Re: [elpa] externals/corfu 6572247 3/4: Bind [tab] and [return]

Re: [elpa] externals/corfu 6572247 3/4: Bind [tab] and [return]

[Stefan Monnier]

> +    (define-key map [tab] #'corfu-complete)

Please avoid binding `tab`: the `tab` event (only generated under GUIs)
is supposed to be remapped to the TAB char-event (aka `C-i`) and this is
what you should bind to if you want your binding to work both under ttys
and GUIs.

More importantly, if you bind to `tab` than this binding will take
precedence over all other bindings to TAB, even those in keymaps that
have higher precedence.

The same holds for `return` vs RET, and `escape` vs ESC.


        Stefan

Re: [elpa] externals/corfu 6572247 3/4: Bind [tab] and [return]

[Daniel Mendler]

On 6/26/21 4:02 PM, Stefan Monnier wrote:
>> +    (define-key map [tab] #'corfu-complete)
> 
> Please avoid binding `tab`: the `tab` event (only generated under GUIs)
> is supposed to be remapped to the TAB char-event (aka `C-i`) and this is
> what you should bind to if you want your binding to work both under ttys
> and GUIs.
> 
> More importantly, if you bind to `tab` than this binding will take
> precedence over all other bindings to TAB, even those in keymaps that
> have higher precedence.
> 
> The same holds for `return` vs RET, and `escape` vs ESC.

Hello Stefan,

I am aware of the unfortunate implications of these bindings.
Originally I avoided these bindings for the exact reasons you mention
and I hoped I could do without those (only binding RET/TAB and
remappings). However I have to use these keys since Org-mode seems to
override these keys too. Otherwise my keymap will not take precedence
over the Org-mode keymap.

Daniel

Re: [elpa] externals/corfu 6572247 3/4: Bind [tab] and [return]

[Stefan Monnier]

> remappings). However I have to use these keys since Org-mode seems to
> override these keys too.

Please report this as a bug.


        Stefan

org-mode-map binds [tab]

[Daniel Mendler]

`org-mode-map` binds `[tab]` which is unnecessary and harmful, since it
takes precendence over bindings of TAB even in keymaps with higher
precedence.

Emacs  : GNU Emacs 27.1 (build 1, x86_64-pc-linux-gnu, GTK+ Version
3.24.5, cairo version 1.16.0)
 of 2021-02-09, modified by Debian
Package: Org mode version 9.4.6 (9.4.6-gab9f2a @
/home/user/.config/emacs/elpa/org-9.4.6/)

----

I've observed this problem with my GNU ELPA package Corfu, which
installs a keymap with higher precedence than org-mode. In order to
override Org, it also has to bind [tab], which is undesired.

On 6/27/21 11:11 AM, Daniel Mendler wrote:
> On 6/26/21 4:02 PM, Stefan Monnier wrote:
>>> +    (define-key map [tab] #'corfu-complete)
>>
>> Please avoid binding `tab`: the `tab` event (only generated under GUIs)
>> is supposed to be remapped to the TAB char-event (aka `C-i`) and this is
>> what you should bind to if you want your binding to work both under ttys
>> and GUIs.
>>
>> More importantly, if you bind to `tab` than this binding will take
>> precedence over all other bindings to TAB, even those in keymaps that
>> have higher precedence.
>>
>> The same holds for `return` vs RET, and `escape` vs ESC.
> 
> Hello Stefan,
> 
> I am aware of the unfortunate implications of these bindings.
> Originally I avoided these bindings for the exact reasons you mention
> and I hoped I could do without those (only binding RET/TAB and
> remappings). However I have to use these keys since Org-mode seems to
> override these keys too. Otherwise my keymap will not take precedence
> over the Org-mode keymap.
> 
> Daniel
> 

Re: org-mode-map binds [tab]

[Nicolas Goaziou]

Hello,

Daniel Mendler <mail@daniel-mendler.de> writes:

> `org-mode-map` binds `[tab]` which is unnecessary and harmful, since it
> takes precendence over bindings of TAB even in keymaps with higher
> precedence.

Thou shalt not have precedence over Org mode!

Fixed.

Thank you!

Regards,
-- 
Nicolas Goaziou

Re: org-mode-map binds [tab]

[Barton, Mark]

With Emacs compiled from master, after Nicolas Goaziou's fix to remove the <tab> binding, my tab key was no longer bound to org-cycle, but instead to outline-cycle from outline.el

If I comment out line 185 in outline.el then TAB will bind to org-cycle.

Does outline.el need to be fixed too?

——snippet from outline.el
(defvar outline-mode-cycle-map
 (let ((map (make-sparse-keymap)))
   (let ((tab-binding `(menu-item
                        "" outline-cycle
                        ;; Only takes effect if point is on a heading.
                        :filter ,(lambda (cmd)
                                   (when (outline-on-heading-p) cmd)))))
     (define-key map [tab]       tab-binding)
     (define-key map (kbd "TAB") tab-binding)
     (define-key map (kbd "<backtab>") #'outline-cycle-buffer))
   map)
 "Keymap used by `outline-mode-map' and `outline-minor-mode-cycle'.")


I don’t see outline as a minor mode listed when I use C-h m while in an org file expecting org-cycle.

Mark

> On Jun 27, 2021, at 8:34 AM, Daniel Mendler <mail@daniel-mendler.de> wrote:
> 
> `org-mode-map` binds `[tab]` which is unnecessary and harmful, since it
> takes precendence over bindings of TAB even in keymaps with higher
> precedence.
> 
> Emacs  : GNU Emacs 27.1 (build 1, x86_64-pc-linux-gnu, GTK+ Version
> 3.24.5, cairo version 1.16.0)
> of 2021-02-09, modified by Debian
> Package: Org mode version 9.4.6 (9.4.6-gab9f2a @
> /home/user/.config/emacs/elpa/org-9.4.6/)
> 
> ----
> 
> I've observed this problem with my GNU ELPA package Corfu, which
> installs a keymap with higher precedence than org-mode. In order to
> override Org, it also has to bind [tab], which is undesired.
> 
> On 6/27/21 11:11 AM, Daniel Mendler wrote:
>> On 6/26/21 4:02 PM, Stefan Monnier wrote:
>>>> +    (define-key map [tab] #'corfu-complete)
>>> 
>>> Please avoid binding `tab`: the `tab` event (only generated under GUIs)
>>> is supposed to be remapped to the TAB char-event (aka `C-i`) and this is
>>> what you should bind to if you want your binding to work both under ttys
>>> and GUIs.
>>> 
>>> More importantly, if you bind to `tab` than this binding will take
>>> precedence over all other bindings to TAB, even those in keymaps that
>>> have higher precedence.
>>> 
>>> The same holds for `return` vs RET, and `escape` vs ESC.
>> 
>> Hello Stefan,
>> 
>> I am aware of the unfortunate implications of these bindings.
>> Originally I avoided these bindings for the exact reasons you mention
>> and I hoped I could do without those (only binding RET/TAB and
>> remappings). However I have to use these keys since Org-mode seems to
>> override these keys too. Otherwise my keymap will not take precedence
>> over the Org-mode keymap.
>> 
>> Daniel
>> 
> 
> 

Re: org-mode-map binds [tab]

[Stefan Monnier]

> Does outline.el need to be fixed too?
[...]
> ——snippet from outline.el
> (defvar outline-mode-cycle-map
>  (let ((map (make-sparse-keymap)))
>    (let ((tab-binding `(menu-item
>                         "" outline-cycle
>                         ;; Only takes effect if point is on a heading.
>                         :filter ,(lambda (cmd)
>                                    (when (outline-on-heading-p) cmd)))))
>      (define-key map [tab]       tab-binding)
>      (define-key map (kbd "TAB") tab-binding)
>      (define-key map (kbd "<backtab>") #'outline-cycle-buffer))
>    map)
>  "Keymap used by `outline-mode-map' and `outline-minor-mode-cycle'.")

Very much so, indeed, thanks.
I just pushed that change to `master`.


        Stefan

Bug: org-mode-map binds tab [9.4.6 (9.4.6-gab9f2a @ /home/user/.config/emacs/elpa/org-9.4.6/)]

[Daniel Mendler]

`org-mode-map` binds `[tab]` which is unnecessary and harmful, since it
takes precendence over bindings of TAB even in keymaps with higher
precedence.

Emacs  : GNU Emacs 27.1 (build 1, x86_64-pc-linux-gnu, GTK+ Version
3.24.5, cairo version 1.16.0)
 of 2021-02-09, modified by Debian
Package: Org mode version 9.4.6 (9.4.6-gab9f2a @
/home/user/.config/emacs/elpa/org-9.4.6/)

----

I've observed this problem with my GNU ELPA package Corfu, which
installs a keymap with higher precedence than org-mode. In order to
override Org, it also has to bind [tab], which is undesired.

On 6/27/21 11:11 AM, Daniel Mendler wrote:
> On 6/26/21 4:02 PM, Stefan Monnier wrote:
>>> +    (define-key map [tab] #'corfu-complete)
>>
>> Please avoid binding `tab`: the `tab` event (only generated under GUIs)
>> is supposed to be remapped to the TAB char-event (aka `C-i`) and this is
>> what you should bind to if you want your binding to work both under ttys
>> and GUIs.
>>
>> More importantly, if you bind to `tab` than this binding will take
>> precedence over all other bindings to TAB, even those in keymaps that
>> have higher precedence.
>>
>> The same holds for `return` vs RET, and `escape` vs ESC.
> 
> Hello Stefan,
> 
> I am aware of the unfortunate implications of these bindings.
> Originally I avoided these bindings for the exact reasons you mention
> and I hoped I could do without those (only binding RET/TAB and
> remappings). However I have to use these keys since Org-mode seems to
> override these keys too. Otherwise my keymap will not take precedence
> over the Org-mode keymap.
> 
> Daniel
>