Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!necntc!ima!haddock!karl
From: karl@haddock.ISC.COM (Karl Heuer)
Newsgroups: comp.unix.questions,comp.lang.c
Subject: Re: Wierd Compilers
Message-ID: <1635@haddock.ISC.COM>
Date: Fri, 6-Nov-87 14:28:01 EST
Article-I.D.: haddock.1635
Posted: Fri Nov  6 14:28:01 1987
Date-Received: Sun, 8-Nov-87 16:32:52 EST
References: <367@white.gcm> <18964@amdcad.AMD.COM> <44177@beno.seismo.CSS.GOV> <1988@killer.UUCP>
Reply-To: karl@haddock.ima.isc.com (Karl Heuer)
Followup-To: comp.lang.c
Organization: Interactive Systems, Boston
Lines: 25
Xref: mnetor comp.unix.questions:4826 comp.lang.c:5314

[Someone noted that "while (i--);" is slower than "while (--i);".  It was
pointed out that the former generates more instructions because of the extra
work to save the old value of the counter.  At least one person claimed that
the postdecrement operator should be avoided for this reason; apparently he
didn't notice that the value was being used, and that the two statements are
not semantically identical.  Rick Adams observed that, even if the value is
not being used, the user need not be concerned with this optimization.]

More than one person writes that an n-trip loop should be written
  ++n;  while (--n) ...;

I'm surprised that nobody has yet mentioned the idiom that I use:
  while (--n >= 0) ...;

This has the speed of predecrement, and avoids the kludge of having an extra
increment.  It requires n to be a signed type, but that's not usually a
problem.  (Also, it doesn't have identical semantics if n is negative, but
from the context I presume that this is not a concern.)

Of course, the other way is to write
  do ...; while (--n != 0);    [or]    do ...; while (--n > 0);
assuming n is known to be initially nonzero.

Karl W. Z. Heuer (ima!haddock!karl or karl@haddock.isc.com), The Walking Lint
Followups to comp.lang.c only.