Pitch alterations use SCM rather than flower rationals

Pitch alterations use SCM rather than flower rationals

This is a first step for replacing the Rational data type with the
built-in rationals of GUILE.


Also contains commits:

Don't call Pitch::Pitch with more arguments than necessary.


Pitch::Pitch (int, int) is declared but not defined

This caters for the definition by making the last argument of
Pitch::Pitch (int, int, Rational) a default argument.

[dak, 2014-06-24 19:54:11]

Hopefully fix ambitus

[Devon Schudy, 2014-06-27 04:06:40]

Using Guile rationals is good on principle, but it's not great for readability.
It makes arithmetic and comparisons much more verbose, and type declarations
less informative. This isn't much of a problem for pitch alterations (since it's
balanced by the reduced hassle of converting between the two representations),
but it would be if all of Lilypond's rational arithmetic were converted to Guile
rationals. Is there a way to make this less awkward?

We can't just overload the arithmetic and comparison operators to work on Scheme
numbers, because SCM is a pointer type so it already has them.

Rational could become a wrapper around a Guile rational, but that's not much
simpler than the status quo.

Convenience functions for things like scm_is_true(scm_less_p(...)) would make
comparisons a little less verbose (and reduce the risk of forgetting
scm_is_true).

[lemzwerg, 2014-06-27 04:21:45]

Or maybe some cpp macro trickery improves the legibility.