🚀 go-pugleaf

This is a repost of an article previously submitted to gnu.emacs.bug.

When running Emacs-19.19 connected to an NCD Xterminal running over a
9600 baud modem using NCD's Xremote software, emacs will eventually
shut itself down.  After tracing the code on a Sparc running
SunOS-4.1.3, the problem has been narrowed to the following snippet of
code from xterm.c, starting at line 3349 (I've marked 3 lines of
interest with leading >'s):

#ifdef HAVE_SELECT
  if (expected && ! event_found)
    {
      /* AOJ 880406: if select returns true but XPending doesn't, it means that
         there is an EOF condition; in other words, that X has died.
         Act as if there had been a hangup. */
      int fd = ConnectionNumber (x_current_display);
      SELECT_TYPE mask;
      EMACS_TIME timeout;

      FD_SET(fd, &mask);
      EMACS_SET_SECS_USECS (timeout, 0, 0);
>     if (0 != select (fd + 1, &mask, (long *) 0, (long *) 0, &timeout)
>         && !XStuffPending ())
>       kill (getpid (), SIGHUP);
    }
#endif /* ! defined (HAVE_SELECT) */

Can anyone confirm whether the comment AOJ made in 1988 is still correct?
That is, if select() returns non-zero, and XStuffPending() returns zero,
then it's time to shut emacs down?  Is there perhaps a race condition
that only manifests itself when X runs over a long-delay line?  Oh, and
for what it's worth, I tried adding
	XFlush (x_current_display);
just before the call to select(), as was suggested recently on comp.windows.x.
It didn't fix the problem.


Thanks for any help,
Chris <jepeway.cs.utk.edu>