...

* What is the type of boost::multiprecision::number_category<B>::type for all the provided backends? Could the specialization for boost::multiprecision::number_category<B>::type be added in the documentation of each backend? and why not add also B::signed_types, B::unsigned_types, B::float_types, B::exponent_type?

OK.

On second thoughts, the members of B::signed_types etc are basically implementation details - optimizations if you will. Rather than listing the types explicitly I'd prefer some more weasel wording along the lines of "This type will interoperate natively with any signed or unsigned built in integer type no wider than a long - this means for example that multiplying by a integer literal is more efficient than constructing an mpz_int temporary from the integer and then multiplying. Likewise for the other arithmetic operations." John.