Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!iuvax!cica!ctrsol!ginosko!uunet!amgraf!cpsolv!rhg
From: rhg@cpsolv.UUCP (Richard H. Gumpertz)
Newsgroups: comp.lang.c
Subject: Re: Obfuscated SWAP: not portable!
Message-ID: <173@cpsolv.UUCP>
Date: 8 Sep 89 15:47:31 GMT
References: <784@skye.ed.ac.uk> <1267@levels.sait.edu.au> <149@cpsolv.UUCP> <1400@levels.sait.edu.au>
Reply-To: rhg@cpsolv.uucp (Richard H. Gumpertz)
Organization: Computer Problem Solving, Leawood, Kansas
Lines: 11

"x ^= y ^= x ^= y" differs from "x = y = z = 0" in that the former makes TWO
assignments to x.  Side effects (i.e., assignments to variables) need only be
completed at "sequence points"; between sequence points, any order of
evaluation (including assignment operators) is allowed!

The following are all legal interpretations of the sequence in question,
yet may yield different results:
      (t = x ^= y, x ^= y ^= t)
      (t = x, y ^= x ^= y, x = t ^ y)
      (t = x ^ y, x ^= y ^= t, x = t)
I am sure there are other interpretations as well.