The bytestring builder API has a large number of entry points for encoding numbers:

      int8Dec
    , int16Dec
    , int32Dec
    , int64Dec
    , intDec
    , integerDec

    , word8Dec
    , word16Dec
    , word32Dec
    , word64Dec
    , wordDec

Contrast this with the text library:

    decimal :: Integral a => a -> Builder

I want to deprecate the 11 *Dec functions in favour of the same single-entry-point API used by text: it's way simpler, and no less efficient.

(Really, the *Hex functions need the same treatment, but they're less of a priority to me.)