  0x0 in ?? ()
>#1  0xef2df2e8 in gdk_im_open () at gdkim.c:313
>#2  0xef2cd63c in gdk_init (argc=0xefffe9ac, argv=0xefffe9a8) at gdk.c:541
>#3  0xef3fd678 in gtk_init (argc=0xefffe9ac, argv=0xefffe9a8) at gtkmain.c:272
>#4  0xef71c260 in gnome_add_gtk_arg_callback ()
>#5  0xef1e27dc in invokeCallbacks ()
>#6  0xef1e2710 in invokeCallbacks ()
>#7  0xef1e38ec in poptGetNextOpt ()
>#8  0xef0f4488 in gnomelib_parse_args ()
>#9  0xef71cc0c in gnome_init_with_popt_table ()
>#10 0x1ecb4 in main (argc=1, argv=0xefffed5c) at main.c:73

>
>gdkim.c:313 is:
>XRegisterIMInstantiateCallback (gdk_display, NULL, NULL, NULL,
>                    gdk_im_instantiate_cb, NULL);


Yup, this is a known issue.  The workaround is to go and rebuild your GLIB/GTK 
libraries starting with a "./configure --disable-xim".  Can't remember who told 
*me* this and figured out what was happening here, but thanks to them.

There appears to be a bug/difference in the way the XIM stuff is handled in 
Solaris that causes a false to be returned when it isn't checked for in gdk, and 
so you get the SEGV.  Switching this off just means GDK never tries to 
instantiate this stuff, so you don't get the crash.

I've got access to the Solaris source, but it doesn't appear to have changed 
from the standard X or the XFree86 code in the same area.  Very odd.  The *real* 
fix though would be for GDK to check for the false return, IMHO.

I've now got E0.15 and all of GNOME running on it, though.

Ta,

-------My opinion - Not sane, intelligent or necessarily useful-------
o o                                      mailto:Moredhel@earthling.net 
/v\ark R. Bowyer.  http://i.am/Moredhel  mailto:Mark.Bowyer@UK.Sun.COM
`-'                                             I'm the dots in .co.uk