RE: [Haskell-cafe] how to avoid overlapping instance error?
Unfortunately, the context of an instance decl is not taken into account when matching instance decls. Yes, it would make sense to do so, but it'd make the system yet more complicated. So Show (table item) overlaps with Show ([] item) Overlap is checked lazily, so if you look for Show (MyTable t) you won't get overlap. Only if you ask for Show (table item), where the type checker still doesn't know what type 'table' is going to be, do you get a problem. Simon | -----Original Message----- | From: haskell-cafe-bounces@haskell.org [mailto:haskell-cafe-bounces@haskell.org] On Behalf Of S. | Alexander Jacobson | Sent: 25 February 2005 17:09 | To: haskell-cafe@haskell.org | Subject: [Haskell-cafe] how to avoid overlapping instance error? | | | Currently, the HAppS.DBMS lib requires the user to provide a Show | instance for their table types. An example might be: | | instance (Show item) => Show (MyTable item) where | showsPrec d table = showsPrec d $ Set.toList $ myTableSet table | | But the Table class itself defines a toSet function so I think I | should be able to do this once for all Table instances: | | instance (Show item,Ord item, Table table item p) => Show (table item) where | showsPrec d table = showsPrec d $ Set.toList $ toSet table | | But I get an error telling me I am overlapping with Show [a]. | | Since [] is not an instance of Table, I don't see why there should be | an overlap. | | -Alex- | | ______________________________________________________________ | S. Alexander Jacobson tel:917-770-6565 http://alexjacobson.com | _______________________________________________ | Haskell-Cafe mailing list | Haskell-Cafe@haskell.org | http://www.haskell.org/mailman/listinfo/haskell-cafe
If you are not using them to prevent overlapping instances, then why require instance decls at all? For example, why does GHC require an instance decl here: instance (Ord x)=>ToSet [] x where toSet = Set.fromList But not here: listToSet x = Set.fromList x Or I suppose, one could rephrase this question as why not simplify instance declarations to be: instance ToSet where toSet = Set.fromList And let the typechecker take care of figuring out what instance is being specified here? -Alex- ______________________________________________________________ S. Alexander Jacobson tel:917-770-6565 http://alexjacobson.com On Mon, 28 Feb 2005, Simon Peyton-Jones wrote:
Unfortunately, the context of an instance decl is not taken into account when matching instance decls. Yes, it would make sense to do so, but it'd make the system yet more complicated.
So Show (table item) overlaps with Show ([] item)
Overlap is checked lazily, so if you look for Show (MyTable t) you won't get overlap. Only if you ask for Show (table item), where the type checker still doesn't know what type 'table' is going to be, do you get a problem.
Simon
| -----Original Message----- | From: haskell-cafe-bounces@haskell.org [mailto:haskell-cafe-bounces@haskell.org] On Behalf Of S. | Alexander Jacobson | Sent: 25 February 2005 17:09 | To: haskell-cafe@haskell.org | Subject: [Haskell-cafe] how to avoid overlapping instance error? | | | Currently, the HAppS.DBMS lib requires the user to provide a Show | instance for their table types. An example might be: | | instance (Show item) => Show (MyTable item) where | showsPrec d table = showsPrec d $ Set.toList $ myTableSet table | | But the Table class itself defines a toSet function so I think I | should be able to do this once for all Table instances: | | instance (Show item,Ord item, Table table item p) => Show (table item) where | showsPrec d table = showsPrec d $ Set.toList $ toSet table | | But I get an error telling me I am overlapping with Show [a]. | | Since [] is not an instance of Table, I don't see why there should be | an overlap. | | -Alex- | | ______________________________________________________________ | S. Alexander Jacobson tel:917-770-6565 http://alexjacobson.com | _______________________________________________ | Haskell-Cafe mailing list | Haskell-Cafe@haskell.org | http://www.haskell.org/mailman/listinfo/haskell-cafe
participants (2)
-
S. Alexander Jacobson -
Simon Peyton-Jones