labels usage

labels usage

[Katsumi Yamaoka]

Hi,

Is this a wrong usage?

(let (fn)
  (labels ((fx (arg) arg))
    (setq fn 'fx))
  (funcall fn 99))

It worked until cl changed.  Now this labels form is expanded into:

(progn
  (defvar --cl---cl-fx----)
  (let ((--cl---cl-fx---- nil))
    (progn (set '--cl---cl-fx---- #'(lambda (arg) arg))
	   (setq fn 'fx))))

Whereas it was:

(let ((--cl---cl-var--86211-- nil))
  (progn (set '--cl---cl-var--86211-- #'(lambda (arg) arg))
	 (setq fn (symbol-value '--cl---cl-var--86211--))))

Did we get to have no way to access the function definition, that
a labels form generates, in the outside of the labels form?
(If so, we have a solution; we have to simply replace labels with
 let in such a Lisp code, i.e., emacs-w3m.)

Re: labels usage

[Andreas Schwab]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

> Hi,
>
> Is this a wrong usage?
>
> (let (fn)
>   (labels ((fx (arg) arg))
>     (setq fn 'fx))
>   (funcall fn 99))

You should use #'fx instead.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: labels usage

[Katsumi Yamaoka]

In [emacs-w3m : No.11847] Andreas Schwab wrote:
> Katsumi Yamaoka <yamaoka@jpl.org> writes:
>> Is this a wrong usage?
>>
>> (let (fn)
>>   (labels ((fx (arg) arg))
>>     (setq fn 'fx))
>>   (funcall fn 99))

> You should use #'fx instead.

Ok, we *should* use #' from now on.  This was what Michael Heerdegen
wrote in the emacs-w3m list a while ago.  Maybe it didn't go well
when I tried it was I forgot to build emacs-w3m from scratch.  I'll
try it again.  Thanks.

Re: labels usage

[Stefan Monnier]

>   (labels ((fx (arg) arg))
>     (setq fn 'fx))

'fx evaluates to the *symbol* `fx', not the *function* `fx'.  In Elisp
until Emacs-24, you can use pretty much interchangeably a function and
the symbol to which it is bound, but with lexical scoping this is not
true any more.  CL's `labels' used to follow the "Elisp tradition" and
looked inside `quote' to turn that symbol into a reference to the
locally defined function.  But that was fundamentally a bug.


        Stefan