[Haskell-cafe] Ambiguous type variable with subclass instance (also: is there a better way to do this?)

Daniel Fischer daniel.is.fischer at web.de
Thu Sep 17 12:48:46 EDT 2009


Am Donnerstag 17 September 2009 18:01:36 schrieb Andy Gimblett:
> On 17 Sep 2009, at 16:50, Daniel Fischer wrote:
> > Yes, the second appearance of 'constructors' is at an unspecified
> > type.
> >
> > instance (Enumerated a) => Target a where
> >    convert n
> >
> >       | n < 0     = Nothing
> >       | otherwise = case drop n constructors of
> >
> >                        (x:_) -> Just x
> >                        _ -> Nothing
> >
> > would make it compile.
>
> Neat trick.  It works: thanks!
>
> > But there'd be a risk that Target is unusable, depending on how
> > instance resolution is
> > done.
>
> Unusable?  How so?  Sorry, but I don't follow...
>

Cf. Section 7.6.3.3 of the user's guide:

"When matching, GHC takes no account of the context of the instance declaration (context1 
etc). GHC's default behaviour is that exactly one instance must match the constraint it is 
trying to resolve. It is fine for there to be a potential of overlap (by including both 
declarations (A) and (B), say); an error is only reported if a particular constraint 
matches more than one.

The -XOverlappingInstances flag instructs GHC to allow more than one instance to match, 
provided there is a most specific one. For example, the constraint C Int [Int] matches 
instances (A), (C) and (D), but the last is more specific, and hence is chosen. If there 
is no most-specific match, the program is rejected."

So for the matching, you have now

instance Target a where ...

which matches everything. Add an instance declaration of the form

instance (SomeClass b) => Target b where ...

and you're hosed.
Though I think that wouldn't compile, at least not without IncoherentInstances.
Actually, I think now that with one-parameter type classes, if it compiles, it will most 
likely work, at least I don't see the problems one can create with multi-parameter type 
classes now.

> -Andy
>



More information about the Haskell-Cafe mailing list