RE: missing Data instances
Simon PJ, Is that Ok? Should I? Ralf > -Original Message- > From: [EMAIL PROTECTED] [mailto:glasgow-haskell- > [EMAIL PROTECTED] On Behalf Of Frederik Eaton > Sent: Thursday, December 29, 2005 11:39 AM > To: glasgow-haskell-bugs@haskell.org > Subject: missing Data instances > > These don't seem to have Data instances in 6.4.1: > > Language.Haskell.TH.Exp > Data.Tree.Tree > > I guess there are probably a lot more, I just ran into these today. Is > there anything which prevents proper 'deriving' clauses from being > added to their definitions? > > Frederik > ___ > Glasgow-haskell-bugs mailing list > Glasgow-haskell-bugs@haskell.org > http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs ___ Glasgow-haskell-bugs mailing list Glasgow-haskell-bugs@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
missing Data instances
These don't seem to have Data instances in 6.4.1: Language.Haskell.TH.Exp Data.Tree.Tree I guess there are probably a lot more, I just ran into these today. Is there anything which prevents proper 'deriving' clauses from being added to their definitions? Frederik ___ Glasgow-haskell-bugs mailing list Glasgow-haskell-bugs@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
Re: Constr and Eq
I see. Maybe there should be a separate function efficientConstrEq :: Constr -> Constr -> Bool so that people don't get confused with the standard Eq definition? How often is the efficient version needed? In any case, it seems like we should document the fact that the Constr Eq instance doesn't work in the usual way. Frederik On Thu, Dec 29, 2005 at 11:23:15AM -0800, Ralf Lammel wrote: > As of today, this is done so to be efficient. > You can easily compare the data types of the constructors to further > constrain equivalence, when this is necessary, while the base check only > compares an Integer value for the constructors. > Ralf > > > > -Original Message- > > From: [EMAIL PROTECTED] > [mailto:glasgow-haskell- > > [EMAIL PROTECTED] On Behalf Of Frederik Eaton > > Sent: Thursday, December 29, 2005 10:57 AM > > To: glasgow-haskell-bugs@haskell.org > > Subject: Constr and Eq > > > > > toConstr True > > True > > > toConstr [()] > > (:) > > > toConstr True == toConstr [()] > > True > > > show (toConstr True) == show (toConstr [()]) > > False > > > > I don't understand the utility of Constr's Eq definition. It isn't > > really documented though. Why return True when the constructors being > > compared are different? Am I supposed to use (show.toConstr) to > > compare constructors? > > > > Here is the documentation I could find: > > > > data Constr > > Representation of constructors > > Instances > > Show Constr > > Eq Constr > > > > This is ghc 6.4.1. > > > > Thanks, > > > > Frederik > > ___ > > Glasgow-haskell-bugs mailing list > > Glasgow-haskell-bugs@haskell.org > > http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs > ___ Glasgow-haskell-bugs mailing list Glasgow-haskell-bugs@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
RE: Constr and Eq
As of today, this is done so to be efficient. You can easily compare the data types of the constructors to further constrain equivalence, when this is necessary, while the base check only compares an Integer value for the constructors. Ralf > -Original Message- > From: [EMAIL PROTECTED] > [mailto:glasgow-haskell- [EMAIL PROTECTED] On Behalf Of > Frederik Eaton > Sent: Thursday, December 29, 2005 10:57 AM > To: glasgow-haskell-bugs@haskell.org > Subject: Constr and Eq > > > toConstr True > True > > toConstr [()] > (:) > > toConstr True == toConstr [()] > True > > show (toConstr True) == show (toConstr [()]) > False > > I don't understand the utility of Constr's Eq definition. It isn't > really documented though. Why return True when the constructors being > compared are different? Am I supposed to use (show.toConstr) to > compare constructors? > > Here is the documentation I could find: > > data Constr > Representation of constructors > Instances > Show Constr > Eq Constr > > This is ghc 6.4.1. > > Thanks, > > Frederik > ___ > Glasgow-haskell-bugs mailing list > Glasgow-haskell-bugs@haskell.org > http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs ___ Glasgow-haskell-bugs mailing list Glasgow-haskell-bugs@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
Constr and Eq
> toConstr True True > toConstr [()] (:) > toConstr True == toConstr [()] True > show (toConstr True) == show (toConstr [()]) False I don't understand the utility of Constr's Eq definition. It isn't really documented though. Why return True when the constructors being compared are different? Am I supposed to use (show.toConstr) to compare constructors? Here is the documentation I could find: data Constr Representation of constructors Instances Show Constr Eq Constr This is ghc 6.4.1. Thanks, Frederik ___ Glasgow-haskell-bugs mailing list Glasgow-haskell-bugs@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
RE: GADTs and type synonyms
Hmm. There's a question here. Constructor args allow you to put strictness annotations data Foo where MkFoo :: !Int -> !Bool -> Foo You want to hide part of the declaration inside a type signature. But then what about the strictness annotations? I suppose you could argue for type Baz = !Bool -> Foo data Foo where MkFoo :: !Int -> Baz But that's a new kind of 'type' declaration. Maybe you could argue that you should be allowed to use a type synonym when you don't need strictness annotations, but that seems a bit of a hack. So GHC currently insists that you write all the argument types of the constructor, and the result, explicitly, with no type synonyms in the way. Do you think that's bad? The error message could be better I suppose. Simon | -Original Message- | From: [EMAIL PROTECTED] [mailto:glasgow-haskell-bugs- | [EMAIL PROTECTED] On Behalf Of Ralf Hinze | Sent: 28 December 2005 07:11 | To: glasgow-haskell-bugs@haskell.org | Subject: GADTs and type synonyms | | Another GADT `bug report': type synonyms are not unfolded | when checking the signature of the data constructors: given | | > type Hello a = a -> a | > data World :: * where | > Msg :: Hello World | | GHC reports | | GADT.lhs:2:1: | Data constructor `Msg' returns type `Hello' | instead of its parent type | When checking the data constructor: Msg | In the data type declaration for `World' | | Applies to both version 6.4.1 and 6.5. | | Cheers, Ralf | ___ | Glasgow-haskell-bugs mailing list | Glasgow-haskell-bugs@haskell.org | http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs ___ Glasgow-haskell-bugs mailing list Glasgow-haskell-bugs@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
