$x = Math::BigInt->bstr( $bigint );
$bits = int( log( $x ) / log( 2 ) ) + 1;
[download]

I'm using whatever BigInt came with ActivePerl... It's likely not the latest. It does occur to me that I can probably do something like:

$bits=int((length($bigint)-1) * 3.3219)+1;

where 3.3219 is log(10)/log(2) and the length of $bigint is relatively large assuming that length($bigint) returns the length of the string / ie number of decimals.

It also occurred to me that I could do a manual conversion to float by something like this:

$float=substr($bigint,1,1).'.'.substr($bigint,2).'e'.(length($bigint)-2);

where the point is to create a float string value out of the bigint string value...

Either of these might work, but they're a bit ugly...

j

As long as you have a fairly recent ActivePerl, you should be ok. I think only relatively old versions of Math::BigInt did weird things sometimes with bstr(). The newest version of Math::BigInt installs like a charm on a Win32 system, btw. So either way, you might be better off installing the newest version.
___________________
Kurt

You can get what you want directly, without log(), using some bit-twiddling:

my $the_num = Math::BigInt->new(1234567890123545678);
sub numbits {
    my $tester = Math::BigInt->new(shift);
    my $i = 0;
    do {$i++} while ($tester >>= 1);
    $i;
}
print numbits($the_num);
[download]

After Compline,
Zaxo

Hmm... I like the idea. It replaces a series of complex string operations with a simple loop. I'll have to run some benchmarks to see which is faster. But I do like this approach.