From xemacs-m  Tue Dec 17 13:26:54 1996
Received: from bert.amazon.com (bert.amazon.com [204.177.154.1]) by xemacs.cs.uiuc.edu (8.8.3/8.8.3) with ESMTP id NAA00626 for <xemacs-beta@xemacs.org>; Tue, 17 Dec 1996 13:26:54 -0600 (CST)
Received: from pc134.amazon.com (eb@pc134.amazon.com [204.177.154.134]) by bert.amazon.com (8.7.6/8.7.3) with SMTP id LAA22456 for <xemacs-beta@xemacs.org>; Tue, 17 Dec 1996 11:26:23 -0800 (PST)
Sender: eb@amazon.com
Message-ID: <32B6F3F1.54B53CF9@amazon.com>
Date: Tue, 17 Dec 1996 11:26:41 -0800
From: Eric Benson <eb@amazon.com>
Organization: Amazon.com Books
X-Mailer: Mozilla 3.0Gold (X11; I; Linux 2.0.18 i586)
MIME-Version: 1.0
To: xemacs-beta@xemacs.org
Subject: Subprocess problems in alpha-dec-osf4.0: Reward offered!
References: <yngengx3zlc.fsf@colargol.idb.hist.no>
		<32AEFD4D.485BA89D@amazon.com>
		<32AF1597.4303C54C@amazon.com> <9612171323.AA19337@fornet.gvc.dec.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Steve Carney wrote:
> 
> eb@amazon.com (Eric Benson) writes, in <32AF1597.4303C54C@amazon.com>:
> 
> >In addition to the assertion failure problem I described in my last
> >message, process_send_signal, or at least interrupt-process, doesn't
> >seem to work at all.
> 
> I had the same problems starting in 15b2 (I think) with DUNIX 3.2D.  I
> reported the crash with trace info, and there's been no fix as far as I
> know.
> 
> The dump seem to happen during subprocess creation or deletion.  From
> the stack info, I couldn't figure out what the problem was.

This problem has been around at least since 19.14.

If we don't find a fix for it, we're going to have to switch to FSF
Emacs here.

I believe I can convince management to pay a bounty to someone who fixes
these problems.  If you have access to Digital Unix 3.2 or 4.0, and have
the time and skill to debug this, please let me know.  The fix would be
available to everyone, of course.  There are three (possibly related)
problems that are currently making XEmacs almost unusable to us:

1. Signals are not being passed to subprocesses.  Control-C Control-C in
a shell buffer does nothing.

2. There's an intermittent assertion failure when starting subprocesses.

3. Control-G interrupts are being passed on to XEmacs's parent process
when not running in X. We have an application that execs XEmacs using
the system() library function.  This application acts as if XEmacs has
quit when you type Control-G.  However, the terminal is still in raw
mode and XEmacs is still running.  Needless to say, this makes it
completely unusable for this application in non-X mode.

Please let me know if you're willing and able to fix these things and
what kind of monetary incentive it would take to get you to fix them.  I
have a figure in mind myself, probably enough to pay for a good chunk of
that PC you've been wanting to buy to install Linux on!

Happy Holidays,
Eric Benson
eb@amazon.com

