Allowing automatic coercion to and from Decimal sounds dangerous and 
complicated to me.  Mightn't it be safer just to disallow this (at least for 
now)?

I think something like trim()  (find the closest rational approximation with 
denominator bounded by a given integer) would be useful to have as a Rational 
method, but probably only for explicit use.

I'm still worried about equality tests:  is it possible to give a good reason 
why Decimal("2.5") == Rational(5, 2) should return False, while Rational(5, 2) 
== 2.5 returns True.  Can someone articulate some workable principle that 
dictates when an int/float/complex/Rational/Decimal instance compares equal to 
some other int/float/complex/Rational/Decimal instance of possibly different 
type but the same numerical value?