From: "Quôc Peyrot" <chojin@lrde.epita.fr>
To: caml-list@yquem.inria.fr
Subject: Re: [Caml-list] The Implicit Accumulator: a design pattern using optional arguments
Date: Wed, 27 Jun 2007 15:18:07 +0200 [thread overview]
Message-ID: <1A1D6F56-B3DB-4552-969C-9859482175AC@lrde.epita.fr> (raw)
In-Reply-To: <200706271314.35134.jon@ffconsultancy.com>
On Jun 27, 2007, at 2:14 PM, Jon Harrop wrote:
>
> I can't find the thread where we were talking about design patterns
> recently
> but I'd like to note a design pattern that works nicely in OCaml.
> I'll call
> it "The Implicit Accumulator".
>
> ML programmers often use nested auxiliary functions or separate
> functions to
> handle base cases. For example, writing rev in terms of rev_append:
>
> # let rec rev_append l1 l2 = match l1 with
> | [] -> l2
> | a :: l -> rev_append l (a :: l2);;
> val rev_append : 'a list -> 'a list -> 'a list = <fun>
> # let rev l = rev_append l [];;
> val rev : 'a list -> 'a list = <fun>
>
> Provided performance is unimportant, you can make the accumulator
> implicit in
> OCaml by specifying the default value in an optional argument
> instead of
> having a separate function:
>
> # let rec rev ?(back=[]) = function
> | [] -> back
> | h::t -> rev ~back:(h::back) t;;
> val rev : ?back:'a list -> 'a list -> 'a list = <fun>
Could you be more specifics about the performance hit?
> When you don't want the auxiliary (rev_append) function, I think
> this style
> results in shorter and clearer code. I used it in the "search"
> function of my
> Sudoku solver, for example:
It's funny that you speak about this, because I recently (few days
ago) used
a pattern similar to yours, but to actually improve performances.
I had something like that (which is quite different than my actual
code, but
the idea is the same):
let encrypt str =
let len = String.length str in
let encrypted = String.create len in
(* ... *)
encrypted
(*...*)
for i = 0 to 10000000 do
let encrypted = encrypt str in
(* do something on the result *)
done
Which is slow due to the string allocation happening each time we
call "encrypt"
So I rewrote it like that:
let encrypt ?encrypted str =
let len = String.length str in
let result = match encrypted with
| None -> String.create len
| Some s -> s
in
(* ... *)
result
(* ... *)
let encrypted = String.create (String.length str) in
for i = 0 to 1000000000 do
let encrypted = encrypt ~encrypted str in
(* ... *)
done
Which gave me more than a 2x speedup while still being able to call a
simple:
let encrypted = encrypt str
during normal usage
I was quite happy with this solution, but maybe there is something
more elegant to do?
(I'm still in the process of learning good optimization patterns in
ocaml which preserve readability)
--
Best Regards,
Quôc
next prev parent reply other threads:[~2007-06-27 13:18 UTC|newest]
Thread overview: 60+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-06-27 12:14 Jon Harrop
2007-06-27 13:18 ` Quôc Peyrot [this message]
2007-06-27 13:53 ` [Caml-list] " Jon Harrop
2007-06-27 14:18 ` Thomas Fischbacher
2007-06-27 15:09 ` Quôc Peyrot
2007-06-27 15:28 ` Mattias Engdegård
2007-06-27 15:38 ` Robert Fischer
2007-06-27 15:48 ` Mattias Engdegård
2007-06-27 16:01 ` Robert Fischer
2007-06-27 16:01 ` Mattias Engdegård
2007-06-27 18:06 ` Jon Harrop
2007-06-27 18:31 ` Brian Hurt
2007-06-27 19:56 ` skaller
2007-06-27 20:17 ` Jonathan Bryant
2007-06-27 22:57 ` Jon Harrop
2007-06-27 16:53 ` Hash-consing (was Re: [Caml-list] The Implicit Accumulator: a design pattern using optional arguments) Daniel Bünzli
2007-06-30 8:19 ` [Caml-list] The Implicit Accumulator: a design pattern using optional arguments Pierre Etchemaïté
2007-06-27 13:55 ` Thomas Fischbacher
2007-06-27 15:06 ` Quôc Peyrot
2007-06-27 15:53 ` Jon Harrop
2007-06-28 11:01 ` Thomas Fischbacher
2007-06-28 11:32 ` Jon Harrop
2007-06-28 11:42 ` Joel Reymont
2007-06-28 12:08 ` Jon Harrop
2007-06-28 13:10 ` Quôc Peyrot
2007-06-28 13:35 ` Thomas Fischbacher
2007-06-28 12:59 ` Thomas Fischbacher
2007-06-28 13:05 ` Jon Harrop
2007-06-28 13:33 ` Thomas Fischbacher
2007-06-28 14:43 ` Jon Harrop
2007-06-28 16:01 ` Thomas Fischbacher
2007-06-28 17:53 ` Jon Harrop
2007-06-27 16:39 ` Thomas Fischbacher
2007-06-27 19:26 ` Quôc Peyrot
2007-06-28 11:39 ` Thomas Fischbacher
2007-06-28 14:44 ` Jon Harrop
2007-06-28 16:03 ` Thomas Fischbacher
2007-06-28 17:20 ` Dirk Thierbach
2007-06-28 22:12 ` Thomas Fischbacher
2007-06-29 1:10 ` Jon Harrop
2007-06-29 10:55 ` Thomas Fischbacher
2007-06-29 6:12 ` Dirk Thierbach
2007-06-27 17:16 ` Book about functional design patterns Gabriel Kerneis
2007-06-27 17:48 ` [Caml-list] " Jon Harrop
2007-06-27 19:33 ` Quôc Peyrot
2007-06-27 19:30 ` Quôc Peyrot
2007-06-27 19:48 ` Brian Hurt
2007-06-27 20:04 ` Quôc Peyrot
2007-06-27 20:35 ` Brian Hurt
2007-06-27 20:55 ` Quôc Peyrot
2007-06-27 20:58 ` Pal-Kristian Engstad
2007-06-27 21:18 ` Quôc Peyrot
2007-06-27 21:18 ` Pal-Kristian Engstad
2007-06-27 21:34 ` Quôc Peyrot
2007-06-27 22:13 ` Pal-Kristian Engstad
2007-06-27 15:18 ` [Caml-list] The Implicit Accumulator: a design pattern using optional arguments Jon Harrop
2007-06-27 16:44 ` Thomas Fischbacher
2007-06-27 18:17 ` Jon Harrop
2007-06-28 11:18 ` Thomas Fischbacher
2007-06-29 13:15 ` Bill Wood
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1A1D6F56-B3DB-4552-969C-9859482175AC@lrde.epita.fr \
--to=chojin@lrde.epita.fr \
--cc=caml-list@yquem.inria.fr \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox