In a couple days it will be the two week mark and I'll be posting a
summary of this conversation on the ticket. At this point I don't see
any disagreement, just questions regarding the completeness of the
patch (so I included many more functions from Numeric). The final
type signatures are:
efloat :: (RealFloat a) => Maybe Int -> a -> Doc
ffloat :: (RealFloat a) => Maybe Int -> a -> Doc
gfloat :: (RealFloat a) => Maybe Int -> a -> Doc
hex :: (Integral a) => a -> Doc
hexPad :: (Integral a) => a -> Int -> Doc
intAtBase :: (Integral a) => a -> (Int -> Char) -> a -> Doc
intAtBasePad :: (Integral a) => a -> (Int -> Char) -> a -> Int -> Doc
oct :: (Integral a) => a -> Doc
octPad :: (Integral a) => Maybe Int -> a -> Int -> Doc
Which follows the Numeric module pretty close (except the *Pad
functions, which are obvious devivations).
Thomas
On Mon, Mar 2, 2009 at 7:01 PM, Thomas DuBuisson
Padding floats could be touchy - perhaps it should be a separate patch by someone who would use such a feature and know what they want.
Thoughts?
Presumably it would be uncontroversial for the float ones to behave the same as the ones from the existing Numeric module.
Done, the new prettyNumeric.patch is with the ticket including hex, oct, ffloat, efloat, gfloat, signed, intByBase. Also, padded versions of hex, oct, intByBase.
Thomas