From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from localhost by hawkwind.utcs.toronto.edu with SMTP id <2773>; Fri, 18 Jun 1993 13:33:11 -0400
Return-Path: TechFak.Uni-Bielefeld.DE!malte
Received: from techfac.TechFak.Uni-Bielefeld.DE ([129.70.132.100]) by hawkwind.utcs.toronto.edu with SMTP id <2771>; Fri, 18 Jun 1993 03:32:38 -0400
Received: from dahlie.techfak.uni-bielefeld.de (idefix.TechFak.Uni-Bielefeld.DE)
	by techfac.TechFak.Uni-Bielefeld.DE id AA21169; Fri, 18 Jun 1993 09:31:54 +0200
Received: by dahlie.techfak.uni-bielefeld.de (5.0/tp.29.0890)
	id AA00685; Fri, 18 Jun 93 09:31:53 +0200
Date:	Fri, 18 Jun 1993 03:31:53 -0400
From:	malte@TechFak.Uni-Bielefeld.DE
Message-Id: <9306180731.AA00685@dahlie.techfak.uni-bielefeld.de>
To:	rc-owner
Subject: trace in rc
Resent-To: rc
Resent-Date: Fri, 18 Jun 1993 13:33:08 -0400
Resent-From: Chris Siebenmann <cks>
Resent-Message-Id: <93Jun18.133311edt.2773@hawkwind.utcs.toronto.edu>

Hello,

I've been thinking about how to implement a kind of "intelligent"
trace facility in rc. Intelligent means I want to execute a function
each time a command fails. One parameter to this function should be
the name of the failed command. I thought about the prompt function,
but that doesn't help in shell scripts.

So I'd like to hear what you think about adding a function to rc
which is called after invocation of any command or builtin, running
interactively or not with the last command's name ( or names if it was a
pipe ) as parameter. Then, fn prompt can be defined in terms of this
function, as well as any trace function.

Malte