How can I convert between Float and [Word8]?

Question

I want to convert Float to a list of bytes [Word8] (and vice versa). I see there is a Storable class which could maybe be used for this, but I'd like to avoid using IO monad as this has nothing to do with IO.

Answer 1

What is it that you want? If you want to convert between the Float and its actual bit representation, you can use data-binary-ieee754 (uses Foreign.Storable and unsafePerformIO under the hood) or cereal-ieee754. The latter doesn't use Storable or IO, it writes the value to an STUArray, casts the array and reads a value of the other type. Both packages give you a conversion Float <-> Word32 (or Double <-> Word64), converting the WordN to [Word8] is easy.

Answer 2

I recommend the Data.Binary library.

See here:

Prelude Data.Binary> encode (13.7 :: Double)
Chunk "\SOH\SOH\NUL\NUL\NUL\NUL\NUL\NUL\NUL\affffff\ESC\255\255\255\255\255\255\255\207" Empty
Prelude Data.Binary Data.ByteString.Lazy> Data.ByteString.Lazy.unpack $ encode (13.7 :: Double)
[1,1,0,0,0,0,0,0,0,7,102,102,102,102,102,102,27,255,255,255,255,255,255,255,207]

If you're wondering why the representation is that big, it's because of the implementation of the Binary instance for Float and Double uses decodeFloat function:

decodeFloat :: RealFloat a => a -> (Integer, Int)

This is done to keep the implementation as generic as the Haskell Report (which doesn't specify IEEE754 or anything like that).

To get the actual binary representation of a Float, you need to use Storable AFAIK.