Re: Scope 2...

On Wed, Jan 29, 2003 at 08:51:54AM +0100, Fabrizio Morbini wrote:
> Hi, thank you very much for all the response to my previous mail!
> 
> Take this example:
> 
> (setq a 6)

Please stop using SETQ like this.  Use LET instead, like you did in the
very last example below.

> (flet (
>        (prova () (print a))
>        )
>       (let ((a 4))
> 	(print a)
> 	(prova)))
> 
> (print a)

I would appreciate if you would format your code properly.  Use Emacs or
similar.

(flet ((prova () (print a)))
  (let ((a 4))
    (print a)
    (prova)))

> The output is:
> 
> print: 4
> prova: 6
> print: 6
> 
> This is because the function prova isn't in the lexical environment
> created by let...
> 
> using defvar instead of setq the output is:
> 
> print: 4
> prova: 4
> print: 6
> 
> *Why the let hidden the dinamic scope variable "a" (defvar) but not when
> this is created by setq? My response is: because let start a new empty
> lexical environment that doesn't contain "a" so let can't hidden a
> variable that can't see... is correct?

No.  When LET is used on a special variable, such as that created with
DEFVAR, it does not create a lexical binding but rather it creates a
special binding.  This new binding is in effect for all subsequent code
(indefinite scope) until the LET form is exited (dynamic extent).

This is one of the reasons for the *VAR* convention of special
variables; so that you don't create a special binding with LET when you
really mean to create a lexical binding.

Consider:

(defvar *a* 1)

(defun g ()
  *a*)

(defun f ()
  (let ((*a* 2))  ; special binding established
    (g)))  ; call G with new special binding in effect
    
Compare the results of (f) and (g).

-- 
; Matthew Danish <[email protected]>
; OpenPGP public key: C24B6010 on keyring.debian.org
; Signed or encrypted mail welcome.
; "There is no dark side of the moon really; matter of fact, it's all dark."