FreeBSD-5.1 problem identified
edscott wilson garcia
edscott at prodigy.net.mx
Thu Aug 7 04:38:02 CEST 2003
On mi? 2003-08-06 at 13:44, Benedikt Meurer wrote:
> On Wed, 06, Aug 2003, edscott wilson garcia wrote:
> > The problem might also affect other xfce4 elements besides xffm. I
> > figured out the problem with FreeBSD-5.1 (which *might* be affecting
> > xfce4-panel, xfce4-session, xfdesktop and xfwm as well). Apparently the
> > sigaction() function does not work to change the default action of
> > SIGTERM (and maybe other signals), but the simplified version signal()
> > works fine. So that by using (in configure.ac):
> > AC_CHECK_FUNCS([sigaction])
> > AC_CHECK_FUNCS([signal])
> > and
> > putting in code like:
> > #ifdef HAVE_SIGNAL
> > signal(SIGUSR1, TuboSemaforo);
> > #else
> > sigaction(SIGUSR1,&act,NULL);
> > #endif
> > xffm with all it's components is now working like a charm in
> > FreeBSD-5.1.
> This is rather useless. Every plattform that has sigaction() has
> signal(), and IMHO its more a FreeBSD bug.
> > The other files that also use sigaction() are xfce4-panel/main.c,
> > xfce4-session/main.c, xfdesktop/main.c and xfwm4/main.c.
> > If anybody has a powerful reason why I should not fix them as well, I
> > will go ahead and do so in a few hours.
> No time to look into this in detail currently. If it serves as a
> workaround for FreeBSD, then "fix" it now, and I'll see to it later
> for a real solution, or maybe someone knows whats the right thing
I'll follow Olivier directive and not proceed with any more corrections.
I think the problem with FreeBSD-5.1 only occurs with signals involving
forked processes, and in that case the other xfce4 modules are most
probably not affected.
> > regards
> > Edscott
More information about the Xfce4-dev