Curried form different from "normal"

Curried form different from "normal"

[Till Varoquaux]

I can across something that puzzled me while develloping a caml application:

Using:
 let non_blank_string=
      if !is_first then
	trailling:=Buffer.contents white
      else
	Buffer.add_buffer buf white;
      Buffer.reset white;
      is_first:=false;
      Buffer.add_string buf

instead of

 let non_blank_string s=
      if !is_first then
	trailling:=Buffer.contents white
      else
	Buffer.add_buffer buf white;
      Buffer.reset white;
      is_first:=false;
      Buffer.add_string buf s

Doesn't yeld the same results (the former is buggy). Now, as I
understand, these two functions should be exactly equivalent. I am
misunderstanding something here?

Cheers,

Till Varoquaux

Re: [Caml-list] Curried form different from "normal"

[David Baelde]

Hi,

Your statement is true only for purely functional definitions. Here,
there is a difference. The following piece of code is evaluated only
one, at the time of the outermost definition. So trailing is
instantiated only once.

>       if !is_first then
>         trailling:=Buffer.contents white
>       else
>         Buffer.add_buffer buf white;
>       Buffer.reset white;
>       is_first:=false;

This kind of effect is sometimes expected, however.

Have fun.
--
David

Re: [Caml-list] Curried form different from "normal"

[Jon Harrop]

On Sunday 05 February 2006 16:40, Till Varoquaux wrote:
>  let non_blank_string=
>       if !is_first then
> 	trailling:=Buffer.contents white
>       else
> 	Buffer.add_buffer buf white;
>       Buffer.reset white;
>       is_first:=false;
>       Buffer.add_string buf
>
>  let non_blank_string s=
>       if !is_first then
> 	trailling:=Buffer.contents white
>       else
> 	Buffer.add_buffer buf white;
>       Buffer.reset white;
>       is_first:=false;
>       Buffer.add_string buf s
>
> Doesn't yeld the same results (the former is buggy). Now, as I
> understand, these two functions should be exactly equivalent. I am
> misunderstanding something here?

The code preceding the call to "Buffer.add_string" is only evaluated once in 
the former case (at the time of definition) whereas it is evaluated at each 
invocation of the latter definition of "non_blank_string".

You should not have expected these two definitions to be "exactly equivalent" 
as the former is equivalent to:

 if !is_first then
   trailling:=Buffer.contents white
 else
   Buffer.add_buffer buf white;;

 Buffer.reset white;;

 is_first:=false;;

 let non_blank_string =
      Buffer.add_string buf

-- 
Dr Jon D Harrop, Flying Frog Consultancy Ltd.
Objective CAML for Scientists
http://www.ffconsultancy.com/products/ocaml_for_scientists

Re: [Caml-list] Curried form different from "normal"

[Till Varoquaux]

Yes, I apologize for bugging all of you. This makes sense and I could
(should) have figured it out by myself.

As david pointed out the equivalence is only true for purely
functional code. I guess this is one of the consequences of flirting
to much with imperative features...

Till Varoquaux