Another option is to just add the instance:

> instance FooBar Char

instead of

> instance (Foo a, Bar a) => FooBar a

Now you don't need any extensions, the disadvantage is that you have to add an instance for each type...

There has also been an proposal for type class synonyms: http://repetae.net/john/recent/out/classalias.html

Maybe you like it,

Gerrit


Sean Seefried wrote:


On 24/03/2006, at 12:45 PM, Fritz Ruehr wrote:

What is the easiest way to name a combination of type classes, i.e., to abbreviate the fact that a certain type is an instance of several classes simultaneously? I have a vague sense that this is do-able, but that I am messing up by trying to use an empty class body as below.

So in the code below, I try to use FooBar to abbreviate the conjunction of Foo and Bar. But while f (which uses a FooBar constraint) has a valid definition, it can't be used. On the other hand, g (which uses the long-winded constraint), is both a valid defined and useable.

(In a real example, imagine that FooBar names a conjunction of a half dozen things, so that the g-like form really is onerous, whereas the f-like form would be sweet and tidy :) .)


Hi Fritz!

You only need to do a couple of things to get this working. Add an instance declaration:

instance (Foo a, Bar a) => FooBar a

But for this to work you need to allow undecidable instances (and - fglasgow-exts).

To have this type class synonym trick work you need both the class and instance declaration:

class    (Foo a, Bar a) => FooBar a
instance (Foo a, Bar a) => FooBar a

The first ensures that members of class FooBar will inherit the methods of classes Foo and Bar. The second ensures that if there is a Foo and a Bar instance then there will be a FooBar instance. You were lacking this in your code hence the error message:

> f 'a'

No instance for (FooBar Char)
      arising from use of `f' at <interactive>:1:0
    Probable fix: add an instance declaration for (FooBar Char)
    In the definition of `it': it = f 'a'

This is a neat trick. I've also used it to reduce onerous contexts.

Sean
_______________________________________________
Haskell-Cafe mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/haskell-cafe


_______________________________________________
Haskell-Cafe mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/haskell-cafe

Reply via email to