Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rutgers!im4u!milano!mcc-pp!tiemann
From: tiemann@mcc-pp.UUCP (Michael Tiemann)
Newsgroups: comp.lang.c
Subject: Re: Finding MSB in bit string
Message-ID: <2040@mcc-pp.UUCP>
Date: Fri, 14-Nov-86 00:16:20 EST
Article-I.D.: mcc-pp.2040
Posted: Fri Nov 14 00:16:20 1986
Date-Received: Fri, 14-Nov-86 06:07:51 EST
References: <800@oakhill.UUCP> <11053@cca.UUCP> <7143@boring.mcvax.UUCP>
Organization: MCC, Austin, TX
Lines: 30
Summary: Saving Guy the trouble

In article <7143@boring.mcvax.UUCP>, guido@mcvax.uucp (Guido van Rossum) writes:
> >What's a good general method?  How would a real programmer do it?
> 
> This has been discussed in this group a while ago.  It goes something like
> 
> 	return count[x&0xff] + count[(x>>8)&0xff] +
> 		count[(x>>16)&0xff] + count[(x>>32)&0xff;
						^^
						24
> 
> (I hope this time I didn't forget some offsets like in my solution of
> the MSB problem.)

Nope, just some arithmetic.

How about

	cnt = count[x&0xff];
	cnt += count[(x>>=8) & 0xff];
	cnt += count[(x>>=8) & 0xff];
	return	cnt + count[x&0xff];

for those poor souls without barrel shifters.

But this is not a *general* method, since there is no parameterization
of the bit-string (!). Not all bit-strings are 32-bit ints. I won't
*even* touch the issue of bit-string alignment...

Michael
tiemann@mcc.com