CODEWITHSUNDEEP
haskelltype-inference
user328062
user328062

Reputation:

Why would Haskell infer a specific type (apparently) inconsistently?

I'm in the process of writing a toy (physics) vector library and am having issues with GHC insisting functions should have Integers in their type. I want vectors to multiply vectors, as well as scalars (just using *), and while this was possible by just having Vector instance Num I'm now getting errors of the sort:

Couldn't match expected type `Integer' with actual type `Double'

After playing around with the code in order to pin down the problem I've gotten it down to this:

data V a = V a a a deriving (Show, Eq, Functor)
scale a (V i j k) = V (a*i) (a*j) (a*k)
(<.>) = scale

Now if we ask GHCi what type these are we get:

>:t scale
scale :: Num a => a -> V a -> V a
>:t (<.>)
(<.>) :: Integer -> V Integer -> V Integer

Where we certainly don't want <.> only acting on Integers. While this is fixable here by including a type signature for <.>, I would like to know what's actually going on.

Upvotes: 7

Views: 204

Answers (1)

rkhayrov
rkhayrov

Reputation: 10260

You are running into the infamous monomorphism restriction. Another solution would be to specify arguments explicitly:

a <.> v = scale a v

Or add -XNoMonomorphismRestriction pragma.

Upvotes: 15

Related Questions