[gnu.emacs.bug] problem losing characters and occasionally quitting unceremoniously

bergman@m2c.m2c.org (Michael Bergman) (03/02/90)

(This is a modification of an article I posted and then cancelled.  My
apologies if it appears in both forms; I don't know how well "cancel"
works, never having tried it before)

I used to have a problem with 18.51 (I think) freezing up on me
periodically (under ultrix on a Vax3600).  And sometimes quitting with
no saving throw, with the error message:

Xlib: sequence lost (0x20000 > 0x12284) in reply type 0x0!

or something similar.  I don't remember if the two problems are
related or not, but I haven't seen the freezing problem for a long
time.  The other one just came back to haunt me again.  Read on...

I posted to this group, and got a couple of useful replies, one of
which said that the problem would be fixed with UWS2.2, and one of
which offered a patch to x11term.c, causing it to look something like
this (ignore the comment bars)

/*	static int server_ping_timer; 	*/
	BLOCK_INPUT_DECLARE ();

/*	if (server_ping_timer > 0)  	*/
/*	  server_ping_timer--;      	*/
/*	else				*/
/*	  {				*/
/*	    server_ping_timer = 100;	*/

	/* Yes, this is really what I mean -- Check to see if we've
	 * lost our connection */
	
	BLOCK_INPUT ();
	XSetErrorHandler(0);
	XSetIOErrorHandler(0);
	...and so on...
/*   }   */

After instituting this change, in 18.54 (as long as I was recompiling
I upgraded), I no longer had freeze-ups or sudden death syndrome, but
would periodically drop a character, apparently sending some sort of
garbage instead.  The result would be as if two characters had been
dropped.  Now that we have UWS2.2, I have recompiled without the
change (hence the comment bars) and the problem is better -- but not
completely gone.  At least the freezing up hasn't returned.  However,
the sudden loss of emacs, with files unsaved, has just returned.  I
have only dropped characters once since recompiling (in about 2 hours
of use) as compared to once every 15 minutes or so before, so there's
a definite improvement -- but meanwhile I was wondering if there was
any insight any of you could offer on this problem.  Since I wrote the
previous sentence, I have had one "death" with the error message given
above.  Help!

--
--mike bergman
	      Massachusetts Microelectronics Center
	      75 North Drive, Westborough, MA  01581, USA +1 (508) 870-0312
	UUCP: (...harvard)!m2c!bergman    INTERNET:   bergman@m2c.org