cvs.gedasymbols.org/archives/browse.cgi   search  
Mail Archives: djgpp/1997/08/13/01:54:00

From: "A. Sinan Unur" <asu1 AT cornell DOT edu>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: [Q] expression evaluation order
Date: Thu, 07 Aug 1997 15:05:28 -0400
Organization: Cornell University
Lines: 118
Sender: asu1 AT cornell DOT edu (Verified)
Message-ID: <33EA1C78.4900@cornell.edu>
References: <5sbb1h$kvr$1 AT newton DOT pacific DOT net DOT sg> <33E98313 DOT 6250 AT pentek DOT com>
NNTP-Posting-Host: 128
Mime-Version: 1.0
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

Charles Krug wrote:
> 
> Victor wrote:
> >
> > Is it safe to assume that the expression
> >
> > if(first_function()==OK && second_function()==OK &&
> > third_function()==OK) return(OK);
> >
> > will be evaluated from left to right?

> No.
> 
> I think that ANSI tries to demand left-to-right evalutaion of &&.  But
> not all compilers do this correctly.  To guarantee left-to-right
> evaluation, you need:
> 
> if (first() == OK)      {
>         if (second() == OK)     {
>                 if (third() == OK)      {
>                         return (OK);
>                 }
>         }
> }
> return (!OK);

this is an abomination to say the least. why the braces?

> 
> This suggests a better (IMO) implementation that's a bit more clear:
> if (first() != OK)
>         return (!OK);
> else if (second != OK)
>         return (!OK);
> else if (third() != OK)
>         return (!OK);
> else
>         return (OK);

can you actually read this code if the blocks contain anything more 
complicated than return's. even if you can, can you tell me exactly which 
else belongs to which if without hesitation? also, what is the difference 
between this and:

if(first() != OK)
  return !OK;
if(second() != OK)
  return !OK;
...

if you have a logical expression that you want to test, just test it. the 
gotcha is that the evaluation of that expression stops once the value is 
known so parts of the expression may not be evaluated. so if you have 
code that depends on all the calls the first(), second() etc actually 
being made in the if statement, you'll be in trouble.

> Don't bother trying to save mythical machine cycles by placing all your
> logicals on the same line.  Remember that function calls are much more
> expensive than logicals in terms of processor time.  This is especially
> true of x86 processors, which are notoriously slow in terms of stack
> performance.

readable code is much more important than whatever cycles we are talking 
about 95% of the time. and to get rid of readability on the basis of a 
wrong premise is, well, plain bad advice.

> The professional in me would also like to see comments explaining what
> first(), second(), and third() do, just in case you put this project
> away for a year.  This is much easier to accomplish when they are on
> seperate lines.

the pro in you should read a book first.
 
> The best explanation of the rules of precedence I've ever heard:
> 
>         "Multiply and divide before you add and subtract"
>         "Put parentheses around everything else"

i suggest you forget about these rules of thumb and stick with what the 
language specification actually says. from the comp.lang.c faq by steve 
summit:

3.4:    Can I use explicit parentheses to force the order of evaluation
        I want?  Even if I don't, doesn't precedence dictate it?

A:      Not in general.

        Operator precedence and explicit parentheses impose only a
        partial ordering on the evaluation of an expression.  In the
        expression

                f() + g() * h()

        although we know that the multiplication will happen before the
        addition, there is no telling which of the three functions will
        be called first.

        When you need to ensure the order of subexpression evaluation,
        you may need to use explicit temporary variables and separate
        statements.

        References: K&R1 Sec. 2.12 p. 49, Sec. A.7 p. 185; K&R2
        Sec. 2.12 pp. 52-3, Sec. A.7 p. 200.


this is the kind of weirdness that always prompts me to say check the 
comp.lang.c faq (for this question 3.4, 3.5, 3.8 would provide the 
answer) and post your general c programming related questions to 
comp.lang.c.moderated once you have determined that the faq does not 
answer it. at least you can be reasonably certain of the quality of 
responses. with a couple of exceptions, most of the responses to the 
question were wrong. read the responses by Erik Max Francis and  
Hans-Bernhard Broeker and avoid the rest. next time, for your own 
benefit, ask your general c related questions in a newsgroup that is 
dedicated to such discussions. and for all the touchy people, this was 
_NOT_A_FLAME_!

  -- Sinan

- Raw text -


  webmaster     delorie software   privacy  
  Copyright © 2019   by DJ Delorie     Updated Jul 2019