NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/lang/c-part1/2927

Internet Message Format | 1996-08-05 | 3.6 KB

Path: newshost.lanl.gov!tanmoy
From: [e-mail] (Tanmoy Bhattacharya)
Newsgroups: comp.lang.c
Subject: Re: Floating point calculation order
Date: 25 Jan 1996 01:10:45 GMT
Organization: Los Alamos National Laboratory
Message-ID: <[e-mail]>
References: <[e-mail]> <[e-mail]>
    <[e-mail]> <[e-mail]>
NNTP-Posting-Host: qcd.lanl.gov
Mime-Version: 1.0
Content-Type: text
In-reply-to: [e-mail]'s message of Wed, 24 Jan 1996 21:14:36 GMT

--text follows this line--
In article <[e-mail]> [e-mail] (Tom
Sanders) writes:
<snip>
   |> I am almost certain you write `a + b * c' contrary to your claim. Do
   |> you mean `a + (b * c)' or `(a + b) * c' when, if, you do it? If you do
   |> not, you are in a very small minority.
<snip>
   Sorry Tanmoy you'd be wrong, I do write `a + (b * c)' not `a + b * c' .

I see. I had not expected such an extreme position: but de gustabis
non est disputandum. I am used to writing long expressions, and after
a point parentheses obscure rather than reveal.

   Do I need to , no of course not.  I prefer to for clarity.  If I specify
   exactly what I want when I write it, there is little doubt in the future
   if I or someone else is debugging a problem.  If I leave the parentheses
   out, I or someone else may wonder whether it was intended to be
   `(a + b) * c' and waste time in a useless direction.

I must admit that I sometimes use space and alignment for these
purposes, but very rarely use parentheses for this purpose. I sometime
do use parentheses when evaluation order does not matter and I want to
stress the logical unity of a piece of the expression: even that is
rare though.

   As to the original equation `p * q / r' , I would argue that parentheses
   are certainly necessary.  There is a slight numeric difference between
   `(p * q) / r' and `p * (q / r)' .  By specifying I am sure the order I
   prefer will be used no matter what compiler I am using.

Even in non-ANSI C, there was no guarantee that the compiler will
respect your parentheses. However, if you meant not `any' compiler:
but any of a number of compilers, at least one of which is
non-conformant, that you are likely to work on, I agree with you.

When I write C, I assume ANSI conformance for parts which are not
time-critical. Parts which are time-critical, I have to code afresh
for every machine anyway: this, at least so I hope, is not the
experience of everyone.

In ANSI C, a compiler is just plain wrong if it treats p * q / r as p
* (q/r), if there is a difference between them (yes, I mean if
numerically, there is a difference between them). The only thing ANSI
C permits is that intermediate results may be help at higher precision
than stored results: no reordering is allowed if it is visible by the
program. The rules in FORTRAN (which I use more often for numerical
work) are different.

In any case, such stylistic issues can become religious. Some of my
styles, I am sure, are religiously hated by other people. I see what
you are saying, I think; but I do not agree with that style.

Cheers
Tanmoy
--
[e-mail](128.165.23.46) DECNET: BETA::"[e-mail]"(1.218=1242)
Tanmoy Bhattacharya O:T-8(MS B285)LANL,NM87545 H:#9,3000,Trinity Drive,NM87544
Others see <gopher://yaleinfo.yale.edu:7700/00/Internet-People/internet-mail>,
<http://alpha.acast.nova.edu/cgi-bin/inmgq.pl>or<ftp://csd4.csd.uwm.edu/pub/
internetwork-mail-guide>. -- <http://nqcd.lanl.gov/people/tanmoy/tanmoy.html>
fax: [phone]   voice: 1 [phone]    [ Home: 1 [phone] ]