Re: table.el

[Tak Ota]

Mon, 3 Dec 2001 01:46:54 -0700 (MST): Richard Stallman <address@hidden> wrote:

>     The reason why I needed wrapper mechanism was that I was advised by
>     you not to use advice in table.el, and that was because
>     `describe-function' of those advised functions mislead novice users to
>     table.el where there is no real information about the original
>     functions.  I personally dislike this shortcoming of
>     `describe-function'.
> 
> This could be a bug in advice.el.  Could you provide a simple test case?
> Perhaps advice.el should do something better to construct the doc string
> of the combined function.

I think the problem is rather in help.el or combination with
advice.el.


Normally C-h f `newline' says:

newline is an interactive compiled Lisp function in `simple'.

Now create a file ~/test.el which contents the following three lines.

(defadvice newline (around newline-wrapper last activate compile)
  "wrapps newline"
  ad-do-it)

Do M-x load-file ~/test.el and C-h f `newline' again.

Now the help message says:

newline is an interactive compiled Lisp function in `~/test.el'.


Same thing happens with table.el.  Before loading table.el

kill-region is an interactive compiled Lisp function in `simple'.

After loading table.el

kill-region is an interactive compiled Lisp function in `table'.


I've asked a question regarding this issue about a year ago on
address@hidden and didn't seem to get sufficient attention to
change anything.  See the attached messages.

-Tak
> --- Begin Message ---
>
>
>
> Subject: 
>
> a weakness in help.el?
>
>
>
>
> Date: 
>
> Mon, 20 Nov 2000 14:14:53 -0800 (PST)
>
>
>
>
> Under Emacs 20.7 `help-follow' fails.
>
> How to reproduce:
>
> 1. Make a test file say "test.el" which content is:
>
> ;; test.el begins here
> ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
> (require 'advice)
> (require 'tabify)
>
> (defadvice untabify (around advised-untabify last activate compile)
>   ad-do-it
>   (message "executed original `untabify'"))
>
> (defmacro test-advice (func)
>   (`
>    (defadvice (, func) (around advised-tabify last activate compile)
>      ad-do-it
>      (message "executed original `tabify'"))))
>
> (test-advice tabify)
> ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
> ;; test.el ends here
>
>
> 2. Start a fresh Emacs by "emacs -q -no-site-file".
>
> 3. M-x describe-function "tabify"
>
> 4. help-follow (hit RET) on the file name on the first line.
>
> 5. Get an error "Cannot find definition of `tabify' in library ..."
>
> While
>
> 6. M-x describe-function "untabify"
>
> 7. help-follow (hit RET) on the file name on the first line.
>
> 8. test.el opens in a buffer and point goes to where `untabify' is
>    advised.
>
>
> Is this an expected limitation in help.el?  How difficult is it to
> find an advice made by a macro?
>
> -Tak
>
> --- End Message ---
> --- Begin Message ---
>
>
>
> Subject: 
>
> Re: a weakness in help.el?
>
>
>
>
> Date: 
>
> 22 Nov 2000 22:27:36 +0000
>
>
>
>
> User-agent: 
>
> Gnus/5.09 (Gnus v5.9.0) Emacs/21.0.92
>
>
>
>
> Things are somewhat different in the current development code, but
> there are known problem with the load history and advice.  We didn't
> think they were important enough to work on, but we'd be happy to see
> someone else contribute a clean solution.
>
> --- End Message ---