--- Manfred Doubar wrote:

BUT I would expect that both integer and fractional _parts *included* the sign - and I tend to think anything else would inevitably lead to confusion.

I set up the current semantics of is_negative, integral_part, and fractional_part to match the semantics of the corresponding typedefs in mixed_number_impl. BTW, integral_part used to be called whole_part because it returned whole numbers, which were were nonnegative.

I have a good hint of why you might have gone down this path,

Now you know exactly why.

but I can only think that its going to be cause for error.

No, the difference between our definitions of integral_part and fractional_part will be the cause for error. It turns out that Mathematica <http://mathworld.wolfram.com/> agrees with you, so I'll go ahead and change the metafunction semantics. Cromwell D. Enage __________________________________ Yahoo! Mail - PC Magazine Editors' Choice 2005 http://mail.yahoo.com