Create instance for list of data types - Haskell

Question

I feel like this should be really easy, but i am stuck at this point for quite some time.

Is it possible in haskell to create an instance of a type class for a list of certain data types?

What i want to achieve is the following.

class Rename a where
    findSub :: a -> a

-- A 'normal' instance would look like this
instance Rename Atom where
    findSub ......

-- Now i want to acchieve something like this
instance Rename ([] Atom) where
    findSub ......

How ever when i write this i get the following error:

* Illegal instance declaration for `Rename [Atom]'
        (All instance types must be of the form (T a1 ... an)
         where a1 ... an are *distinct type variables*,
         and each type variable appears at most once in the instance head.
         Use FlexibleInstances if you want to disable this.)
    * In the instance declaration for `Rename ([] Atom)'
Failed, modules loaded: none.

I would like to know how to fix this and why this is not allowed.

Thanks in advance.

luqui · Accepted Answer · 2017-06-04 23:21:39Z

4

It just requires a language extension, since this form of instance is not technically allowed by the Haskell 98 standard. When the error message says to enable an extension, usually just do it[1]. FlexibleInstances is a very common and accepted extension, so just put

{-# LANGUAGE FlexibleInstances #-}

at the top of the file.

Alternatively, often times such instances can be defined "recursively", in which case you don't need the extension, but besides that you get more instances for free, and more guarantees because the type is more general. See if you can define

instance (Rename a) => Rename [a] where
    findSub ...

instead, using only the fact that a is a Rename rather than it being specifically an Atom. When you can define instances this way, it is a good sign that your design is on the right track.

[1] There are a few exceptions, such as UndecidableInstances or IncoherentInstances, which sometimes indicates that you have a wrong idea about how typeclass resolution works.

edited Jun 4, 2017 at 23:21

answered Jun 4, 2017 at 23:01

luqui

60.7k12 gold badges153 silver badges210 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

dfeuer Over a year ago

I don't tend to see UndecidableInstances showing up in mistaken code; I tend to see it much more in perfectly good code that Haskell's extremely limited termination criteria are insufficient to prove good.

luqui Over a year ago

@dfeuer, Once you're comfortable with typeclasses I agree, but often beginners tend to think of typeclass resolution backwards or as being much more powerful than it is (for example instance Vector a => Num a and the like), which quickly leads to these extensions being "needed" (though when they are enabled it still doesn't work as intended). No?

dfeuer Over a year ago

I tend to think that much more about OverlappingInstances, IncoherentInstances, AllowAmbiguousTypes, and (when not using multiparameter type classes) FlexibleInstances. But you may be right as well.

Collectives™ on Stack Overflow

Create instance for list of data types - Haskell

1 Answer 1

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related