Yes, that is similar to what the Decimal library does. It maintains a precision value and a scaled integer.

So € 10.02 could be represented as:

Decimal 2 1002

It also has nice functions for dividing out currency into equal parts, or ratios, while preserving the property that the total of the parts should add up to the original value.

I just thought it might be interesting to encode the precision in the type, and just use an integer in the representation. Then it's even more efficient at runtime. Maybe I'll write my own library. Or perhaps there's another one that does that already.

The Data.Number.Fixed library encodes the precision in the type, but it uses rational number for the values, and only guarantees the precision within an epsilon range, so that "two digits of precision" means it could be +/- .01, which is not good for currency.

On Sat, Mar 5, 2016 at 12:26 AM, Imants Cekusins <imantc@gmail.com> wrote:

> I prefer the Decimal library for currency.

here is another option to consider:

data Cur = Eur | ...
data Money = Mint Cur Int | Mtup Cur (Int,Int)

precision::Cur -> Int
precision Eur = 2
precision _ = 2

€ 10.02 would be:
Mint Eur 1002
Mtup Eur (10,2)

This adds overhead to computations. However precision and correct currency is guaranteed.


_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe