[comp.std.c] for

karl@haddock.ima.isc.com (Karl Heuer) (05/25/90)

In article <5897@buengc.BU.EDU> bph@buengc.bu.edu (Blair P. Houghton) writes:
>If I were to write a compiler under K&R1 and provide a no-optimization mode,
>`for(;;)' would be an unconditional branch and `while(1)' would compare 1 to
>0 and then branch-if-unequal, because that's what the syntax implies.

I suppose you would also compile `if (x==y)' into code that compares x against
y, sets a temporary to either 0 or 1, and then compares that value against
zero, right?

Your statements are vacuously true, because according to your definitions,
nobody has ever written a "non-optimizing" compiler.

Karl W. Z. Heuer (karl@ima.ima.isc.com or harvard!ima!karl), The Walking Lint

diamond@tkou02.enet.dec.com (diamond@tkovoa) (05/25/90)

How do we know if they're a draw?  All you people who are arguing
which is better:  one of you code an infinite loop with for(;;)
and one of you code an infinite loop with while(1).  When they
finish, wake me (from the dead) and tell me.

The rest of us can go RTFS.  TFS says that both forms are legal.
This is the only answer that belongs in ThisFNewsgroup.

Some of us will even use optimizers when we can get our hands on them.
Though this does not belong in ThisFNewsgroup.

-- 
Norman Diamond, Nihon DEC     diamond@tkou02.enet.dec.com
Proposed group comp.networks.load-reduction:  send your "yes" vote to /dev/null.

exspes@gdr.bath.ac.uk (P E Smee) (05/25/90)

In article <12971@smoke.BRL.MIL> gwyn@smoke.BRL.MIL (Doug Gwyn) writes:
>If a compiler is going
>to generate a useless comparison, it is in fact more likely
>to do so for while(1) than for for(;;).

One benefit of 'for(;;)' over 'while (1)' which I haven't seen
mentioned is that most (in my experience) versions of 'lint' will
cheerfully accept 'for(;;)' without comment, while giving a 'constant
in conditional' warning for 'while(1)'.

You could plausibly argue that lint is broken and should recognize this
use of while as a special case.  You could also plausibly argue that C
should have an explicit 'do forever' construct.  However, neither of
these seem to be true, and I have no control over them; I can control
my code and like it to be lint-free (or as near as possible -- 'There
are some things you just cant get lint to shut up about' from lint(1)).

(Yeah, I know how to hide things from lint.  I don't like to do that,
though, for fear that I'll make a REAL error which it would have
caught, somewhere where it can't see it because I've hidden it.)

-- 
Paul Smee, Computing Service, University of Bristol, Bristol BS8 1UD, UK
 P.Smee@bristol.ac.uk - ..!uunet!ukc!bsmail!p.smee - Tel +44 272 303132

aipdc@castle.ed.ac.uk (Paul D. Crowley) (05/25/90)

I agree: this thread properly belongs in alt.religion.computers.
-- 
\/ o\ Paul Crowley aipdc@uk.ac.ed.castle
/\__/ "Trust me, I know what I'm doing" - Sledge Hammer

jsdy@hadron.COM (Joseph S. D. Yao) (05/25/90)

I just lied to 'rn'.  The following, like >75% of what preceded it,
isn't really worth sending or reading, but I thought I'd have my say.

When I was a beginning programmer, I thought the following was cute, so
I used it all over:
	#define ever	(;;)
	...
		for ever {
			...
		}

For that matter, I also define'd:
	#define until(x)	while (!(x))
	#define unless(x)	if (!(x))
to answer people who said "it doesn't have an "UNTIL" keyword, so what
good is it?"

Also, if our universe of discourse IS ENGLISH, then while(1) makes no
sense, because after the subordinate conjuction "while", one expects a
predicate, not a noun or noun phrase.  IN C, however, we all know that
predicates eventually evaluate to the-equivalent-of 1 or 0 ... but I
still teach people that, if they're going to do this kind of thing,
they should define TRUE and use while (TRUE) instead, it makes more
"intuitive sense", whatever that means (different for each of our
intuitions).

NOW, AS I SAID, ALL THIS MAKES NO DIFFERENCE WHATSOEVER.  It has been
shown that while(1) [or, better, while (TRUE)] and for (;;) [or, worse,
for ever] are logically equivalent by the definitions in both K&R and
X3J11.  The purposes of coding a program are (1) to communicate to the
computer what you intend for it to do, and (2) to communicate to other
people working on the same program (now or later) what you intend for
the computer to do.  [The latter is necessary, otherwise we'd compile
programs and throw the source away.  I HOPE you don't do that!]  BOTH
of these constructs do that.  Which is preferable is largely a cultural
thing.

On the other hand, I still prefer for (;;) ...		;-)

	Joe Yao				jsdy@hadron.COM
	( jsdy%hadron.COM@{uunet.UU.NET,decuac.DEC.COM} )
	arc,arinc,att,avatar,blkcat,cos,decuac,\
	dtix,ecogong,grebyn,inco,insight,kcwc,  \
	lepton,lsw,netex,netxcom,phw5,research,  >!hadron!jsdy
	rlgvax,seismo,sms,smsdpg,sundc,telenet, /
	uunet				       /
(Last I counted ...)

steve@taumet.COM (Stephen Clamage) (05/25/90)

In article <Z4O3P:2@xds13.ferranti.com> peter@ficc.ferranti.com (Peter da Silva) writes:
>... [this would mean that]
>the compiler can legally generate any code it likes for anything, so long
>as the semantics match the abstract machine. The compiler is permitted
>to generate any code it likes ...

Yup.  That's exactly what the ANSI standard says.  The standard makes
heavy use of the "as-if" rule for this reason.  It doesn't tell the
compiler implementor how to do the job, only what the result must be.
As in your hyperbolic example, a conforming compiler could send mail
to Dennis Ritchie (or to Bob Hope, for that matter) to get a value for
an expression whose semantics are undefined.
-- 

Steve Clamage, TauMetric Corp, steve@taumet.com