Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!mcvax!ukc!its63b!simon
From: simon@its63b.ed.ac.uk (ECSC68 S Brown CS)
Newsgroups: comp.bugs.4bsd,comp.unix.questions
Subject: $@ vs. $* (was - Re: nohup (from Bourne shell))
Message-ID: <164@its63b.ed.ac.uk>
Date: Sun, 7-Dec-86 10:36:52 EST
Article-I.D.: its63b.164
Posted: Sun Dec  7 10:36:52 1986
Date-Received: Mon, 8-Dec-86 02:10:31 EST
References: <181@haddock.UUCP> <106@quacky.UUCP>
Reply-To: simon@its63b.ed.ac.uk (ECSC68 S Brown CS)
Followup-To: comp.unix.questions
Distribution: world
Organization: I.T. School, Univ. of Edinburgh, U.K.
Lines: 34
Keywords: sh, positional parameters
Xref: mnetor comp.bugs.4bsd:65 comp.unix.questions:279

In article <106@quacky.UUCP> dce@quacky.UUCP (David Elliott) writes:
>
>There is one slight problem with "$@", which can be shown by the following
>
>	[... example of the wll-known "feature" ...]
>
>Anyway, the following statement can be used to fix this bug in most
>versions of sh.
>
>	if (strcmp (as, "\"$@\"") == 0 && dolc == 0) {
>		return(fixstak());
>	}
>
>Add these to the beginning of the subroutine macro() in macro.c, and the
>bug goes away. It is somewhat kludgy, but it works.
>

However, if you fix this in your version of sh, it makes any shell-scripts
that you may write assuming the "fixed" behaviour inherently non-portable,
'cos they'll immediately bomb out on standard shells.

So, until the fix described above (or some equivalent thereof) is official,
its probably better just to be very careful when using $* or $@ in scripts.
One foolproof way is to always use ${1+"$@"} in place of $* or $@ or "$@"
- ok, its ugly, I know... :-)


--
Simon Brown
Department of Computer Science, University of Edinburgh, Scotland.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Life's distressing - that's no blessing". [Anon.]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~