From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from proxy4.ba.best.com ([206.184.139.15]) by hawkwind.utcs.utoronto.ca with SMTP id <24785>; Fri, 10 Dec 1999 03:36:56 -0500 Received: from peanut.rakitzis.com (dynamic24.pm03.san-mateo.best.com [205.149.174.152]) by proxy4.ba.best.com (8.9.3/8.9.2/best.out) with ESMTP id XAA06330; Thu, 9 Dec 1999 23:45:42 -0800 (PST) Received: (from byron@localhost) by peanut.rakitzis.com (8.8.7/8.8.7) id XAA29345; Thu, 9 Dec 1999 23:45:04 -0800 Date: Fri, 10 Dec 1999 02:45:04 -0500 From: Byron Rakitzis Message-Id: <199912100745.XAA29345@peanut.rakitzis.com> To: fosterd@hartwick.edu, rc@hawkwind.utcs.toronto.edu Subject: Re: tilde featuritis alert >It seems to me that a simple rule would suffice here. An '=' in the >first word of a command is interpreted as a metacharacter. An '=' in >the rest of it is not. Oh, but there is the issue of 'foo = bar' (I >find myself trying to do that frequently in sh now.) OK, an unquoted >'=' is a metacharacter iff (it's part of the first word || it's the >first character in the second word). That should continue to allow >'foo =bar' as well. For better or for worse, rc's grammar allows for multiple temporary assignments: ; foo=1 bar=2 echo $foo $bar 1 2 ; echo $foo $bar ; So you have to account for that as well. With sufficient tweaking to the parser and lexer perhaps you could allow unquoted '='s in a parameter list, but I tried my hand at it years ago and I wasn't able to come up with anything remotely pleasant. Byron.