bug with 5 code lines

[Tor Lillqvist]

> on win32 it's easy to reproduce it.

Note that the situation with dbus on Windows is a bit if a mess. There
are several forks. So when you say "on Win32" it is not clear what you
actually mean.

You should tell us exactly where you got your dbus implementation
from. (As KDE seems to be involved based on the file name you mention,
most likely you are using their port/fork, it might be more useful for
you to report this problem on some KDE-on-Windows mailing list.)

> Should i report a bug ?

Yes, but to those maintaining the fork you are using.

(For what it's worth, in my build of "master" of the reference
implementation of dbus with the dbus4win fork merged back in, your
code works fine on Windows. At least in a minimal test program that
does nothing except what you posted. Is you problem in such a minimal
program, or do you see the problem only when your code snippet is part
of a larger program? It is always best to include a *complete* and
*minimal* sample program that exhibits a problem, if one wants others
to be able to reproduce it.)

(Here is your code with stuff added to make it actually compile, this
is what I tested:

#include <glib.h>
#include <glib-object.h>
#include <dbus/dbus.h>

int
main (int argc, char **argv)
{
  DBusError e;
  DBusConnection *conn = NULL;

  g_type_init ();
  dbus_error_init(&e);

  conn = dbus_bus_get(DBUS_BUS_SESSION, &e);

  return 0;
}

)

--tml