Re: compiling okay, success running gtk-app computer dependant

[Keith Sharp <kms passback co uk>]

On Sat, 2004-11-27 at 15:37 +0100, edward hage wrote:

> ----------------------------------------
> 1: // name this program = time.c
> 2: #include <time.h>
> 3: #include <gtk/gtk.h>
> 4:
> 5: // compiled with: gcc -g time.c -o time `pkg-config gtk+-2.0 --cflags --libs`
> 6:
> 7: int main()
> 8: {
> 9:  struct tm tms;
> 10:  time_t current_time;
> 11:  GDate *date;
> 12:
> 13:  current_time = time (NULL);
> 14:
> 15:  tms = *localtime (&current_time);
> 16:  g_print("time is %d\n", time(NULL));
> 17:
> 18:  g_date_set_time (date, current_time);
> 19:  g_print("day = %d\n", date->day);
> 20:  return 0;
> 21: }
> ---------------------------------------
>
> The problem does not appear when I do not define tms (blank line 9 and 15)!! But the 
> problem is that in my 'real' program I want to use both tms and current_time.
>
> The segmentation fault occurs after that the first line is printed. So this must be at 
> g_date_set_time.
>
> The big question is in my view: Is it a bug in glib / g_date_set_time that it can not deal 
> with tms being defined or someting that is not okay with my computer?

I think the problem is that localtime(3) returns a pointer to a
statically allocated tm struct.  You should change line 9 to read:

        struct tm *tms;

and line 15 to read:

        tms = localtime (&current_time);

With these changes your program works on my Fedora Core 2 system.
Without the changes, your original code, I get a segmentation fault.

Keith.