Re: The need to specify 'rec' in a recursive function defintion

Re: The need to specify 'rec' in a recursive function defintion

[oleg]

Fortunately OCaml is (much more) than simply-typed lambda
calculus. Almost any feature of OCaml -- recursive data types,
recursive types, reference cells, mutable records, exceptions,
objects, recursive modules and polymorphic variants -- can be used to
express the fixpoint combinator. Sometimes there is more than one way
to use the same feature to express the fixpoint combinator. The more
and less known ways of expressing fix are collected in the following
document:

	http://okmij.org/ftp/ML/fixpoints.ml

The need to specify 'rec' in a recursive function defintion

[Saptarshi Guha]

Hello,
 I was wondering why recursive functions need to be specified with
"rec". According to Practical Ocaml, to "inform the compiler that the function
exists". But when entering the function definition, can't the compiler note that
the function is being defined so that when it sees the function calling itself,
it wont say "Unbound value f"?

How is the knowledge of a function being rec taken advantage of (in
ocaml) as opposed to other languages
(leaving aside tail call optimization).

Wouldn't one of way of detecting a recursive function would be to see
if the indeed the function calls itself?

These are very much beginners' questions.
Thank you
Saptarshi

Re: The need to specify 'rec' in a recursive function defintion

[Stefan Monnier]

> Wouldn't one of way of detecting a recursive function would be to see
> if the indeed the function calls itself?

That's what Haskell does, yes.


        Stefan

Re: [Caml-list] Re: The need to specify 'rec' in a recursive function defintion

[Till Varoquaux]

On Wed, Feb 10, 2010 at 5:01 PM, Stefan Monnier
<monnier@iro.umontreal.ca> wrote:
>> Wouldn't one of way of detecting a recursive function would be to see
>> if the indeed the function calls itself?
>
> That's what Haskell does, yes.
>
>

Let's make things clear here: the "rec" *really* is a feature; it is
very convenient to reuse the same identifier to express something in
the process of being built (e.g. something going through a pipeline).

For instance:

let g () =
  let f s =
    if !debug then
       Printf.printf "f is called with value: %s\n%!" s
    f s
  in
  .......

Some (including me) would even argue that it is sad that type
definitions don't use "rec".

Till
>        Stefan
>
> _______________________________________________
> Caml-list mailing list. Subscription management:
> http://yquem.inria.fr/cgi-bin/mailman/listinfo/caml-list
> Archives: http://caml.inria.fr
> Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
> Bug reports: http://caml.inria.fr/bin/caml-bugs
>

Re: The need to specify 'rec' in a recursive function defintion

[Stefan Monnier]

>>> Wouldn't one of way of detecting a recursive function would be to see
>>> if the indeed the function calls itself?
>> That's what Haskell does, yes.
> Let's make things clear here: the "rec" *really* is a feature;

Nobody said otherwise.  Eliminating the "rec" is also a feature.
Those two features are mostly incompatible, and many reasonable people
disagree on which one of the two is more important.


        Stefan "who extensively used that feature in SML, but happens
                to prefer the other feature nevertheless"

Re: [Caml-list] Re: The need to specify 'rec' in a recursive function defintion

[Jon Harrop]

On Monday 15 February 2010 15:46:58 Stefan Monnier wrote:
> Till Varoquaux had written:
> > Let's make things clear here: the "rec" *really* is a feature;
>
> Nobody said otherwise.  Eliminating the "rec" is also a feature.
> Those two features are mostly incompatible, and many reasonable people
> disagree on which one of the two is more important.
>
>         Stefan "who extensively used that feature in SML, but happens
>                 to prefer the other feature nevertheless"

Standard ML doesn't have the feature that Till described.

-- 
Dr Jon Harrop, Flying Frog Consultancy Ltd.
http://www.ffconsultancy.com/?e

Re: The need to specify 'rec' in a recursive function defintion

[Stefan Monnier]

>> Till Varoquaux had written:
>> > Let's make things clear here: the "rec" *really* is a feature;
>> Nobody said otherwise.  Eliminating the "rec" is also a feature.
>> Those two features are mostly incompatible, and many reasonable people
>> disagree on which one of the two is more important.
>> Stefan "who extensively used that feature in SML, but happens
>> to prefer the other feature nevertheless"
> Standard ML doesn't have the feature that Till described.

It sure does, tho not with "fun" but only with "var" definitions.


        Stefan

Re: The need to specify 'rec' in a recursive function defintion

[Stefan Monnier]

> It sure does, tho not with "fun" but only with "var" definitions.
                                                  ^^^
                                                  val

        Stefan "blush!"