From: David Flanagan <david@...>
Date: 2007-06-06T13:37:41+09:00
Subject: Re: currying in Ruby

Thanks for your response, Victor.

Could you clarify about currying and instance_eval?  What is the "usual 
problem"

The >> and << are (obviously) and experimental syntax.  I like that it 
specifies whether the argument being curried goes on the left hand side 
(>>) or the right hand side (<<) of the argument list. I gather that 
true functional programming fans like currying to happen automatically 
if you invoke a function with too few arguments.  So I was looking for a 
compact syntax, akin to invocation, for currying.

I don't have a way to curry arguments in the middle of the list, but I 
suspect that this is not, in practice, that big a problem.  I would 
argue that the overhead of allowing arbitrary "holes" as the rubymurray 
implementation does is too great, and that when you need to do this 
you'd be better off just writing your own custom proc wrapper to do what 
you need.

My complaints with rubymurray are:

1) its overkill--just too long

2) it bothers me that currying a Proc returns a Curry object rather than 
a Proc object.  And that the resulting Curry object is not itself Curryable.

	David

Victor "Zverok" Shepelev wrote:
> From: David Flanagan [mailto:david@davidflanagan.com]
> Sent: Wednesday, June 06, 2007 3:35 AM
>> I've written a little argument currying module for Procs and Methods.  I
>> think my only real contribution here is the use of << and >> operators
>> as a shorthand, which I haven't seen done elsewhere.
>>
>> Anyone care to review the code and comment?

> 
> * The usual problem with proc currying: instance_eval (and instance_exec in
> Ruby 1.9 and Facets imitation).
> 
> * Personally I dislike >> and << syntax. What this should read like? "shift
> difference to 10"? And what about currying of "middle" parameters?
> 
> * What's wrong with existing currying module[1]? It's syntax is a bit more
> verbose, but reads more natural, I think (especially if alias HOLE and
> BLACKHOLE with shorter constants).
> 
> V.
> 
> 1. http://rubymurray.rubyforge.org/ 
> 
>