bug#33314: 27.0.50; wrong compiler warning

bug#33314: 27.0.50; wrong compiler warning

[Andreas Röhler]

Get a compiler Warning: Unused lexical variable ‘regexp’

WRT a form like below:

(defun foo (regexp &optional something)
   "..."
   (unless (bobp)
     (when (empty-line-p) (skip-chars-backward " \t\r\n\f"))
     (let* ((orig (point))
	   (regexpvalue (if (eq regexp 'py-clause-re) (symbol-value 
'py-extended-block-or-clause-re) (symbol-value regexp)))
...

;;

As visible, regexp is evaluated in last clause.

Thanks,
Andreas
GNU Emacs 27.0.50 (build 1, i686-pc-linux-gnu, GTK+ Version 3.14.5) of 
2018-10-13

bug#33314: 27.0.50; wrong compiler warning

[Alan Mackenzie]

Hello, Andreas.

In article <mailman.3649.1541688141.1284.bug-gnu-emacs@gnu.org> you wrote:
> Get a compiler Warning: Unused lexical variable ‘regexp’

> WRT a form like below:

> (defun foo (regexp &optional something)
>   "..."
>   (unless (bobp)
>     (when (empty-line-p) (skip-chars-backward " \t\r\n\f"))
>     (let* ((orig (point))
>           (regexpvalue (if (eq regexp 'py-clause-re) (symbol-value 
> 'py-extended-block-or-clause-re) (symbol-value regexp)))
> ...

Reformatting that to make things clearer:

(defun foo (regexp &optional something)
  "..."
  (unless (bobp)
    (when (empty-line-p) (skip-chars-backward " \t\r\n\f"))
    (let* ((orig (point))
           (regexpvalue
            (if (eq regexp 'py-clause-re)
                (symbol-value 'py-extended-block-or-clause-re)
              (symbol-value regexp)))

> ;;

> As visible, regexp is evaluated in last clause.

Does the rest of that let* form use regexpvalue?  If not, the lack of
use of regexpvalue will, I believe, transfer to regexp.

Incidentally, why write (symbol-value 'py-extended-block-or-clause-re)
rather than just py-extended-block-or-clause-re?

> Thanks,
> Andreas
> GNU Emacs 27.0.50 (build 1, i686-pc-linux-gnu, GTK+ Version 3.14.5) of 
> 2018-10-13

-- 
Alan Mackenzie (Nuremberg, Germany).

bug#33314: 27.0.50; wrong compiler warning

[Andreas Röhler]

On 08.11.2018 16:54, Alan Mackenzie wrote:
> Hello, Andreas.
> 
> In article <mailman.3649.1541688141.1284.bug-gnu-emacs@gnu.org> you wrote:
>> Get a compiler Warning: Unused lexical variable ‘regexp’
> 
>> WRT a form like below:
>
> Reformatting that to make things clearer:
> 
> (defun foo (regexp &optional something)
>    "..."
>    (unless (bobp)
>      (when (empty-line-p) (skip-chars-backward " \t\r\n\f"))
>      (let* ((orig (point))
>             (regexpvalue
>              (if (eq regexp 'py-clause-re)
>                  (symbol-value 'py-extended-block-or-clause-re)
>                (symbol-value regexp)))
> 
>> ;;
> 
>> As visible, regexp is evaluated in last clause.
> 
> Does the rest of that let* form use regexpvalue?  If not, the lack of
> use of regexpvalue will, I believe, transfer to regexp.
> 
> Incidentally, why write (symbol-value 'py-extended-block-or-clause-re)
> rather than just py-extended-block-or-clause-re?
> 

Hi Alan,

that form is used to make a decision at a certain point:

(cond ((and
		     (member
		      regexp
		      (list
		       'py-block-or-clause-re
		   ...
		       ))
		     (looking-at regexpvalue))

So regexpvalue is used too. May send the complete function if interested.

bug#33314: 27.0.50; wrong compiler warning

[Noam Postavsky]

tags 33314 unreproducible moreinfo
quit

Andreas Röhler <andreas.roehler@online.de> writes:

> So regexpvalue is used too. May send the complete function if interested.

Please do, I can't reproduce the warning you claimed in your original
report (and frankly, I think you've been doing this long enough that you
should know by now that such an abbreviated report is not very helpful).

Trying to fill in your example like this:

    ;;; -*- lexical-binding: t -*-

    (defun foo (regexp &optional something)
       "..."
       (unless (bobp)
         (when (empty-line-p) (skip-chars-backward " \t\r\n\f"))
         (let* ((orig (point))
                (regexpvalue (if (eq regexp 'py-clause-re)
                                 (symbol-value
                                  'py-extended-block-or-clause-re)
                               (symbol-value regexp))))
           t)))

gives

    bug-33314-bcomp-warning.el:3:1:Warning: Unused lexical variable ‘regexpvalue’
    bug-33314-bcomp-warning.el:3:1:Warning: Unused lexical variable ‘orig’
    bug-33314-bcomp-warning.el:3:1:Warning: Unused lexical argument ‘something’

    In end of data:
    bug-33314-bcomp-warning.el:13:1:Warning: the function ‘empty-line-p’ is not
        known to be defined.

`regexp' is not mentioned, but the other unused variables are, as
expected.

bug#33314: 27.0.50; wrong compiler warning

[Andreas Röhler]

On 09.11.2018 02:11, Noam Postavsky wrote:
> tags 33314 unreproducible moreinfo
> quit
> 
> Andreas Röhler <andreas.roehler@online.de> writes:
> 
>> So regexpvalue is used too. May send the complete function if interested.
> 
> Please do, I can't reproduce the warning you claimed in your original
> report (and frankly, I think you've been doing this long enough that you
> should know by now that such an abbreviated report is not very helpful).
> 
> Trying to fill in your example like this:
> 
>      ;;; -*- lexical-binding: t -*-
> 
>      (defun foo (regexp &optional something)
>         "..."
>         (unless (bobp)
>           (when (empty-line-p) (skip-chars-backward " \t\r\n\f"))
>           (let* ((orig (point))
>                  (regexpvalue (if (eq regexp 'py-clause-re)
>                                   (symbol-value
>                                    'py-extended-block-or-clause-re)
>                                 (symbol-value regexp))))
>             t)))
> 
> gives
> 
>      bug-33314-bcomp-warning.el:3:1:Warning: Unused lexical variable ‘regexpvalue’
>      bug-33314-bcomp-warning.el:3:1:Warning: Unused lexical variable ‘orig’
>      bug-33314-bcomp-warning.el:3:1:Warning: Unused lexical argument ‘something’
> 
>      In end of data:
>      bug-33314-bcomp-warning.el:13:1:Warning: the function ‘empty-line-p’ is not
>          known to be defined.
> 
> `regexp' is not mentioned, but the other unused variables are, as
> expected.
> 

Hmm, sorry, can't reproduce it any more. As this happens from time to 
time, should send a complete report indeed. Tried to replay for now, but 
couldn't find a saved state.

Thanks for your care,

Andreas