use ShortByteString to hold zeros and named 'primBBounded word64Dec'

Data/ByteString/Builder/RealFloat.hs

@@ -70,12 +70,12 @@ module Data.ByteString.Builder.RealFloat
  , generic
  ) where
 
-import Data.ByteString.Builder.Internal (Builder, byteString)
+import Data.ByteString.Builder.Internal (Builder, shortByteString)
 import qualified Data.ByteString.Builder.RealFloat.Internal as R
-import qualified Data.ByteString.Char8 as BC
 import qualified Data.ByteString.Builder.RealFloat.F2S as RF
 import qualified Data.ByteString.Builder.RealFloat.D2S as RD
 import qualified Data.ByteString.Builder.Prim as BP
+import qualified Data.ByteString.Short as BS
 import GHC.Float (roundTo)
 import GHC.Word (Word64)
 import GHC.Show (intToDigit)
@@ -265,18 +265,18 @@ showStandard m e prec =
  | e <= 0
  -> string7 "0."
  <> zeros (-e)
- <> BP.primBounded BP.word64Dec m
+ <> buildDigits m
  | e >= olength
- -> BP.primBounded BP.word64Dec m
+ -> buildDigits m
  <> zeros (e - olength)
  <> string7 ".0"
  | otherwise -> let
  wholeDigits = m `div` (10 ^ (olength - e))
  fractDigits = m `mod` (10 ^ (olength - e))
- in BP.primBounded BP.word64Dec wholeDigits
+ in buildDigits wholeDigits
  <> char7 '.'
  <> zeros (olength - e - R.decimalLength17 fractDigits)
- <> BP.primBounded BP.word64Dec fractDigits
+ <> buildDigits fractDigits
  Just p
  | e >= 0 ->
  let (ei, is') = roundTo 10 (p' + e) ds
@@ -296,5 +296,6 @@ showStandard m e prec =
  ds = digits m
  digitsToBuilder = fmap (char7 . intToDigit)
 
- zeros n = byteString $ BC.take n $ BC.replicate 308 '0'
+ zeros n = shortByteString $ BS.take n $ BS.replicate 308 48
  olength = R.decimalLength17 m
+ buildDigits = BP.primBounded BP.word64Dec