Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!tut.cis.ohio-state.edu!ZURICH.AI.MIT.EDU!cph
From: cph@ZURICH.AI.MIT.EDU (Chris Hanson)
Newsgroups: gnu.emacs.bug
Subject: (none)
Message-ID: <8908181759.AA13573@zurich.ai.mit.edu>
Date: 18 Aug 89 17:59:08 GMT
Sender: daemon@tut.cis.ohio-state.edu
Distribution: gnu
Organization: GNUs Not Usenet
Lines: 60

Here's a couple of messages that should be of interest.

From cph Fri Aug 18 13:36:45 1989
To: rjh@ai.mit.edu
CC: hp-bugs@ai.mit.edu, system-hackers@ai.mit.edu, rjh@ai.mit.edu
In-reply-to: Robert J. Hall's message of Fri, 18 Aug 89 13:05:26 edt <8908181705.AA23199@praz.ai.mit.edu>
Subject: clisp-mode broken on HPs

   From: rjh@ai.mit.edu (Robert J. Hall)
   Date: Fri, 18 Aug 89 13:05:26 edt

   CLISP mode in gnu emacs used to work, but now something appears
   screwed up in the inter-process communication.  Doing c-C c-E
   to the following form results in some erroneous error message.
   It looks like only so many characters are getting sent to the
   child process, then it starts interpreting the remainder as new
   input without processing any of the first characters.

   (setq barf
	 '(a
	   a a a a a a a a a a a a a a aa a a as
	   a a a a a a a a a a a a a a aa a a as
	   a a a a a a a a a a a a a a aa a a as
	   a a a a a a a a a a a a a a aa a a as
	   a a a a a a a a a a a a a a aa a a as
	   a a a a a a a a a a a a a a aa a a as
	   a a a a a a a a a a a a a a aa a a as
	   a a a a a a a a a a a a a a aa a a as
	   a a a a a a a a a a a a a a aa a a as
	   a a a a a a a a a a a a a a aa a a as
	   a a a a a a a a a a a a a a aa a a as
	   a a a a a a a a a a a a a a aa a a as))

   Delete one of the `a' lines from the above and it works.
   NOTE that it seems unlikely that this will fix itself when
   we get the new version of Lucid, as it seems to be an
   emacs problem.  It appears to have started with the big changeover.

   -- Bob

I wouldn't be surprised if this happens on the Suns also.  We had a
similar problem with the Scheme subprocess -- and the culprit turned
out to be that nearly every version of unix had bugs in their PTY
implementation.

The fix we used for Scheme was to bind `process-connection-type' to
nil when the subprocess was started, thus forcing the use of pipes for
communication.


Return-Path: 
From: rjh@ai.mit.edu (Robert J. Hall)
Date: Fri, 18 Aug 89 13:58:08 edt
To: cph
In-Reply-To: Chris Hanson's message of Fri, 18 Aug 89 13:36:49 edt <8908181736.AA13561@zurich.ai.mit.edu>
Subject: clisp-mode broken on HPs

Your fix to clisp mode worked great.  Thanks.

-- Bob