2009/3/9 Noah Slater <nslater@apache.org>:
> My company produce price feeds using decimals, and we need those decimals to be
> decimals. For that reason, and after much messing about and research, I decided
> that we should use strings for all non-int numerical values. I would be
> supportive of a canonicalisation format that followed this rationale.
I've had good results using a scheme where decimals are made up of
pairs of ints, the first representing a coefficient and the second a
power of 10. Basically calculator e notation, but with an integer
instead of a float, adjusted accordingly.
--
Kerr
|