# [racket] Function composition in Racket

To use prop:procedure, just give a function which will handle the application of the structure to some arguments. The define-values is only there because the for/fold has two accumulators (sum and x) and will therefore return two values (the values of those accumulators). This means its context has to expect two values, even though we don't really care about the second value. The x* is just a name.
---
Justin Slepak
PhD student, Computer Science dept.
----- Original Message -----
From: Gregory Woodhouse <gregwoodhouse at me.com>
To: Justin R. Slepak <jrslepak at ccs.neu.edu>
Sent: Sun, 14 Oct 2012 22:07:59 -0400 (EDT)
Subject: Re: [racket] Function composition in Racket
Thanks! This does what I want. To tell you the truth, I've shied away from prop:procedure (probably more due to my own confusion than anything else!). The define-values here seems a bit mysterious, but I assume the point is to support the recursive polynomial evaluation function? Is x* a special syntax here or just a name. I see sum in both for/fold and values, but the x* in define-values is mysterious.
On Oct 14, 2012, at 4:57 PM, "Justin R. Slepak" <jrslepak at ccs.neu.edu> wrote:
>* Instead of trying to peek inside a lambda, you could implement polynomials as (transparent) structs and use prop:procedure to allow them to be applied to numbers:
*>*
*>* (struct polynomial (coeffs)
*>* #:transparent
*>* #:property prop:procedure
*>* (lambda (poly num)
*>* (define-values (result x*)
*>* (for/fold ([sum 0]
*>* [x 1])
*>* ([c (polynomial-coeffs poly)])
*>* (values (+ sum (* c x))
*>* (* x num))))
*>* result))
*>*
*>* This would let you implement functions that crunch on polynomials. As for using existing operators' names, maybe generics can get you that?
*>*
*>* ---
*>* Justin Slepak
*>* PhD student, Computer Science dept.
*