Excerpts from Edward Z. Yang's message of Fri Jul 30 14:48:34 -0400 2010:

Const x xs is x:xs (constructor)

That should be a Cons, not Const. :o) Cheers, Edward

On Fri, Jul 30, 2010 at 8:54 PM, michael rice wrote:

Thanks all,

Now that I have a (very) rudimentary understanding of Haskell, I figured I'd back up and have a closer (conceptual) look at type definitions to see what they have in common, and just happen to pick Maybe and List.

I also noticed Maybe has a list of "Instances"

Monad Maybe Functor Maybe Typeable1 Maybe MonadFix Maybe MonadPlus Maybe etc.

while List has none, at least I don't see any in Data.List. Same reason?

From "Learn You A Haskell:"

"If a type is a part of a typeclass, that means it supports and implements the behavior the typeclass describes."

I'm way out on a limb here, but isn't Monad a typeclass? and if, as we say above, that Maybe is an instance of Monad, wouldn't there have to be

instance Monad Maybe where  return = ...  -- return for Maybe  >>= = ...     -- bind for Maybe  etc.

somewhere? Where? It's not in Data.Maybe. Is there some kind of scheme for defining this stuff, i.e., this goes here, that goes there?

It *is* in Data.Maybe: http://hackage.haskell.org/packages/archive/base/4.2.0.1/doc/html/src/Data-M... Generally speaking a type class instance should go either where the type is defined, or where the class is defined, although there are exceptions. In any case, if you can get the instance loaded in ghci, you can find out where its defined: ghci> :i Maybe data Maybe a = Nothing | Just a -- Defined in Data.Maybe instance (Eq a) => Eq (Maybe a) -- Defined in Data.Maybe instance Monad Maybe -- Defined in Data.Maybe instance Functor Maybe -- Defined in Data.Maybe instance (Ord a) => Ord (Maybe a) -- Defined in Data.Maybe instance (Read a) => Read (Maybe a) -- Defined in GHC.Read instance (Show a) => Show (Maybe a) -- Defined in GHC.Show

Hi Daniel, OK, now I'm getting somewhere. I was looking here: http://haskell.org/ghc/docs/6.12.1/html/libraries/base-4.2.0.0/Data-Maybe.ht... instead of in the source. Seem to be on the right track. I'm going to do some more poking around. Thanks a lot, guys. Michael --- On Fri, 7/30/10, Daniel Fischer wrote: From: Daniel Fischer Subject: Re: [Haskell-cafe] Definition of List type? To: haskell-cafe@haskell.org Cc: "michael rice" Date: Friday, July 30, 2010, 4:23 PM On Friday 30 July 2010 21:54:20, michael rice wrote:

Thanks all,

Now that I have a (very) rudimentary understanding of Haskell, I figured I'd back up and have a closer (conceptual) look at type definitions to see what they have in common, and just happen to pick Maybe and List.

I also noticed Maybe has a list of "Instances"

Monad Maybe Functor Maybe Typeable1 Maybe MonadFix Maybe MonadPlus Maybe etc.

while List has none, at least I don't see any in Data.List. Same reason?

Data.List provides only functions for working with lists, while Data.Maybe also contains the definition of the type. Type class instances are documented - at the type class - at the data type Since the list datatype isn't documented like the other types (probably because it's not defined in valid Haskell but built-in), the list instances appear only at the type classes (and at the Monad documentation, you find "Monad []" listed as the first instance.

...

From "Learn You A Haskell:"

"If a type is a part of a typeclass, that means it supports and implements the behavior the typeclass describes."

I'm way out on a limb here, but isn't Monad a typeclass?

Yes.

and if, as we say above, that Maybe is an instance of Monad, wouldn't there have to be

instance Monad Maybe where  return = ...  -- return for Maybe  >>= = ...     -- bind for Maybe  etc.

somewhere?

Yes, there is. In GHC at least, it's defined in Data.Maybe: instance  Monad Maybe  where     (Just x) >>= k      = k x     Nothing  >>= _      = Nothing     (Just _) >>  k      = k     Nothing  >>  _      = Nothing     return              = Just     fail _              = Nothing

Where? It's not in Data.Maybe. Is there some kind of scheme for defining this stuff, i.e., this goes here, that goes there?

Instance declarations should be in the same module where - the class is defined, or - the type is defined if possible.

On Fri, Jul 30, 2010 at 7:50 PM, Gregory Crosswhite wrote:

The reason why this definition never actually appears is because it defines the constructors using operators rather than names, which is not allowed in vanilla Haskell.  (There is an extension, TypeOperators, however, that does allow this.)

 Nope: see Data.Complex.Complex; only infix *type* constructors are nonstandard. The thing about lists that makes them impossible to define in normal Haskell is the [a] syntax, which is some kind of "outfix" type constructor, which no amount of currently available extensions will allow. In addition, the constructor [] for the empty list isn't a normal constructor, syntactically, because it doesn't start with an uppercase character or a colon. Basically, lists are so ubiquitous in Haskell that they have their own special syntax, which cannot be defined like any other data type. It is simple, as others have said, to define a new data type that works identically to lists.