Xref: utzoo gnu.emacs:657 comp.emacs:5676
Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!bbn!jr@bbn.com
From: jr@bbn.com (John Robinson)
Newsgroups: gnu.emacs,comp.emacs
Subject: Re: GNU Emacs hands under BSD 4.3 (help)
Summary: the well-known, easily-fixed setpgrp() bug
Keywords: GNU, BSD, Emacs, Vax
Message-ID: <37603@bbn.COM>
Date: 22 Mar 89 18:25:15 GMT
References: <10897@lanl.gov> <1324@cunixc.cc.columbia.edu>
Sender: news@bbn.COM
Reply-To: jr@bbn.com (John Robinson)
Followup-To: gnu.emacs
Organization: BBN Systems and Technologies Corporation, Cambridge MA
Lines: 52
In-reply-to: sethr@cunixc.cc.columbia.edu (Seth Robertson)

In article <1324@cunixc.cc.columbia.edu>, sethr@cunixc (Seth Robertson) writes:
>In addition, it might well be a good idea if Richard Stallman
>or some Important Person at MIT/UUNET/etc could modify the code
>and redistribute the bug-free code to all points of interest.
>(18.53 1/2)

Well, gee, I'm not very important or anything, but do you think anyone
would mind if I distributed the patch?  It's purty small, so I reckon
there's no good to sending a new copy of the whole file nowhere.

While I'm here, let me add the ChangLog entry that goes with this
change.  It looks like it was put in with good intentions, but without
knowledge of the grim side-effects it would have on shell scripts or
other programs that run emacs in inferior forks.  I would never do
such a thing - rather sit in emacs all day long...

 Sat Feb 18 09:03:52 1989  Richard Stallman  (rms at sugar-bombs.ai.mit.edu)

	 * emacs.c (main) [BSD]: Do setpgrp.  May avoid some Unix signal bugs.

Here's the change (only one made in emacs.c, so you can undo this by
putting your release 18.52 copy of emacs.c into your 18.53
distribution area):

diff -rc2N dist-18.52/src/emacs.c dist-18.53/src/emacs.c
*** dist-18.52/src/emacs.c	Fri Aug 26 18:31:50 1988
--- dist-18.53/src/emacs.c	Sat Feb 18 09:03:50 1989
***************
*** 225,228 ****
--- 225,233 ----
  #endif HIGHPRI
  
+ #ifdef BSD
+   /* interrupt_input has trouble if we aren't in a separate process group.  */
+   setpgrp (getpid (), getpid ());
+ #endif
+ 
    inhibit_window_system = 0;
  
So it seems that some systems that use interrupt_input will behave
worse without this setpgrp() call, so there should be some folks out
there would prefer to leave this code in.  None as much as owned up
yet, though.

Just run that feller through patch and remake yer emacs, and this
shellscript problem'll be fixed.

Shucks.
--
/jr
jr@bbn.com or bbn!jr
C'mon big money!