Proposal: merge [S-tab] and [backtab]

Proposal: merge [S-tab] and [backtab]

[Paul Griepentrog]

  Just a tidbit of computer evolution: the [backtab] key does not
exist on modern keyboards.  Of course, Emacs uses this key in
several modes (grep, compile, erc, ...).  So, in the place of
[backtab] within Emacs, [S-tab] is the conventional binding for a
poor man's backtab.  For convenience, some terminals (X,
Windows, and recently: NextStep) translate the keypress of
[S-tab] into [backtab] and nobody is the wizer.

Unless... you want to bind [S-tab] in a mode.  Then you find out
[S-tab] is NOT the key sequence you thought it was.  The binding
worked on X and Windows, but not at the terminal (or another
platform which did not link [S-tab] to [backtab]).  What
happened?!  Well, so you bind [backtab] as well.

My proposal is to universally translate [S-tab] and [S-iso-tab]
into [backtab] across all terminals.  At least then you can bind
[backtab] and know it will work across platforms and terminals.
Included in this proposal is removing obsolete bindings to
[S-tab] and [S-iso-tab] for modes within Emacs and updating the
documentation.  I include [S-iso-tab] since it is yet another
abused synonym for [backtab].

Note: this change will NOT ADD or REMOVE any bindings.  It is to
unify the keypresses of [S-tab], [S-iso-tab] to [backtab] where
they already exist.

I don't pretend to know the history of the keybindings or
specific terminals, but a uniform view of [S-tab] vs [backtab]
makes sense to me.

Thoughts?

Re: Proposal: merge [S-tab] and [backtab]

[Leo]

On 2010-08-02 05:22 +0100, Paul Griepentrog wrote:
>  Just a tidbit of computer evolution: the [backtab] key does not
> exist on modern keyboards.  Of course, Emacs uses this key in
> several modes (grep, compile, erc, ...).  So, in the place of
> [backtab] within Emacs, [S-tab] is the conventional binding for a
> poor man's backtab.  For convenience, some terminals (X,
> Windows, and recently: NextStep) translate the keypress of
> [S-tab] into [backtab] and nobody is the wizer.
>
> Unless... you want to bind [S-tab] in a mode.  Then you find out
> [S-tab] is NOT the key sequence you thought it was.  The binding
> worked on X and Windows, but not at the terminal (or another
> platform which did not link [S-tab] to [backtab]).  What
> happened?!  Well, so you bind [backtab] as well.
>
> My proposal is to universally translate [S-tab] and [S-iso-tab]
> into [backtab] across all terminals.  At least then you can bind
> [backtab] and know it will work across platforms and terminals.
> Included in this proposal is removing obsolete bindings to
> [S-tab] and [S-iso-tab] for modes within Emacs and updating the
> documentation.  I include [S-iso-tab] since it is yet another
> abused synonym for [backtab].
>
> Note: this change will NOT ADD or REMOVE any bindings.  It is to
> unify the keypresses of [S-tab], [S-iso-tab] to [backtab] where
> they already exist.
>
> I don't pretend to know the history of the keybindings or
> specific terminals, but a uniform view of [S-tab] vs [backtab]
> makes sense to me.
>
> Thoughts?

For example, eshell-cmpl-initialize has:

  ;; jww (1999-10-19): Will this work on anything but X?
  (if (featurep 'xemacs)
      (define-key eshell-mode-map [iso-left-tab] 'pcomplete-reverse)
    (define-key eshell-mode-map [(shift iso-lefttab)] 'pcomplete-reverse)
    (define-key eshell-mode-map [(shift control ?i)] 'pcomplete-reverse))

And yet it still fails on some systems.

Leo

RE: Proposal: merge [S-tab] and [backtab]

[Drew Adams]

> My proposal is to universally translate [S-tab] and [S-iso-tab]
> ...
> I include [S-iso-tab] since it is yet another
> abused synonym for [backtab].

FWIW - 

I think I was the one who mentioned S-iso-tab (in the bug thread).
But I meant `S-iso-lefttab'.

Re: Proposal: merge [S-tab] and [backtab]

[Stefan Monnier]

> I think I was the one who mentioned S-iso-tab (in the bug thread).
> But I meant `S-iso-lefttab'.

BTW, could someone explain to me what is the iso-lefttab key?
What does it look like, where is it "typically" located?


        Stefan

RE: Proposal: merge [S-tab] and [backtab]

[Drew Adams]

> > I think I was the one who mentioned S-iso-tab (in the bug thread).
> > But I meant `S-iso-lefttab'.
> 
> BTW, could someone explain to me what is the iso-lefttab key?
> What does it look like, where is it "typically" located?

Not I.

(I just know about it because some Icicles users brought it to my attention.
Icicles uses [S-tab], and that wasn't available to users who had only
[S-iso-lefttab].)

Re: Proposal: merge [S-tab] and [backtab]

[Chad Brown]

On Aug 2, 2010, at 1:11 PM, Stefan Monnier wrote:

>> I think I was the one who mentioned S-iso-tab (in the bug thread).
>> But I meant `S-iso-lefttab'.
> 
> BTW, could someone explain to me what is the iso-lefttab key?
> What does it look like, where is it "typically" located?

I believe it stems from the practice on some (older?) keyboards of
printing a pair of arrows on the `tab key', one pointing each
direction, mimicking the placement of (for example) `!' over `1' on
the US-qwerty keyboard.

A little web searching suggests that it originated in the early days
of X:

	There is no universal standard for "backward tab" in the X
	Window System. On some systems shift+tab gives the keysym "ISO
	Left Tab", on others it gives a private "BackTab" keysym and
	on others it gives "Tab" and applications tell from the shift
	state that it means backward-tab rather than forward-tab. In
	the RFB protocol the latter approach is preferred. Viewers
	should generate a shifted Tab rather than ISO Left
	Tab. However, to be backwards-compatible with existing
	viewers, servers should also recognise ISO Left Tab as meaning
	a shifted Tab.

That same web searching also suggests that Emacs customization is the
most common source of `caring' about iso-lefttab in the present.

Hope this helps,
*Chad

Re: Proposal: merge [S-tab] and [backtab]

[Paul Griepentrog]

  On 8/2/10 11:09 AM, Drew Adams wrote:
> > My proposal is to universally translate [S-tab] and [S-iso-tab]
> > ...
> > I include [S-iso-tab] since it is yet another
> > abused synonym for [backtab].
>
>  FWIW -
>
>  I think I was the one who mentioned S-iso-tab (in the bug thread).

Yep, I got it from there.

I saw Stefan's commit on /srv/bzr/emacs/emacs-23 r99957, which looks like it
is a more comprehensive way to solve the problem than re-working a few 
modes.
I'll go from there.

>  But I meant `S-iso-lefttab'.