Glynn Clements <glynn(a)sensei.co.uk> writes:
> > 2 --- SIGHUP (Hangup) ---
> > 3 --- SIGCONT (Continued) ---
> > 4 <... select resumed> ) = -1 ENOSYS (Function not
implemented)
>
> It is getting the SIGHUP but it is not dying... (Are signal handler
> settings preseverd over fork/xexec.
According to Stevens[1]:
a) All signal dispositions are inherited across a fork().
b) exec() resets any signal for which a handler is installed to its
default disposition. Any signal which was ignored remains so.
Then that may be it. Since some betas XEmacs now lets SIGHUP to remain
SIG_IGN when it was. However gnuserv SHOULD also die in that case.
That is probably the real problem problem in that gnuserv gets its
standard streams in a weird state.
giacomo, can you reproduce the problem using
nohup xemacs -f gnuserv-start &
from an xterm ?
Jan