Posted by fritz-bayer@web.de on 11/18/67 11:41

Tim Roberts wrote:
> "fritz-bayer@web.de" <fritz-bayer@web.de> wrote:
> >
> ><?php
> >$a = 5543039447;
> >$b = 2147483648;
> >$result = ($a & $b);
> >echo "$result\n" ;
> >?>
> >
> >to perl? In perl almost the exactly same code:
> >
> >#!/usr/bin/perl
> >$a = 5543039447;
> >$b = 2147483648;
> >$result = ($a & $b);
> >print "$result\n" ;
> >
> >returns 2147483648.
> >
> >How do I have to modify the perl code, so that it will actually do a
> >bitwise and? I don't get - any workaround? Somethings I'm not getting
> >here? I'm using linux debian/sarge with an intel 32 bit prozessor.
>
> The basic problem here is that what you are asking for is nonsense. GIGO.
>
> Both of your numbers are too large to fit into an unsigned 32-bit value.
> When that happens, both languages will automatically convert to using
> floating point to store those values. $a and $b are FLOATING POINT.
>
> However, bitwise operations do not make sense for floating point values.
> Thus, both languages must convert the floats back to 32-bit integers. But,
> we already know that the values are too big for that. So, there is an
> overflow. It just turns out that the two languages handle this overflow
> differently.
>
> PHP handles the overflow by throwing away the upper bits, leaving you with
> only the bottom 32 bits of the converted value. Perl handles the overflow
> by "pegging"; it returns 2^32-1, or 4,294,967,295. That value happens to
> have all 32 bit set, so the bitwise "and" returns the second value.
>
> If you need to handle integers larger than 32-bits in Perl, you can use the
> Math::BigInt module to do that.
> --
> - Tim Roberts, timr@probo.com
> Providenza & Boekelheide, Inc.

Thanks Tim! Any ideas how I could emulate the overflow, so that I get
the same overflow behaviour in perl, maybe by writting a function for
it?!

[Back to original message]