Hello,

I like this. But where can I try it ? Squeaksource and squeakmap do  
not know about it. I would like to see more about it.

> LambdaMessageSend, aka FunctionalTalk, implements a specialized  
> MessageSend subclass and an associated API for functional  
> programming in Squeak.

Few months ago I worked on a module calculus. I first did an  
implementation in Squeak, but I had to switch for Scheme, mainly  
because of the closure.

>
> you can do things like:
>
> | x y f g j |
>
> x _ Lambda x.
> y _ Lambda y.

Why do you use the term lambda to create a variable ?

> f _ ((x*x) + (y*y)) sqrt.

Why is there no term named lambda here ?

> self assert: f printString = '(((<x> * <x>) + (<y> * <y>)) sqrt)'.
> self assert: {3 . 4} ~~> f = 5.
> g := ('y' -> 0) ~>> f.

I like this mechanism of curryfication.

> self assert: g printString = '(((<x> * <x>) + 0) sqrt)' .
> self assert: 3 ~> g = 3.
>

> j := ('y' -> (Lambda z1 + Lambda z2)) ~>> f.

I did not get this one... Could you explain a bit ?

> self assert: j printString = '(((<x> * <x>) + ((<z1> + <z2>) *  
> (<z1> + <z2>))) sqrt)' .
> self assert: {3 . 2 . 2} ~~> j = 5.
>
>
> see the test cases in FunctionalTalkTest for comprehensive  
> documentation.

Where are they ?


> I would be interested in feedback and criticism from languages gurus.
>
> the idea here is to define a MessageSend those receiver, selector  
> and arguments can themselves be MessageSend. It seems to me that we  
> get this way a clean lambda calculus framework but of course I may  
> be wrong.

It seems to me that, what is tricky with lambda calculus are the  
evaluation strategies...

Cheers,
Alexandre


>
> what do you think ?
>
>
> Stef
>
>

-- 
_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel  http://www.cs.tcd.ie/Alexandre.Bergel
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.