Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!cmcl2!rutgers!bellcore!faline!ulysses!sfmag!sfsup!mpl
From: mpl@sfsup.UUCP (M.P.Lindner)
Newsgroups: comp.lang.c
Subject: Re: typeof isn't enough to define swap correctly
Message-ID: <2020@sfsup.UUCP>
Date: Wed, 9-Sep-87 20:19:32 EDT
Article-I.D.: sfsup.2020
Posted: Wed Sep  9 20:19:32 1987
Date-Received: Sat, 12-Sep-87 00:37:02 EDT
References: <557@rocky.STANFORD.EDU> <1880@sol.ARPA> <109@umigw.MIAMI.EDU> <2019@sfsup.UUCP>
Organization: AT&T-IS, Summit N.J. USA
Lines: 17
Summary: oops


OK, as I read further I discovered people were complaining about not handling
name clashes, arrays, and register variables.  My method won't handle register
variables (can't take the address of a register), but it WILL handle arrays,
and as for name clashes, might I suggest the following kludge based on a trick
used by C++ to avoid name clashes (no, I wouldn't really do this, any more than
I would really use that swap routine):

#define swap(a,b) { \
	int _unlikely_string_/**/a/**/b; \
	etc.

blech, I have a bad taste in my mouth.  Note that this scheme (and much of
C++) fails if the arguments to the macro have space around them (ie
swap(x, y) instead of swap(x,y) ).

Mike (feeling sick now) Lindner