Re: [stack] Self Types

Hi Manfred,

What you call "X" I believe corresponds directly to the classical "M"
combinator according to Brent Kerby's concatenative combinator
mapping. The SK and Lambda Calculus definitions of "M" can be found at
http://www.angelfire.com/tx4/cus/combinator/birds.html

On 6/14/07, Manfred Von Thun <[email protected]> wrote:
>
> Further to my previous reply:
>
> Cat seems to be related to Joy in the same way as
> Haskell and (the functional subset of) ML are related
> to (the functional subset of) Lisp/Scheme: Cat is
> strongly typed and Joy is not.

Yep.

> A good question to ask: Even if y and x are possible in Joy
> just as y is possible in Lisp/Scheme, it might just be that
> y and x (your dup apply) are not possible in Cat.

I have to extend the type system to handle them. The current type
system can't handle it yet. I feel I am very close to having them
working.

> Am I
> right in believing that y is not possible in Haskell and ML?

I think that Y and M are possible in Haskell using GADT's, but I am
unsure. I know that Haskell can't infer the types. I can't say
anything about ML.

> If so, then perhaps your dup apply problem would
> disappear ­ simply by prohibiting it.

That is definitely a valid approach. However, I find that they Y and M
useful combinators (for example you can define looping constructs
using M). M can be seen as a classical combinatorial definition of
recursion. Providing a meaningful type for it is desirable for my
purposes.

> Are there any Haskellers or MLers on board?
>
> - Manfred

 Christoper