Re: empty-line-p

[Andreas Roehler]

Xavier Maillard schrieb:
>    >  wrap it in a `save-match-data'.
>    >
>    > David
>    >   
>
>    OK, thanks. So I got this now:
>
> Here is what I would do instead:
>
> (defun empty-line-p (&optional ispec)
>   "Returns t if cursor is at an empty line, nil otherwise.
> Displays result in minibuffer when called interactive."
>   (interactive "p")
>   (save-excursion
>     (beginning-of-line)
>     (save-match-data
>       (let ((res (looking-at empty-line-p-chars)))
>         (or (and ispec
>                  (message "%s" res))
>             res)))))
>
>
> Xavier

That would probably work, however, as far as the form
is conceived as a subroutine, it doesn't seem the best
solution.

Assume you introduced `let', because in

   (when ispec
     (message "%s" (looking-at empty-line-p-chars)))
     (looking-at empty-line-p-chars))))

`looking-at' is performed two times.

Thought that repeat won't matter, because when called
interactively, time isn't at stake.

Otherwise `let' would maybe called again and
again from inside a program, whereas `let' isn't really
needed here.

Meanwhile thought to reconcile
`save-match-data' proposed by David Hansen with Stefan
Monnier's objections concerning speed.

The form below should `save-match-data', when called as

(empty-line-p t)

and without arg not.

(defun empty-line-p (&optional arg ispec)
 "Returns t if cursor is at an empty line, nil otherwise.
Displays result in minibuffer when called interactive."
 (interactive "P\np")
 (save-excursion
   (beginning-of-line)
   (if arg
   (save-match-data
     (if ispec
         (message "%s" (looking-at empty-line-p-chars))
       (looking-at empty-line-p-chars)))
     (if ispec
     (message "%s" (looking-at empty-line-p-chars))
   (looking-at empty-line-p-chars)))))

(defcustom empty-line-p-chars "^[ \t\f\r]*$"
 "empty-line-p-chars"
:type 'regexp
:group 'convenience)


Thanks all

Andreas Roehler