[PATCH] eww-mode-map doesn't bind TAB on tty.

[PATCH] eww-mode-map doesn't bind TAB on tty.

[Mario Lang]

Hi.

Just built Emacs from trunk to check out eww.
It appears that the usage of [tab] and [backtab] does not work for my
TTY Emacs.  If I replace that with (kbd "TAB") and (kbd "M-TAB")
jumping between links works as expected on TTY Emacs.

--- a/lisp/net/eww.el
+++ b/lisp/net/eww.el
@@ -412,8 +412,8 @@ word(s) will be searched for via `eww-search-prefix'."
     (suppress-keymap map)
     (define-key map "q" 'quit-window)
     (define-key map "g" 'eww-reload)
-    (define-key map [tab] 'shr-next-link)
-    (define-key map [backtab] 'shr-previous-link)
+    (define-key map (kbd "TAB") 'shr-next-link)
+    (define-key map (kbd "M-TAB") 'shr-previous-link)
     (define-key map [delete] 'scroll-down-command)
     (define-key map [?\S-\ ] 'scroll-down-command)
     (define-key map "\177" 'scroll-down-command)

-- 
CYa,
  ⡍⠁⠗⠊⠕

Re: [PATCH] eww-mode-map doesn't bind TAB on tty.

[Lars Ingebrigtsen]

Mario Lang <mlang@delysid.org> writes:

> Just built Emacs from trunk to check out eww.
> It appears that the usage of [tab] and [backtab] does not work for my
> TTY Emacs.  If I replace that with (kbd "TAB") and (kbd "M-TAB")
> jumping between links works as expected on TTY Emacs.

[...]

> -    (define-key map [tab] 'shr-next-link)
> -    (define-key map [backtab] 'shr-previous-link)
> +    (define-key map (kbd "TAB") 'shr-next-link)
> +    (define-key map (kbd "M-TAB") 'shr-previous-link)

Hm.  I don't quite understand the patch.  [tab] should work in Emacs,
also under tty, shouldn't it?  But you're right -- [tab] doesn't work in
this case.

Does anybody know why, and is TAB a special case here?

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

Re: [PATCH] eww-mode-map doesn't bind TAB on tty.

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Date: Tue, 04 Mar 2014 20:39:29 +0100
> Cc: emacs-devel@gnu.org
> 
> > -    (define-key map [tab] 'shr-next-link)
> > -    (define-key map [backtab] 'shr-previous-link)
> > +    (define-key map (kbd "TAB") 'shr-next-link)
> > +    (define-key map (kbd "M-TAB") 'shr-previous-link)
> 
> Hm.  I don't quite understand the patch.  [tab] should work in Emacs,
> also under tty, shouldn't it?  But you're right -- [tab] doesn't work in
> this case.
> 
> Does anybody know why, and is TAB a special case here?

[tab] is a function key, not the character TAB.  Try ?\t instead.

Re: [PATCH] eww-mode-map doesn't bind TAB on tty.

[Eli Zaretskii]

> Date: Tue, 04 Mar 2014 22:07:49 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: mlang@delysid.org, emacs-devel@gnu.org
> 
> [tab] is a function key, not the character TAB.  Try ?\t instead.

And [?\M-\t] for backtab.

Re: [PATCH] eww-mode-map doesn't bind TAB on tty.

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

>> [tab] is a function key, not the character TAB.  Try ?\t instead.
>
> And [?\M-\t] for backtab.

Right; thanks.

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

Re: [PATCH] eww-mode-map doesn't bind TAB on tty.

[Mario Lang]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>>> [tab] is a function key, not the character TAB.  Try ?\t instead.
>>
>> And [?\M-\t] for backtab.
>
> Right; thanks.

`eww-text-map', `eww-textarea-map' and `shr-map' likely have the same
problem as `eww-mode-map' used to have:

--- a/lisp/net/eww.el
+++ b/lisp/net/eww.el
@@ -592,8 +592,8 @@ appears in a <link> or <a> tag."
     (define-key map [(control a)] 'eww-beginning-of-text)
     (define-key map [(control c) (control c)] 'eww-submit)
     (define-key map [(control e)] 'eww-end-of-text)
-    (define-key map [tab] 'shr-next-link)
-    (define-key map [backtab] 'shr-previous-link)
+    (define-key map [?\t] 'shr-next-link)
+    (define-key map [?\M-\t] 'shr-previous-link)
     map))
 
 (defvar eww-textarea-map
@@ -601,8 +601,8 @@ appears in a <link> or <a> tag."
     (set-keymap-parent map text-mode-map)
     (define-key map "\r" 'forward-line)
     (define-key map [(control c) (control c)] 'eww-submit)
-    (define-key map [tab] 'shr-next-link)
-    (define-key map [backtab] 'shr-previous-link)
+    (define-key map [?\t] 'shr-next-link)
+    (define-key map [?\M-\t] 'shr-previous-link)
     map))
 
 (defvar eww-select-map
--- a/lisp/net/shr.el
+++ b/lisp/net/shr.el
@@ -147,8 +147,8 @@ cid: URL as the argument.")
     (define-key map "a" 'shr-show-alt-text)
     (define-key map "i" 'shr-browse-image)
     (define-key map "z" 'shr-zoom-image)
-    (define-key map [tab] 'shr-next-link)
-    (define-key map [backtab] 'shr-previous-link)
+    (define-key map [?\t] 'shr-next-link)
+    (define-key map [?\M-\t] 'shr-previous-link)
     (define-key map [follow-link] 'mouse-face)
     (define-key map [mouse-2] 'shr-browse-url)
     (define-key map "I" 'shr-insert-image)


And here is the result of grepping for the pattern in all Emacs Lisp files:

lisp/calendar/todo-mode.el:    (define-key map [tab] 'todo-next-button)
lisp/eshell/em-cmpl.el:  (define-key eshell-command-map [tab] 'pcomplete-expand-and-complete)
lisp/eshell/em-cmpl.el:  (define-key eshell-mode-map [tab] 'eshell-pcomplete)
lisp/gnus/gnus-html.el:    (define-key map [tab] 'widget-forward)
lisp/image-dired.el:  (define-key image-dired-thumbnail-mode-map [tab] 'image-dired-jump-original-dired-buffer)
lisp/image-dired.el:  (define-key dired-mode-map [tab] 'image-dired-jump-thumbnail-buffer)
lisp/international/quail.el:    (define-key map [tab] 'quail-completion)
lisp/kmacro.el:    (define-key map [tab] 'act-repeat)
lisp/play/5x5.el:    (define-key map [tab]                     #'5x5-right)

-- 
CYa,
  ⡍⠁⠗⠊⠕

Re: [PATCH] eww-mode-map doesn't bind TAB on tty.

[Eli Zaretskii]

> From: Mario Lang <mlang@delysid.org>
> Date: Wed, 05 Mar 2014 16:02:54 +0100
> Cc: emacs-devel@gnu.org
> 
> And here is the result of grepping for the pattern in all Emacs Lisp files:
> 
> lisp/calendar/todo-mode.el:    (define-key map [tab] 'todo-next-button)
> lisp/eshell/em-cmpl.el:  (define-key eshell-command-map [tab] 'pcomplete-expand-and-complete)
> lisp/eshell/em-cmpl.el:  (define-key eshell-mode-map [tab] 'eshell-pcomplete)
> lisp/gnus/gnus-html.el:    (define-key map [tab] 'widget-forward)
> lisp/image-dired.el:  (define-key image-dired-thumbnail-mode-map [tab] 'image-dired-jump-original-dired-buffer)
> lisp/image-dired.el:  (define-key dired-mode-map [tab] 'image-dired-jump-thumbnail-buffer)
> lisp/international/quail.el:    (define-key map [tab] 'quail-completion)
> lisp/kmacro.el:    (define-key map [tab] 'act-repeat)
> lisp/play/5x5.el:    (define-key map [tab]                     #'5x5-right)

Thanks for the footwork.  I think image-dired is the only one that
might not need changing, because thumbnails are only supported on GUI
terminals which do have TAB as a function key.

Re: [PATCH] eww-mode-map doesn't bind TAB on tty.

[Lars Ingebrigtsen]

Mario Lang <mlang@delysid.org> writes:

> And here is the result of grepping for the pattern in all Emacs Lisp files:

Thanks; I'll replace all those (except the image-dired one)...

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

Re: [PATCH] eww-mode-map doesn't bind TAB on tty.

[Lars Ingebrigtsen]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Mario Lang <mlang@delysid.org> writes:
>
>> And here is the result of grepping for the pattern in all Emacs Lisp files:
>
> Thanks; I'll replace all those (except the image-dired one)...

Actually...  before I do that...

Some modes define both [?\M-\t] (or "\e\t"), which are the same thing, I
think, as well as [backtab].  Same for [tab].  Like:

    (define-key map "\t" 'widget-forward)
    (define-key map "\e\t" 'widget-backward)
    (define-key map [(shift tab)] 'widget-backward)
    (put 'widget-backward :advertised-binding [(shift tab)])
    (define-key map [backtab] 'widget-backward)

Are there situations where just removing the latter binding is wrong?
I.e., is there a difference between [?\M-\t] and [backtab] that should
be preserved?

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

Re: [PATCH] eww-mode-map doesn't bind TAB on tty.

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Date: Wed, 05 Mar 2014 19:10:01 +0100
> 
>     (define-key map "\t" 'widget-forward)
>     (define-key map "\e\t" 'widget-backward)
>     (define-key map [(shift tab)] 'widget-backward)
>     (put 'widget-backward :advertised-binding [(shift tab)])
>     (define-key map [backtab] 'widget-backward)
> 
> Are there situations where just removing the latter binding is wrong?
> I.e., is there a difference between [?\M-\t] and [backtab] that should
> be preserved?

If both are present, I don't think you should delete one of them.
E.g., some terminal might define an escape sequence to [backtab].

Re: [PATCH] eww-mode-map doesn't bind TAB on tty.

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Lars Ingebrigtsen <larsi@gnus.org>
>> Date: Wed, 05 Mar 2014 19:10:01 +0100
>> 
>>     (define-key map "\t" 'widget-forward)
>>     (define-key map "\e\t" 'widget-backward)
>>     (define-key map [(shift tab)] 'widget-backward)
>>     (put 'widget-backward :advertised-binding [(shift tab)])
>>     (define-key map [backtab] 'widget-backward)
>> 
>> Are there situations where just removing the latter binding is wrong?
>> I.e., is there a difference between [?\M-\t] and [backtab] that should
>> be preserved?
>
> If both are present, I don't think you should delete one of them.
> E.g., some terminal might define an escape sequence to [backtab].

Hm.  Should we always have two bindings each for [tab] and [backtab],
then?  This all seems kinda inconsistent.

And should perhaps wait until after the feature freeze, anyway.

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

Re: [PATCH] eww-mode-map doesn't bind TAB on tty.

[Stefan Monnier]

> Hm.  Should we always have two bindings each for [tab] and [backtab],
> then?

No.  We should basically always use just TAB (aka ?\t).  `tab' should
only be used in special cases where it's justified with a comment
explaining why it's needed.


        Stefan