Re: Weired side-effect of using the remap feature in keymaps

Re: Weired side-effect of using the remap feature in keymaps

[Richard Stallman]

    1. Start Emacs with emacs -Q

    2. Evaluate the following Lisp code:

	     ;; Define a keymap
	     (defvar my-dummy-map (make-keymap))

	     ;; Now use the remap feature to remap delete-backward-char
	     (define-key my-dummy-map
	       (vector 'remap 'delete-backward-char) 'backward-char)

	     ;; Create a buffer and actually install that map.  This step
	     ;; seems to be necessary for the bug to occcur.
	     (switch-to-buffer (get-buffer-create "foo"))
	     (use-local-map my-dummy-map)

    3. Now, start calc with `M-x calc RET'.

    4. If you now check the definition of the DEL key, for example with
	`C-h k DEL', you will see that in calc-mode, this key is now bound
	to `backward-char', even though the remapping was done in
	`my-dummy-map'.  When calc binds the DEL key, it actually scans the
	global map for any key bound to `delete-backward-char' using

	  (where-is-internal 'delete-backward-char global-map)

	This search seems to fail after the code above, i.e. something
	seems to have changed in global map which should not have changed.

Nothing has changed in global-map.  When I do the same call to
where-is-internal in some other buffer, it correctly returns ([127]).

So it seems there is a bug in where-is-internal.  When called that way,
it ought to ignore the local map, but it does not.

Can someone fix this bug, then ack?

Re: Weired side-effect of using the remap feature in keymaps

[Carsten Dominik]

On Mar 30, 2007, at 23:23, Richard Stallman wrote:

>     1. Start Emacs with emacs -Q
>
>     2. Evaluate the following Lisp code:
>
> 	     ;; Define a keymap
> 	     (defvar my-dummy-map (make-keymap))
>
> 	     ;; Now use the remap feature to remap delete-backward-char
> 	     (define-key my-dummy-map
> 	       (vector 'remap 'delete-backward-char) 'backward-char)
>
> 	     ;; Create a buffer and actually install that map.  This step
> 	     ;; seems to be necessary for the bug to occcur.
> 	     (switch-to-buffer (get-buffer-create "foo"))
> 	     (use-local-map my-dummy-map)
>
>     3. Now, start calc with `M-x calc RET'.
>
>     4. If you now check the definition of the DEL key, for example with
> 	`C-h k DEL', you will see that in calc-mode, this key is now bound
> 	to `backward-char', even though the remapping was done in
> 	`my-dummy-map'.  When calc binds the DEL key, it actually scans the
> 	global map for any key bound to `delete-backward-char' using
>
> 	  (where-is-internal 'delete-backward-char global-map)
>
> 	This search seems to fail after the code above, i.e. something
> 	seems to have changed in global map which should not have changed.
>
> Nothing has changed in global-map.  When I do the same call to
> where-is-internal in some other buffer, it correctly returns ([127]).

I forgot to mention this:  I tried the same, and also found that it
would return ([127]).  However, when running the *compiled* calc.elc,
it returns nil.

- Carsten

Re: Weired side-effect of using the remap feature in keymaps

[Richard Stallman]

    >
    > Nothing has changed in global-map.  When I do the same call to
    > where-is-internal in some other buffer, it correctly returns ([127]).

    I forgot to mention this:  I tried the same, and also found that it
    would return ([127]).  However, when running the *compiled* calc.elc,
    it returns nil.

I get the correct result in other buffers,
and the wrong result in the buffer that your example creates
which uses my-dummy-map.