[PRL] Bob Harper's new blog

David Van Horn dvanhorn at ccs.neu.edu
Thu Mar 17 10:39:55 EDT 2011


> functionality metastasizes throughout a program.

I read this as a critique of the shape of programs going from:

(define (f x)
   (cond [(rect? x) ...rect code...]
         [(circ? x) ...circ code...]
         [(trap? x) ...trap code...]))

to

(class Rect
   (define (f) ...rect code...))

(class Circ
   (define (f) ...circ code...))

(class Trap
   (define (f) ...trap code...))

where these class definitions could be spread out over thousands of 
lines of code or across the file system.

How that is "anti-modular" or "anti-parallel", I don't get.

As for Will's comment about dynamic method dispatch at run time, you get 
that in functional languages too, right?.  It's just that you, the 
programmer, have to write it and are given the chance to screw it up.

Reading these blog posts, one of the things I take away is this: show, 
don't tell, in your writing.  Heated assertions with no evidence aren't 
persuasive -- and I'm a sympathetic reader.  If Bob can't reach the PRL 
with his imperatives, he's got little hope of changing the world.

David



More information about the PRL mailing list