[Haskell-cafe] computational overhead of the "data" declarations

[Don Stewart]

eford:
> > So, to answer your question, the only computational overhead with a
> > data declaration is the extra memory and time to store and process the
> > constructor tags.  It usually isn't noticeable, but sometimes the
> > difference can be important.
> 
> Is there also potentially overhead due to laziness?  I mean, in this
> case:
> 
> 	newtype Fast = Fast Int
> 
> If you have a Fast value, you know you've got a real Int, not a thunk,
> right?  But here:
> 
> 	data Slow = Slow Int
> 
> Doesn't a Slow value remain unevaluated until you pattern match out the
> Int?

Also, remember that GHC is an optimising, strictness-analysing compiler,
making 'data' cheap or non-existent. Tags are also represented as bits in
pointers to heap values, so 'tags' are very cheap, if they're still needed
(e.g. when testing on Just/Nothing).

E.g.

    Slow = Slow { unSlow :: Int }
        deriving Show

    bigger :: Slow -> Slow
    bigger (Slow n) = Slow (n * 52)

    small = Slow 7

    main = print (unSlow $ bigger small)

Is compiled too:

    lvl :: String
    lvl = $wshowSignedInt 0 364 ([] @ Char)

So there's no runtime overhead.

-- Don