NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/lang/c++-part1/9098

Internet Message Format | 1996-08-05 | 1.4 KB

Path: news.tek.com!mdhost!gaulandm
From: [e-mail] (Michael Alan Gauland)
Newsgroups: comp.lang.c++
Subject: Can I overload "*" without class overhead?
Date: 28 Feb 1996 17:13:55 GMT
Organization: Tektronix, Inc, Beaverton, OR, USA
Sender: gaulandm@mdhost (Michael Alan Gauland)
Distribution: world
Message-ID: <4h22gj$[e-mail]>
NNTP-Posting-Host: hardtime.cse.tek.com

I need to implement some fixed-point math routines.  I've got a set of
16-bit data, with an implied decimal point in the middle of the number.
To multiply two such values requires a right shift so the decimal point
of the result is in the right place:

    result = ((int32)a * b) >> 8;

This looks sloppy, and every place in the code I put such an operation
increases the chances of getting it wrong.  It seems a natural place
to take advantage of C++'s operator overloading.  My attempts to date,
however, have been slow (though elegant).   I expect this is due to the
added overhead of a struct vs. an int.  Is there any way to work around
this?  Of course, I could use a fuction like "FixedPointMultiply(a,b)",
but I'll lose the added security of rigid typing.

Can anyone offer any helpful advice, or is this just something elegant C++ can't
do as efficiently as sloppy C?

Thanks,
Mike

---------------------------------------------------
Michael A. Gauland    [e-mail]
AA7JF            [phone]