Suppose I have 64-bit numbers (some signed, some unsigned) in big-endian format. If I know that the actual value fits in 52 bits, I can use floating-point "doubles" to do the real work.
So, what's the easiest/fastest/best way to "unpack" the 8-byte value into a Perl scalar that holds a normal floating point number?
Ideas:
unpack two L's and multiply them out.
arrange the bits into the mantissa of a floating point value, and unpack with "d". Anybody know the binary format of the normal IEEE 64-bit double format?
Convince someone to add quad support to ActiveState Perl for Windows.