Version française
Home     About     Download     Resources     Contact us    
Browse thread
Avoiding shared data
[ Home ] [ Index: by date | by threads ]
[ Search: ]

[ Message by date: previous | next ] [ Message in thread: previous | next ] [ Thread: previous | next ]
Date: -- (:)
From: Pal-Kristian Engstad <pal_engstad@n...>
Subject: Re: Ant: Re: [Caml-list] Avoiding shared data
On Friday 30 September 2005 03:57 pm, Oliver Bandel wrote:
> On the other hand: writing mor funtional/recursive code will
> make you more used to to this...

I've always thought that this was a really bad argument from the ML camp. The 
logic of complicated control-paths is very easily made a zillion times worse 
by writing in a tail-recursive style. It is *not* a good programming practice 
to make hard-to-read code!

I encourage people to read the paper by Olin Shivers: "The Anatomy of a Loop - 
A story of scope and control", which was presented at ICFP 2005, and can be 
found at http://www.cc.gatech.edu/~shivers/papers/loop.pdf.

The author argues that "Writing loops with tail-recursive function calls is 
the equivalent of writing them with goto’s." and gives an example that I've 
rewritten from Scheme-ish into OCaml-ish:

let myfunc l =
  let rec loop rest result =
    match rest with 
      | [] -> List.rev result
      | x::xs -> 
	  if xpred x then
	    let y = verbose_code_using_x x in
	      if ypred y then 
		let z = verbose_code_using_y y in
		  loop xs (z_expression :: result)
	      else
		loop xs result
	  else
	    loop xs result
  in
    loop l [] 
;;

Obviously, one would like to refactor this into HOF, but in this situation it 
is hard to see how one should do it. The author instead proposes to use loops 
in a monadic style, which I've again rewritten:

let myfunc l =
   loop [ for x in l
	; when xpred x 
	;   let y = verbose_code_using_x x
	;   when ypred y
	;     let z = verbose_code_using_y y
	;     save z
	]

The point being (syntax aside) that this code is much more readible, easier to 
change and easier to verify than the code given above. [Of course, Haskell 
has the really cool list-comprehension syntax that alleviates some of ML's 
problems, but still.]

Thanks,

PKE.
-- 
  _       
  \`.       PÃ¥l-Kristian Engstad, Lead Programmer,
   \ `|     Naughty Dog, Inc., 1601 Cloverfield Blvd, 6000 North,
  __\ |`.   Santa Monica, CA 90404, USA. (310) 633-9112. 
    /  /o   mailto:engstad@naughtydog.com http://www.naughtydog.com
   /  '~    mailto:mrengstad@yahoo.com    http://www.engstad.com
  / ,'      Hang-gliding Rulez!
  ~'