From mboxrd@z Thu Jan  1 00:00:00 1970
Received: (from majordomo@localhost) by pauillac.inria.fr (8.7.6/8.7.3) id EAA20499; Sat, 14 Jul 2001 04:52:20 +0200 (MET DST)
X-Authentication-Warning: pauillac.inria.fr: majordomo set sender to owner-caml-list@pauillac.inria.fr using -f
Received: from nez-perce.inria.fr (nez-perce.inria.fr [192.93.2.78]) by pauillac.inria.fr (8.7.6/8.7.3) with ESMTP id EAA20491 for <caml-list@pauillac.inria.fr>; Sat, 14 Jul 2001 04:52:19 +0200 (MET DST)
Received: from str12.sobor.org (adsl-63-198-183-99.dsl.snfc21.pacbell.net [63.198.183.99])
	by nez-perce.inria.fr (8.11.1/8.10.0) with ESMTP id f6E2qIj25997;
	Sat, 14 Jul 2001 04:52:18 +0200 (MET DST)
Received: from quasar.ipa.nw.ru (anza.sobor.org [192.168.123.51])
	by str12.sobor.org (Postfix) with ESMTP
	id B43C0AC6D2; Fri, 13 Jul 2001 19:39:05 -0700 (PDT)
Message-ID: <3B4FB39D.10D1F6D5@quasar.ipa.nw.ru>
Date: Fri, 13 Jul 2001 19:51:09 -0700
From: "Alexander V. Voinov" <avv@quasar.ipa.nw.ru>
Organization: Fear of the Lord is the beginning of wisdom
X-Mailer: Mozilla 4.75 [en] (Windows NT 5.0; U)
X-Accept-Language: en
MIME-Version: 1.0
To: Daniel de Rauglaudre <daniel.de_rauglaudre@inria.fr>
Cc: "caml-list@inria.fr" <caml-list@inria.fr>
Subject: Re: [Caml-list] Generation of streams is slow
References: <3B4F70D7.82E58A91@quasar.ipa.nw.ru> <20010714042701.A24266@verdot.inria.fr>
Content-Type: multipart/mixed;
 boundary="------------8AE30CFF45EF624A3F321161"
Sender: owner-caml-list@pauillac.inria.fr
Precedence: bulk

This is a multi-part message in MIME format.
--------------8AE30CFF45EF624A3F321161
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Hi Daniel,

Daniel de Rauglaudre wrote:
> 
> Hi,
> 
> On Fri, Jul 13, 2001 at 03:06:15PM -0700, Alexander V. Voinov wrote:
> 
> > I've run the following program and found it to be extremely slow. Why?
> 
> To be optimized, streams and parsers must be preprocessed by Camlp4.
> If they are not, their code is longer, not tail recursive and can be
> very very slow indeed.
> 
> Compile you file with the option -pp camlp4o.

Yes, the difference is substantial, but it gets longer and longer anyway
with more elements parsed (for a slightly more complex program, which
parses the stream as it is generated) and crashes at some point.

I've attached the program (change parse_stream to parse_stream' and vice
versa).

Alexander
--------------8AE30CFF45EF624A3F321161
Content-Type: text/plain; charset=us-ascii;
 name="test_streams.ml"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
 filename="test_streams.ml"

open Printf

let gen_stream n =
  let rec gs i n =
    if i = n then begin
      printf "making last element of the stream\n"; 
      flush stdout;
      [< 'n >]
    end
    else begin
      if i mod 1000 = 0 then begin
	printf "making next element of the stream: %d\n" i; 
	flush stdout
      end;
      [< 'i; (gs (i+1) n) >] 
    end 
  in
  gs 1 n
	
let rec print_list = function
    []   -> printf "[]\n"
  | h::t -> printf "%d::" h; 
            print_list t

let rec parse_stream s =
  match s with parser
    [< 'i; s' >] -> begin 
      
      if i mod 1000 = 0 then begin
	printf "picked element %d\n" i; 
	flush stdout
      end; 
      
      i::(parse_stream s')
    end
  | [< >] -> (printf "end\n"; flush stdout; [])

let rec parse_stream' s =
  match s with parser
    [< 'i; tail = parse_stream' >] -> begin 
      
      if i mod 1000 = 0 then begin
	printf "picked element %d\n" i; 
	flush stdout
      end; 
      
      i::tail
    end
  | [< >] -> (printf "end\n"; flush stdout; [])

let _ = 
  print_list (parse_stream' (gen_stream 700000))

--------------8AE30CFF45EF624A3F321161--

-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr