(re-) Add pow functionality for rational

[jeking3]

See: #24

In 2015 a pull request added pow() and reciprocal(). In 2018 it was reviewed by CMT and reciprocal() was found to be unnecessary. This left boost::pow() for rational, however it introduced an issue resolving std::pow from within boost namespaces. This happened close to the 1.68.0 release (during beta) so it was removed.

Originally recommend looking at adding operator ^= as a modifying operator to raise the rational value to an integer exponent. In addition operator ^ could take an existing rational and raise it to an exponent, returning a new rational. - note comments below.

[swatanabe]

AMDG

On 07/14/2018 09:22 AM, James E. King III wrote: <snip> Recommend looking at adding `operator ^=` as a modifying operator to raise the rational value to an integer exponent. In addition `operator ^` could take an existing rational and raise it to an exponent, returning a new rational.

I strongly disagree with this as operator^ means something completely different and has the wrong precedence. The correct name is pow, as long as the issues are dealt with. In Christ, Steven Watanabe

[jeking3]

I was thinking the commonly used mathematic notation of (x^2) would be reasonable here, however the precedence issue is interesting.

[jeking3]

I think perhaps just moving pow into another namespace (ideally boost::rational) would resolve the issue sufficiently and avoid the collision that was happening.

[swatanabe]

AMDG

On 08/01/2018 07:22 AM, James E. King III wrote: I think perhaps just moving pow into another namespace (ideally boost::rational)

boost::rational is a class template. The idea is sound in principle, but pow does need to be found by ADL. I suppose that you could just document that pow lives in an unspecified associated namespace of rational.

would resolve the issue sufficiently and avoid the collision that was happening.

In Christ, Steven Watanabe