I really like the approach Perl 6 takes with FatRat[1]. You basically have an object which holds a numerator and denominator, so all arithmetic calculations do not loose precision.
Having access to rational numbers is great and there are times they are incredibly useful, but I don't think it's a good default.
This means that the denominator won't get arbitrarily large.
For serious uses of rationals you need to have control over when the top and bottom are divided by the GCD. For currency you want to keep the denominator at 100. 50/100 wants to remain as that and not get simplified to 1/2.
The link to the scheme documentation that goes over this is quite good.
Honestly, most all older manuals and documents that were not related to systems programming seem to cover numeric programming decently enough. I get the impression that languages used to worry more about that sort of thing, but then the truly heavy lifting of numeric programming moved into the specialized domains of Matlab and friends, and then programmers slowly unlearned most of what was learned regarding numbers.