Poll: new keybinding for org-insert-structure-template?

Poll: new keybinding for org-insert-structure-template?

[Eric Abrahamsen]

Hello all,

Rasmus' org-tempo patch has landed, and it's time to make a few
adjustments to org-insert-structure-template: I think we're moving to
string keys in the template alist, rather than character keys, and also
some people noted that the keybinding "C-c C-x w" is too close to "C-c
C-x C-w", which kills a whole subtree. Not a good command to run
accidentally!

Given that the template system has been re-instated, are there any good
suggestions out there for a new keybinding for
`org-insert-structure-template'? This is the command that wraps existing
text in blocks.

Thanks,
Eric

Re: Poll: new keybinding for org-insert-structure-template?

[Rasmus]

Eric Abrahamsen <eric@ericabrahamsen.net> writes:

> Hello all,
>
> Rasmus' org-tempo patch has landed, and it's time to make a few
> adjustments to org-insert-structure-template: I think we're moving to
> string keys in the template alist, rather than character keys,

Let me know when you want me to do the change to strings.

> and also
> some people noted that the keybinding "C-c C-x w" is too close to "C-c
> C-x C-w", which kills a whole subtree. Not a good command to run
> accidentally!

Two other things that might be nice to changer are

    - Perhaps there could be a space after at least "#+begin_src" so that
      it’s ready to insert :keys.
    - Perhaps there could be a newline between "#+begin_x" and "#+end_x".

These things might already have been discussed in which case I missed
them.

> Given that the template system has been re-instated, are there any good
> suggestions out there for a new keybinding for
> `org-insert-structure-template'? This is the command that wraps existing
> text in blocks.

These aren’t too bad to typo on my keyboard: C-c C-. or C-c C-,

C-c C-x s might make sense.  It’s OKish to type on my keyboard layout, but
no harder than C-c C-x C-w.

free-keys.el can be used to identify combinations, although it isn’t
bullet-proof.

Rasmus

-- 
However beautiful the theory, one should occasionally look at the evidence

Re: Poll: new keybinding for org-insert-structure-template?

[Eric Abrahamsen]

Rasmus <rasmus@gmx.us> writes:

> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>
>> Hello all,
>>
>> Rasmus' org-tempo patch has landed, and it's time to make a few
>> adjustments to org-insert-structure-template: I think we're moving to
>> string keys in the template alist, rather than character keys,
>
> Let me know when you want me to do the change to strings.

No time like the present! I don't think there's anything else remaining
to be done.

>> and also
>> some people noted that the keybinding "C-c C-x w" is too close to "C-c
>> C-x C-w", which kills a whole subtree. Not a good command to run
>> accidentally!
>
> Two other things that might be nice to changer are
>
>     - Perhaps there could be a space after at least "#+begin_src" so that
>       it’s ready to insert :keys.
>     - Perhaps there could be a newline between "#+begin_x" and "#+end_x".
>
> These things might already have been discussed in which case I missed
> them.

The first was mentioned, not the second; I think they're both good
ideas. There's a clause in o-i-s-t that puts point at the begin line for
certain structure types, that would be a fine spot to insert a space.

>> Given that the template system has been re-instated, are there any good
>> suggestions out there for a new keybinding for
>> `org-insert-structure-template'? This is the command that wraps existing
>> text in blocks.
>
> These aren’t too bad to typo on my keyboard: C-c C-. or C-c C-,

I like either of these...

Re: Poll: new keybinding for org-insert-structure-template?

[Rasmus]

Hi,

>> Let me know when you want me to do the change to strings.
>
> No time like the present! I don't think there's anything else remaining
> to be done.

One stumbling block is how to input key strings to
org-insert-structure-template.  Currently, it employs read-key which is
really fast.  The other easy alternative is completing-read with the keys.
However, there you’d have to click RET at the end.  The ideal interface
for inputting the keys is something like the ox expert export UI.  I
looked at the code and it seems a bit complex.

Nicolas, are there any public interfaces to make ox export like
interfaces?

Rasmus

-- 
m-mm-mmm-mmmm bacon!

Re: Poll: new keybinding for org-insert-structure-template?

[Nicolas Goaziou]

Hello,

Rasmus <rasmus@gmx.us> writes:

> Nicolas, are there any public interfaces to make ox export like
> interfaces?

No, there isn't. However, Org provides `org-mks'.

Regards,

-- 
Nicolas Goaziou

Re: Poll: new keybinding for org-insert-structure-template?

[Rasmus]

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> Hello,
>
> Rasmus <rasmus@gmx.us> writes:
>
>> Nicolas, are there any public interfaces to make ox export like
>> interfaces?
>
> No, there isn't. However, Org provides `org-mks'.

Good point.  It isn’t quite as nice as the export dispatcher.  Let’s wait
and see what Eric prefers.

Rasmus

-- 
Send from my Emacs

Key binding popup interface (Was: Re: Poll: new keybinding for org-insert-structure-template on org mode list)

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 1454 bytes --]

Copying this on both Emacs devel and Org mode list. Hopefully this
discussion is eligible for that.

Problem statement: Need to have a pretty interface that shows the available
valid key bindings.

We have something custom developed for Org export interface. Magit has its
own popup interface. I know verilog-mode.el has something in-house for
picking the skeleton for insertion, but the interface is not as good as
hydra's.  Projectile.el (not on GNU Elpa) has its projectile commander,
another in-house attempt to get a similar pop-up interface.

Of all those options and more that I've seen out there, hydra.el looks like
the most elegant and *configurable* option.

Can hydra.el be merged into Emacs core, so that all the packages can start
taking advantage of that? We need to stop having each package develop their
own way to do this same thing.

It's already on GNU Elpa[1] , so there won't be any legal concerns.

[1]: https://elpa.gnu.org/packages/hydra.html

On Tue, Dec 12, 2017, 8:50 AM Rasmus <rasmus@gmx.us> wrote:

> Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:
>
> > Hello,
> >
> > Rasmus <rasmus@gmx.us> writes:
> >
> >> Nicolas, are there any public interfaces to make ox export like
> >> interfaces?
> >
> > No, there isn't. However, Org provides `org-mks'.
>
> Good point.  It isn’t quite as nice as the export dispatcher.  Let’s wait
> and see what Eric prefers.
>
-- 

Kaushal Modi

[-- Attachment #2: Type: text/html, Size: 2170 bytes --]

Re: Key binding popup interface

[Eric S Fraga]

[-- Attachment #1: Type: text/plain, Size: 373 bytes --]

On Tuesday, 12 Dec 2017 at 14:02, Kaushal Modi wrote:
> Copying this on both Emacs devel and Org mode list. Hopefully this
> discussion is eligible for that.
>
> Problem statement: Need to have a pretty interface that shows the available
> valid key bindings.

which-key-mode works well for me.

-- 
Eric S Fraga via Emacs 27.0.50, Org release_9.1.3-206-g1bb9cf

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 194 bytes --]

Re: Key binding popup interface

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 1391 bytes --]

On Tue, Dec 12, 2017 at 1:48 PM Eric S Fraga <esflists@gmail.com> wrote:

> On Tuesday, 12 Dec 2017 at 14:02, Kaushal Modi wrote:
> > Copying this on both Emacs devel and Org mode list. Hopefully this
> > discussion is eligible for that.
> >
> > Problem statement: Need to have a pretty interface that shows the
> available
> > valid key bindings.
>
> which-key-mode works well for me.
>

I use which-key too, but that serves a different purpose.

which-key shows *all* bindings that begin with a prefix.

hydra creates a quick keymap that can be bound to any key of user's choice.
Then the user needs to just set that main binding, and all the sub-bindings
in that keymap stay the same.

- Hydra basics[1]
- Projectile hydra example[2]
- Rectangle operations[3]

(See many more examples in that wiki.)

Very vaguely speaking, which-key is a read-only utility, hydra is a create
+ read utility. We need the "create" portion for the org stucture template
replacement discussed on the Org thread. I understand that while hydra
cannot be integrated in emacs 26 and so Org cannot immediately start using
it, it will be very much useful to many packages in future if first hydra
is baked into the emacs core.

[1]: https://github.com/abo-abo/hydra/wiki/Basics
[2]: https://github.com/abo-abo/hydra/wiki/Projectile
[3]: https://github.com/abo-abo/hydra/wiki/Rectangle-Operations
-- 

Kaushal Modi

[-- Attachment #2: Type: text/html, Size: 2201 bytes --]

Re: Key binding popup interface

[Alan E. Davis]

[-- Attachment #1: Type: text/plain, Size: 2217 bytes --]

I discovered accidentally some years ago, when I had been using a popup
help function that I cobbled together myself with a hot key, that
<Prefix>-F1 loads a list of bindings under that prefix.  It's an awesome
feature of GNU/Emacs.

Alan Davis

On Tue, Dec 12, 2017 at 10:56 AM, Kaushal Modi <kaushal.modi@gmail.com>
wrote:

> On Tue, Dec 12, 2017 at 1:48 PM Eric S Fraga <esflists@gmail.com> wrote:
>
>> On Tuesday, 12 Dec 2017 at 14:02, Kaushal Modi wrote:
>> > Copying this on both Emacs devel and Org mode list. Hopefully this
>> > discussion is eligible for that.
>> >
>> > Problem statement: Need to have a pretty interface that shows the
>> available
>> > valid key bindings.
>>
>> which-key-mode works well for me.
>>
>
> I use which-key too, but that serves a different purpose.
>
> which-key shows *all* bindings that begin with a prefix.
>
> hydra creates a quick keymap that can be bound to any key of user's
> choice. Then the user needs to just set that main binding, and all the
> sub-bindings in that keymap stay the same.
>
> - Hydra basics[1]
> - Projectile hydra example[2]
> - Rectangle operations[3]
>
> (See many more examples in that wiki.)
>
> Very vaguely speaking, which-key is a read-only utility, hydra is a create
> + read utility. We need the "create" portion for the org stucture template
> replacement discussed on the Org thread. I understand that while hydra
> cannot be integrated in emacs 26 and so Org cannot immediately start using
> it, it will be very much useful to many packages in future if first hydra
> is baked into the emacs core.
>
> [1]: https://github.com/abo-abo/hydra/wiki/Basics
> [2]: https://github.com/abo-abo/hydra/wiki/Projectile
> [3]: https://github.com/abo-abo/hydra/wiki/Rectangle-Operations
> --
>
> Kaushal Modi
>



-- 
[Fill in the blanks]

The use of corrupt manipulations and blatant rhetorical ploys ...---
outright lying, flagwaving, personal attacks, setting up phony
alternatives, misdirection, jargon-mongering, evading key issues, feigning
disinterested objectivity, willful misunderstanding of other points of
view---suggests that ... lacks both credibility and evidence.

             ---- Edward Tufte (in context of making presentations)

[-- Attachment #2: Type: text/html, Size: 3993 bytes --]

Re: Key binding popup interface

[Eric Abrahamsen]

Kaushal Modi <kaushal.modi@gmail.com> writes:

> Copying this on both Emacs devel and Org mode list. Hopefully this discussion is eligible for that. 
>
> Problem statement: Need to have a pretty interface that shows the available valid key bindings. 
>
> We have something custom developed for Org export interface. Magit has its own popup interface. I know verilog-mode.el has something in-house for picking the skeleton for insertion, but the interface is not as good as hydra's. 
> Projectile.el (not on GNU Elpa) has its projectile commander, another in-house attempt to get a similar pop-up interface. 
>
> Of all those options and more that I've seen out there, hydra.el looks like the most elegant and *configurable* option. 
>
> Can hydra.el be merged into Emacs core, so that all the packages can start taking advantage of that? We need to stop having each package develop their own way to do this same thing. 

Hydra definitely sounds nice, but unless I'm misunderstanding how it
works, it can't be a replacement for either Org's export interface, or
Magit's popups. Both of those can set various state variables before
choosing an actual action -- Hydra can't do that, can it?

Did I hear that Magit was breaking its popup interface out into a
separate library? If so, that would also be a nice thing.

Re: Key binding popup interface

[Kaushal Modi]

[-- Attachment #1.1: Type: text/plain, Size: 4310 bytes --]

On Tue, Dec 12, 2017 at 3:20 PM Eric Abrahamsen <eric@ericabrahamsen.net>
wrote:

>
> Hydra definitely sounds nice, but unless I'm misunderstanding how it
> works, it can't be a replacement for either Org's export interface, or
> Magit's popups. Both of those can set various state variables before
> choosing an actual action -- Hydra can't do that, can it?
>

hydra can do anything you can do in elisp before/after the action.

For example, here's a hydra I use for rectangle actions:

(defhydra hydra-rectangle (:body-pre (rectangle-mark-mode 1)
                           :color pink
                           :post (deactivate-mark)
                           :hint nil)
  "
  Rectangle:
  ^_p_^        _d_   delete      _s_tring        _c_/_C_ (delete/kill) and
replace with space
_b_   _f_      _k_   cut         _r_eset         _o_pen (create blank
rectangle and push text in region to the right)
  ^_n_^        _w_   copy        e_x_change      _X_ delete whitespace
starting from the left edge of the rectangle
^^^^           _y_   paste       _e_xtend        Prefix rectangle lines
with _N_umbers
"
  ("b"   backward-char)
  ("f"   forward-char)
  ("p"   previous-line)
  ("n"   next-line)

  ("d"   delete-rectangle :color blue)
  ("k"   kill-rectangle :color blue)
  ("w"   copy-rectangle-as-kill :color blue)
  ("y"   yank-rectangle :color blue)
  ("s"   string-rectangle :color blue)
  ("t"   string-rectangle :color blue)
  ("r"   (if (region-active-p)
             (deactivate-mark)
           (rectangle-mark-mode 1)))
  ("x"   ora-ex-point-mark)
  ("e"   modi/extend-rectangle-to-end)
  ("c"   clear-rectangle)
  ("C"   modi/kill-rectangle-replace-with-space :color blue)
  ("o"   open-rectangle :color blue)
  ("X"   delete-whitespace-rectangle :color blue)
  ("N"   rectangle-number-lines :color blue)
  ("q"   nil "cancel" :color blue))

Note the use of :body-pre, :post in above example. More details here:
https://github.com/abo-abo/hydra/wiki/internals#body-pre

<https://github.com/abo-abo/hydra/wiki/internals#body-pre>
But this email wasn't to propose replace the matured popups like magit and
org export with hydra. The proposal is to integrate hydra into emacs, so
that many packages that self-implement these popups can do the same using
hydra more easily. Dired, ibuffer can make use of hydra and make all the
bindings more discoverable and memorable. In future, may be hydra can be
use for the org template insertion piece.

I can talk of verilog-mode as I use it everyday. In verilog-mode.el, there
is this code:

(defvar verilog-template-map
  (let ((map (make-sparse-keymap)))
    (define-key map "a" 'verilog-sk-always)
    (define-key map "b" 'verilog-sk-begin)
  ..
  (define-key map "D" 'verilog-sk-define-signal)
    map)
  "Keymap used in Verilog mode for smart template operations.")

In my config, I replace the same with hydra and it looks like this:

    (defhydra hydra-verilog-template (:color blue
                                      :hint nil)
      "
_i_nitial        _?_ if             _j_ fork
_A_ssign                _uc_ uvm-component
_b_egin          _:_ else-if        _m_odule
_I_nput                 _uo_ uvm-object
_a_lways         _f_or              _g_enerate         _O_utput
^^               _w_hile            _p_rimitive        _=_ inout
^^               _r_epeat           _s_pecify
_S_tate-machine         _h_eader
^^               _c_ase             _t_ask
_W_ire                  _/_ comment
^^               case_x_            _F_unction         _R_eg
^^               case_z_            ^^                 _D_efine-signal
"
      ("a"   verilog-sk-always)
      ("b"   verilog-sk-begin)
  ...
      ("D"   verilog-sk-define-signal)
      ("q"   nil nil :color blue)
      ("C-g" nil nil :color blue))

Which gives in these interface which makes the bindings much more memorable
and accessible:

[image: image.png]
There are so many places like this that can get a facelift from hydra.

Did I hear that Magit was breaking its popup interface out into a
> separate library? If so, that would also be a nice thing.
>

That would be great too. We are basically missing a configurable library to
bind temporary key maps, that can make the bindings present in a
discoverable way to the user. So far hydra.el fits the bill.
-- 

Kaushal Modi

[-- Attachment #1.2: Type: text/html, Size: 6539 bytes --]

[-- Attachment #2: image.png --]
[-- Type: image/png, Size: 20877 bytes --]

Re: Key binding popup interface

[Eric Abrahamsen]

Kaushal Modi <kaushal.modi@gmail.com> writes:

> On Tue, Dec 12, 2017 at 3:20 PM Eric Abrahamsen <eric@ericabrahamsen.net> wrote:
>
>  Hydra definitely sounds nice, but unless I'm misunderstanding how it
>  works, it can't be a replacement for either Org's export interface, or
>  Magit's popups. Both of those can set various state variables before
>  choosing an actual action -- Hydra can't do that, can it?

[...]

> But this email wasn't to propose replace the matured popups like magit and org export with hydra. The proposal is to integrate hydra into emacs, so that many packages that self-implement these popups can do the same using hydra
> more easily. Dired, ibuffer can make use of hydra and make all the bindings more discoverable and memorable. In future, may be hydra can be use for the org template insertion piece.

Okay, interesting. I agree that "discoverable keymaps" and "action
dispatchers" (whatever Org and Magit do) aren't quite the same thing,
and it would be nice to have both. I guess in the meantime, though,
we're still going to have to use something else for the Org structure
thing.

Re: Key binding popup interface

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 589 bytes --]

On Tue, Dec 12, 2017 at 3:47 PM Eric Abrahamsen <eric@ericabrahamsen.net>
wrote:

>
> Okay, interesting. I agree that "discoverable keymaps" and "action
> dispatchers" (whatever Org and Magit do) aren't quite the same thing,
> and it would be nice to have both. I guess in the meantime, though,
> we're still going to have to use something else for the Org structure
> thing.
>

Correct. That's why I spawned off this "something for future" thread, and
did not talk about this in the original thread[1] :)

[1]: http://lists.gnu.org/r/emacs-orgmode/2017-12/msg00221.html
-- 

Kaushal Modi

[-- Attachment #2: Type: text/html, Size: 1127 bytes --]

Re: Key binding popup interface

[Stefan Monnier]

> Problem statement: Need to have a pretty interface that shows the available
> valid key bindings.
[...]
> Of all those options and more that I've seen out there, hydra.el looks like
> the most elegant and *configurable* option.

While Hydra might include the feature you're looking for, AFAICT it does
something else than what you're looking for.

> Can hydra.el be merged into Emacs core, so that all the packages can start
> taking advantage of that?

I have no opinion on that part, but if a part of Hydra would be useful
for other packages, then it'd be good to split it out: for me "Hydra"
means "define a Hydra via defhydra or something like that", so if you
tell me to use Hydra to "show the available valid key bindings",
I wouldn't know where to start.


        Stefan

Re: Key binding popup interface

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 1275 bytes --]

On Tue, Dec 12, 2017 at 4:05 PM Stefan Monnier <monnier@iro.umontreal.ca>
wrote:

> > Problem statement: Need to have a pretty interface that shows the
> available
> > valid key bindings.
> [...]
> > Of all those options and more that I've seen out there, hydra.el looks
> like
> > the most elegant and *configurable* option.
>
> While Hydra might include the feature you're looking for, AFAICT it does
> something else than what you're looking for.
>

I don't follow "it does something else than what you're looking for". hydra
is used to define interfaces to temporary keymaps with well-formatted hints.

I have no opinion on that part, but if a part of Hydra would be useful
> for other packages, then it'd be good to split it out: for me "Hydra"
> means "define a Hydra via defhydra or something like that", so if you
> tell me to use Hydra to "show the available valid key bindings",
> I wouldn't know where to start.
>

I meant that use hydra to define such interfaces.. the package authors
would need to use hydra to define these interfaces. All the user would need
to know is to hit some X binding to get that popup interface.

I quickly went though hydra.el.. isn't defhydra mainly what it is? What
would you suggest splitting out of that library?
-- 

Kaushal Modi

[-- Attachment #2: Type: text/html, Size: 1953 bytes --]

Re: Key binding popup interface

[Stefan Monnier]

>> While Hydra might include the feature you're looking for, AFAICT it does
>> something else than what you're looking for.
> I don't follow "it does something else than what you're looking for". hydra
> is used to define interfaces to temporary keymaps with well-formatted hints.

I tend to think of Hydra as "bindings that stick around" (to take the
wording on the first line of hydra.el), rather than "ways to show
available bindings of the current submap".  So, yes, I think it does
something else (something more) than what I understand you want.

And, BTW, if I take a hydra like

    (defhydra hydra-zoom (global-map "<f6>")
      "zoom"
       ("g" text-scale-increase "in")
       ("l" text-scale-decrease "out"))

and I press `f6` I don't get any help in the echo area (nor in the "lv"
area).  I only get that help after pressing `f6 g` or `f6 l`, so I need
some other mechanism to find those "initial" key bindings.

So in this respect, I think it does something less than what
I understand you'd want.

> I quickly went though hydra.el.. isn't defhydra mainly what it is? What
> would you suggest splitting out of that library?

I don't know enough about it to have a clear opinion on that.


        Stefan

Re: Key binding popup interface

[Robert Weiner]

[-- Attachment #1: Type: text/plain, Size: 823 bytes --]

On Tue, Dec 12, 2017 at 4:01 PM, Stefan Monnier <monnier@iro.umontreal.ca>
wrote:

>
> > Can hydra.el be merged into Emacs core, so that all the packages can
> start
> > taking advantage of that?
>
> I have no opinion on that part, but if a part of Hydra would be useful
> for other packages, then it'd be good to split it out: for me "Hydra"
> means "define a Hydra via defhydra or something like that", so if you
> tell me to use Hydra to "show the available valid key bindings",
> I wouldn't know where to start.
>

​One limitation of hydra right now is that it doesn't interface with the
standard way of showing help for key bindings since its keys aren't
actually bound but handled via internal hydra event handling.  With a bit
of thought though, I think it could be integrated well.

Bob
​

[-- Attachment #2: Type: text/html, Size: 1752 bytes --]

Re: Poll: new keybinding for org-insert-structure-template?

[Eric Abrahamsen]

Rasmus <rasmus@gmx.us> writes:

> Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:
>
>> Hello,
>>
>> Rasmus <rasmus@gmx.us> writes:
>>
>>> Nicolas, are there any public interfaces to make ox export like
>>> interfaces?
>>
>> No, there isn't. However, Org provides `org-mks'.
>
> Good point.  It isn’t quite as nice as the export dispatcher.  Let’s wait
> and see what Eric prefers.

I don't have to own this! I don't have a strong feeling about it, and
left to my own devices would just keep the single-character keys. But I
see why people would want more/longer keys, and better discoverability.

The main desires are:

1. String keys of arbitrary length.
2. Discoverability: either a window of choices pops up automatically, or
   you can do it with TAB, etc.
3. Ideally, you don't have to hit an extra key to finish.

Given that, it seems to me that `org-mks' is just fine. It doesn't do as
much as the export dispatcher, but I don't see that it needs to.

The only downside is that it might make the choices a bit harder to
customize, as the table argument is more fiddly. But maybe not a big
deal.

Eric

Re: Key binding popup interface

[John Wiegley]

>>>>> "SM" == Stefan Monnier <monnier@iro.umontreal.ca> writes:

SM> I have no opinion on that part, but if a part of Hydra would be useful for
SM> other packages, then it'd be good to split it out: for me "Hydra" means
SM> "define a Hydra via defhydra or something like that", so if you tell me to
SM> use Hydra to "show the available valid key bindings", I wouldn't know
SM> where to start.

Plus, there are other, less intrusive way to achieve "show the available valid
key bindings", such as the which-key package.

I think one area where Hydras come into their own is that once you have the
Hydra active, it stays active, allowing you to chain together many single-
letter commands.

For example, you could have a hydra for "large-scale syntactic structures",
where repeated uses of n/p are mapped to what C-c C-n and C-c C-p usually do
in modes like cc-mode.

-- 
John Wiegley                  GPG fingerprint = 4710 CF98 AF9B 327B B80F
http://newartisans.com                          60E1 46C4 BD1A 7AC1 4BA2

Re: Key binding popup interface

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 4223 bytes --]

On Tue, Dec 12, 2017 at 4:49 PM Stefan Monnier <monnier@iro.umontreal.ca>
wrote:

> I tend to think of Hydra as "bindings that stick around" (to take the
> wording on the first line of hydra.el), rather than "ways to show
> available bindings of the current submap".  So, yes, I think it does
> something else (something more)


Well, that's correct ..

than what I understand you want.
>

Though I now think I did a very bad job at constructing that problem
statement. But I know for sure that hydra.el fits the bill perfectly (I've
been using it ever since Oleh released it).

Integrating hydra into emacs with help package developers do these:
- Create interfaces for transitional keymaps
- The bindings could be sticky or not (can be configured in hydra)
- Allow customizing the descriptions of the bindings (@JWiegley as you
later suggest which-key, which-key does not allow that.. well it does.. but
the user will need to tweak the description, etc.. I see which-end more as
a tool only for the end-user. hydra can be used both by package developers
and end users).
- Pick and choose which bindings to "show" in that interface.. you don't
have to show all. The package developer may choose to have duplicate
bindings for some function in that keymap, but prefer to show only the
preferred binding in the hydra popup (cannot do that in which-key.. it
shows *everything*).


> And, BTW, if I take a hydra like
>
>     (defhydra hydra-zoom (global-map "<f6>")
>       "zoom"
>        ("g" text-scale-increase "in")
>        ("l" text-scale-decrease "out"))
>
> and I press `f6` I don't get any help in the echo area (nor in the "lv"
> area).  I only get that help after pressing `f6 g` or `f6 l`, so I need
> some other mechanism to find those "initial" key bindings.
>

That's because you used Style 1 (as explained in this hydra Wiki:
https://github.com/abo-abo/hydra/wiki/Binding-Styles). Use the Style 2 to
take care of the issue you stated:

(defhydra hydra-zoom ()
  "zoom"
  ("g" text-scale-increase "in")
  ("l" text-scale-decrease "out"))

(global-set-key (kbd "C-c") 'hydra-zoom/body)

In Style 1, you allow the hydra to share the "keymap space" with other
bindings not related to that hydra.

In Style 2, the hydra takes over the whole "keymap space". In above Style 2
example, the "C-c" space is completely ruled by the hydra-zoom hydra.


> So in this respect, I think it does something less than what
> I understand you'd want.
>

No. It does everything that I need to do. But of course it has a lot of
features, which might be suitable for different applications. See keys like
:pre, :post and more described in
https://github.com/abo-abo/hydra/wiki/internals.

I can see something like this begin added to smerge.el if and when hydra.el
gets merged to the core (See the use of :pre and :post):

(defhydra hydra-smerge (:color pink
                        :hint nil
                        :pre (smerge-mode 1)
                        ;; Disable `smerge-mode' when quitting hydra if
                        ;; no merge conflicts remain.
                        :post (smerge-auto-leave))
  "
^Move^       ^Keep^               ^Diff^                 ^Other^
^^-----------^^-------------------^^---------------------^^-------
_n_ext       _b_ase               _<_: upper/base        _C_ombine
_p_rev       _u_pper              _=_: upper/lower       _r_esolve
^^           _l_ower              _>_: base/lower        _k_ill current
^^           _a_ll                _R_efine
^^           _RET_: current       _E_diff
"
  ("n" smerge-next)
  ("p" smerge-prev)
  ("b" smerge-keep-base)
  ("u" smerge-keep-upper)
  ("l" smerge-keep-lower)
  ("a" smerge-keep-all)
  ("RET" smerge-keep-current)
  ("\C-m" smerge-keep-current)
  ("<" smerge-diff-base-upper)
  ("=" smerge-diff-upper-lower)
  (">" smerge-diff-base-lower)
  ("R" smerge-refine)
  ("E" smerge-ediff)
  ("C" smerge-combine-with-next)
  ("r" smerge-resolve)
  ("k" smerge-kill-current)
  ("q" nil "cancel" :color blue))


> > I quickly went though hydra.el.. isn't defhydra mainly what it is? What
> > would you suggest splitting out of that library?
>
> I don't know enough about it to have a clear opinion on that.
>

OK.

Thanks.
-- 

Kaushal Modi

[-- Attachment #2: Type: text/html, Size: 6504 bytes --]

Re: Key binding popup interface

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 582 bytes --]

On Tue, Dec 12, 2017 at 6:51 PM Robert Weiner <rsw@gnu.org> wrote:

>
> ​One limitation of hydra right now is that it doesn't interface with the
> standard way of showing help for key bindings since its keys aren't
> actually bound but handled via internal hydra event handling.  With a bit
> of thought though, I think it could be integrated well.
>

I don't follow what the functional limitations are of what you said. How
does your concern affect a package author or the end user?

Adding @Oleh back to the thread.. hopefully he can weigh in.
-- 

Kaushal Modi

[-- Attachment #2: Type: text/html, Size: 1122 bytes --]

Re: Key binding popup interface

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 818 bytes --]

On Wed, Dec 13, 2017 at 1:53 AM John Wiegley <johnw@gnu.org> wrote:

> Plus, there are other, less intrusive way to achieve "show the available
> valid
> key bindings", such as the which-key package.
>

As I mentioned in a recent response to Stefan in this thread, I did a poor
job with that problem statement.. I don't want to *just* show the available
bindings.

I think one area where Hydras come into their own is that once you have the
> Hydra active, it stays active, allowing you to chain together many single-
> letter commands.
>

That's correct. It's up to the package author as to how much of hydra
features they want to pack into their package's hydras. hydra is not *just*
for showing available keys and/or *just* for chaining many single letter
commands. It's much more, in a good way.
-- 

Kaushal Modi

[-- Attachment #2: Type: text/html, Size: 1383 bytes --]

Re: Key binding popup interface

[Stefan Monnier]

> (defhydra hydra-zoom ()
>   "zoom"
>   ("g" text-scale-increase "in")
>   ("l" text-scale-decrease "out"))
> (global-set-key (kbd "C-c") 'hydra-zoom/body)

But now this one suffers from other problems:
- `C-h k` does not let me directly find the binding of `C-c g`.
- A `C-c a` binding in a lower keymap is now made unavailable.
- A `C-c a` binding in a higher keymap now makes this hydra unavailable.

> In Style 1, you allow the hydra to share the "keymap space" with other
> bindings not related to that hydra.
> In Style 2, the hydra takes over the whole "keymap space". In above Style 2
> example, the "C-c" space is completely ruled by the hydra-zoom hydra.

I want it all.  Don't get me wrong: I fully understand that hydra.el
works within the constraints of what Emacs provides and so it has to
choose between style 1 or style 2, but if we're talking about
integrating this into Emacs, then we should take the opportunity to lift
those restrictions.


        Stefan

Re: Key binding popup interface

[Robert Weiner]

[-- Attachment #1: Type: text/plain, Size: 1204 bytes --]

On Wed, Dec 13, 2017 at 10:40 AM, Kaushal Modi <kaushal.modi@gmail.com>
wrote:

> On Tue, Dec 12, 2017 at 6:51 PM Robert Weiner <rsw@gnu.org> wrote:
>
>>
>> ​One limitation of hydra right now is that it doesn't interface with the
>> standard way of showing help for key bindings since its keys aren't
>> actually bound but handled via internal hydra event handling.  With a bit
>> of thought though, I think it could be integrated well.
>>
>
> I don't follow what the functional limitations are of what you said. How
> does your concern affect a package author or the end user?
>

​C-h k shows the documentation for a key binding in a help buffer rather
than just a laid out summary of all the keys in a particular hydra map.
C-h k is important to know the details of what individual keys do, so
hydra needs a way of integrating with standard key documentation lookup
capabilities if it gets integrated with core Emacs, which I would like to
see.

Personally, I also think the red, blue, etc. color notation should change to
something more descriptive of what they do and then the particular colors
associated could be customized to suit any particular visual theme.

Bob

[-- Attachment #2: Type: text/html, Size: 2985 bytes --]

Re: Poll: new keybinding for org-insert-structure-template?

[Rasmus]

Eric Abrahamsen <eric@ericabrahamsen.net> writes:

> Rasmus <rasmus@gmx.us> writes:
>
>> Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:
>>
>>> Hello,
>>>
>>> Rasmus <rasmus@gmx.us> writes:
>>>
>>>> Nicolas, are there any public interfaces to make ox export like
>>>> interfaces?
>>>
>>> No, there isn't. However, Org provides `org-mks'.
>>
>> Good point.  It isn’t quite as nice as the export dispatcher.  Let’s wait
>> and see what Eric prefers.
>
> I don't have to own this! I don't have a strong feeling about it, and
> left to my own devices would just keep the single-character keys.

It’s pretty nice and quick with only one character.

> But I see why people would want more/longer keys, and better
> discoverability.

Discoverability is nice to have, but it’s a secondary issue.


> The main desires are:
>
> 1. String keys of arbitrary length.
> 2. Discoverability: either a window of choices pops up automatically, or
>    you can do it with TAB, etc.
> 3. Ideally, you don't have to hit an extra key to finish.
>
> Given that, it seems to me that `org-mks' is just fine. It doesn't do as
> much as the export dispatcher, but I don't see that it needs to.

It doesn’t seem to offer a way to distinguish between a single "a" and
"ab".

(org-mks '(("a" "a.. templates")
           ("a" "export ascii" "")
           ("ab" "abstract" "")
           ("l" "latex" ""))
         "Select a template"
         "Template key: "
         '(("q" "Abort")))

Perhaps the best way would be to automatically reserve upper case letters
for blocks with keys longer than one character, although that seems a bit
arbitrary.

(org-mks '(("A" "a.. templates")
           ("a" "export ascii" "")
           ("Ab" "abstract" "")
           ("l" "latex" ""))
         "Select a template"
         "Template key: "
         '(("q" "Abort")))

Rasmus

-- 
And I faced endless streams of vendor-approved Ikea furniture. . .

Re: Poll: new keybinding for org-insert-structure-template?

[Eric Abrahamsen]

Rasmus <rasmus@gmx.us> writes:

> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>
>> Rasmus <rasmus@gmx.us> writes:
>>
>>> Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:
>>>
>>>> Hello,
>>>>
>>>> Rasmus <rasmus@gmx.us> writes:
>>>>
>>>>> Nicolas, are there any public interfaces to make ox export like
>>>>> interfaces?
>>>>
>>>> No, there isn't. However, Org provides `org-mks'.
>>>
>>> Good point.  It isn’t quite as nice as the export dispatcher.  Let’s wait
>>> and see what Eric prefers.
>>
>> I don't have to own this! I don't have a strong feeling about it, and
>> left to my own devices would just keep the single-character keys.
>
> It’s pretty nice and quick with only one character.
>
>> But I see why people would want more/longer keys, and better
>> discoverability.
>
> Discoverability is nice to have, but it’s a secondary issue.
>
>
>> The main desires are:
>>
>> 1. String keys of arbitrary length.
>> 2. Discoverability: either a window of choices pops up automatically, or
>>    you can do it with TAB, etc.
>> 3. Ideally, you don't have to hit an extra key to finish.
>>
>> Given that, it seems to me that `org-mks' is just fine. It doesn't do as
>> much as the export dispatcher, but I don't see that it needs to.
>
> It doesn’t seem to offer a way to distinguish between a single "a" and
> "ab".
>
> (org-mks '(("a" "a.. templates")
>            ("a" "export ascii" "")
>            ("ab" "abstract" "")
>            ("l" "latex" ""))
>          "Select a template"
>          "Template key: "
>          '(("q" "Abort")))
>
> Perhaps the best way would be to automatically reserve upper case letters
> for blocks with keys longer than one character, although that seems a bit
> arbitrary.
>
> (org-mks '(("A" "a.. templates")
>            ("a" "export ascii" "")
>            ("Ab" "abstract" "")
>            ("l" "latex" ""))
>          "Select a template"
>          "Template key: "
>          '(("q" "Abort")))

Right, you can't have a string act as both a "sub-menu dispatcher" and a
menu item at the same time.

I don't see that it's a big deal, though. For the default value, at
least, it's pretty clear how we would arrange the values: SRC and EXPORT
need sub-menus, and everything else doesn't. Ie:

'(("s" "Source Code")
  ("se" "Elisp" "src elisp")
  ("sp" "Python" "src python")
  ...etc
  ("e" "Export Block")
  ("eh" "HTML" "export html")
  ("el" "LaTeX" "export latex")
  ...etc
  ("v" "Verbatim" "verbatim")
  ("q" "Quote" "quote")
  ("E" "Example" "example")
  ...etc
  )

That's not so bad, is it?

Eric

Re: Poll: new keybinding for org-insert-structure-template?

[Rasmus]

Eric Abrahamsen <eric@ericabrahamsen.net> writes:

> That's not so bad, is it?

The only way it’s "bad" is in the sense it limits the flexibility of
snippets, like "<s".  As you point out, if I add "<sr" as an "src R" block
I can no longer have "<s" as plain "src" block.  Unless, the meaning of
key in org-structure-template-alist is handled differently in the case of
snippet expansion and insertion via org-insert-structure-template.

Rasmus

-- 
In theory, practice and theory are the same. In practice they are not

Re: Poll: new keybinding for org-insert-structure-template?

[Kaushal Modi]

[-- Attachment #1: Type: text/plain, Size: 665 bytes --]

On Fri, Dec 15, 2017 at 6:23 AM Rasmus <rasmus@gmx.us> wrote:

> The only way it’s "bad" is in the sense it limits the flexibility of
> snippets, like "<s".  As you point out, if I add "<sr" as an "src R" block
> I can no longer have "<s" as plain "src" block.


May be have "<ss" insert plain source block, "<ee", plain export block?

I missed this in the original conversation, but in the cases where plain
source/export blocks are inserted, does the point move to where the
language/backend needs to be typed?

i.e. you insert a plain source block and point ends up where "|" is shown
below:

#+BEGIN_SRC |

#+END_SRC

-- 

Kaushal Modi

[-- Attachment #2: Type: text/html, Size: 1230 bytes --]

Re: Poll: new keybinding for org-insert-structure-template?

[Rasmus]

Kaushal Modi <kaushal.modi@gmail.com> writes:

> On Fri, Dec 15, 2017 at 6:23 AM Rasmus <rasmus@gmx.us> wrote:
>
>> The only way it’s "bad" is in the sense it limits the flexibility of
>> snippets, like "<s".  As you point out, if I add "<sr" as an "src R" block
>> I can no longer have "<s" as plain "src" block.
>
>
> May be have "<ss" insert plain source block, "<ee", plain export block?

I wouldn’t really like that.

> I missed this in the original conversation, but in the cases where plain
> source/export blocks are inserted, does the point move to where the
> language/backend needs to be typed?

When using Org Tempo an extra space is always inserted.  It probably
should only do it with blocks that support arguments, actually.  ATM I
don’t think spaces are inserted when using
‘org-insert-structure-template’, but I they’ll probably be consistent
eventually.

Rasmus

-- 
To err is human. To screw up 10⁶ times per second, you need a computer

Re: Poll: new keybinding for org-insert-structure-template?

[Eric Abrahamsen]

Rasmus <rasmus@gmx.us> writes:

> Kaushal Modi <kaushal.modi@gmail.com> writes:
>
>> On Fri, Dec 15, 2017 at 6:23 AM Rasmus <rasmus@gmx.us> wrote:
>>
>>> The only way it’s "bad" is in the sense it limits the flexibility of
>>> snippets, like "<s".  As you point out, if I add "<sr" as an "src R" block
>>> I can no longer have "<s" as plain "src" block.

I don't see any way around that. Any system that allows string keys of
arbitrary length is going to run into that problem.

>> May be have "<ss" insert plain source block, "<ee", plain export block?
>
> I wouldn’t really like that.
>
>> I missed this in the original conversation, but in the cases where plain
>> source/export blocks are inserted, does the point move to where the
>> language/backend needs to be typed?
>
> When using Org Tempo an extra space is always inserted.  It probably
> should only do it with blocks that support arguments, actually.  ATM I
> don’t think spaces are inserted when using
> ‘org-insert-structure-template’, but I they’ll probably be consistent
> eventually.

I think the idea was that for src and export blocks, where you're likely
to be adding additional parameters, point is moved to the end of the
#+begin line. Actually I think that's what happens now.

Re: Poll: new keybinding for org-insert-structure-template?

[Eric Abrahamsen]

Eric Abrahamsen <eric@ericabrahamsen.net> writes:

> Rasmus <rasmus@gmx.us> writes:
>
>> Kaushal Modi <kaushal.modi@gmail.com> writes:
>>
>>> On Fri, Dec 15, 2017 at 6:23 AM Rasmus <rasmus@gmx.us> wrote:
>>>
>>>> The only way it’s "bad" is in the sense it limits the flexibility of
>>>> snippets, like "<s".  As you point out, if I add "<sr" as an "src R" block
>>>> I can no longer have "<s" as plain "src" block.
>
> I don't see any way around that. Any system that allows string keys of
> arbitrary length is going to run into that problem.

One possible fix, a bit arbitrary: in the default value, provide <TAB>
as an artificial "stop key" in the sub-menus. So "s" starts the "source
code" sub-menu, and a <TAB> after that simply inserts "#+begin_src", and
leaves point after that.

Re: Poll: new keybinding for org-insert-structure-template?

[Rasmus]

Eric Abrahamsen <eric@ericabrahamsen.net> writes:

> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>
>> Rasmus <rasmus@gmx.us> writes:
>>
>>> Kaushal Modi <kaushal.modi@gmail.com> writes:
>>>
>>>> On Fri, Dec 15, 2017 at 6:23 AM Rasmus <rasmus@gmx.us> wrote:
>>>>
>>>>> The only way it’s "bad" is in the sense it limits the flexibility of
>>>>> snippets, like "<s".  As you point out, if I add "<sr" as an "src R" block
>>>>> I can no longer have "<s" as plain "src" block.
>>
>> I don't see any way around that. Any system that allows string keys of
>> arbitrary length is going to run into that problem.
>
> One possible fix, a bit arbitrary: in the default value, provide <TAB>
> as an artificial "stop key" in the sub-menus. So "s" starts the "source
> code" sub-menu, and a <TAB> after that simply inserts "#+begin_src", and
> leaves point after that.

Yeah, I tried to suggest that earlier (unless I didn’t say it), but I
might not have expressed the idea in an understandable manner :)

I think that would be the best approach, but there’s no infrastructure
that I know of that does this ATM (but I haven’t had a lot of time lately,
so my knowledge on this issue is limited!).

Rasmus

-- 
9000!

Re: Poll: new keybinding for org-insert-structure-template?

[Eric Abrahamsen]

Rasmus <rasmus@gmx.us> writes:

> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>
>> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>>
>>> Rasmus <rasmus@gmx.us> writes:
>>>
>>>> Kaushal Modi <kaushal.modi@gmail.com> writes:
>>>>
>>>>> On Fri, Dec 15, 2017 at 6:23 AM Rasmus <rasmus@gmx.us> wrote:
>>>>>
>>>>>> The only way it’s "bad" is in the sense it limits the flexibility of
>>>>>> snippets, like "<s".  As you point out, if I add "<sr" as an "src R" block
>>>>>> I can no longer have "<s" as plain "src" block.
>>>
>>> I don't see any way around that. Any system that allows string keys of
>>> arbitrary length is going to run into that problem.
>>
>> One possible fix, a bit arbitrary: in the default value, provide <TAB>
>> as an artificial "stop key" in the sub-menus. So "s" starts the "source
>> code" sub-menu, and a <TAB> after that simply inserts "#+begin_src", and
>> leaves point after that.
>
> Yeah, I tried to suggest that earlier (unless I didn’t say it), but I
> might not have expressed the idea in an understandable manner :)

Maybe I missed it!

> I think that would be the best approach, but there’s no infrastructure
> that I know of that does this ATM (but I haven’t had a lot of time lately,
> so my knowledge on this issue is limited!).

Can't we do this with tempo? It will have to be "handmade", not
automatic, but:

'(("s" "Source Code")
  ("se" "Elisp" "src elisp")
  ("sp" "Python" "src python")
  ("TAB" "Empty" "src ")
  ...etc
  ("e" "Export Block")
  ("eh" "HTML" "export html")
  ("el" "LaTeX" "export latex")
  ("TAB" "Empty" "export ")
  ...etc
  ("v" "Verbatim" "verbatim")
  ("q" "Quote" "quote")
  ("E" "Example" "example")
  ...etc
  )

Re: Poll: new keybinding for org-insert-structure-template?

[Eric Abrahamsen]

Eric Abrahamsen <eric@ericabrahamsen.net> writes:

> Rasmus <rasmus@gmx.us> writes:
>
>> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>>
>>> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>>>
>>>> Rasmus <rasmus@gmx.us> writes:
>>>>
>>>>> Kaushal Modi <kaushal.modi@gmail.com> writes:
>>>>>
>>>>>> On Fri, Dec 15, 2017 at 6:23 AM Rasmus <rasmus@gmx.us> wrote:
>>>>>>
>>>>>>> The only way it’s "bad" is in the sense it limits the flexibility of
>>>>>>> snippets, like "<s".  As you point out, if I add "<sr" as an "src R" block
>>>>>>> I can no longer have "<s" as plain "src" block.
>>>>
>>>> I don't see any way around that. Any system that allows string keys of
>>>> arbitrary length is going to run into that problem.
>>>
>>> One possible fix, a bit arbitrary: in the default value, provide <TAB>
>>> as an artificial "stop key" in the sub-menus. So "s" starts the "source
>>> code" sub-menu, and a <TAB> after that simply inserts "#+begin_src", and
>>> leaves point after that.
>>
>> Yeah, I tried to suggest that earlier (unless I didn’t say it), but I
>> might not have expressed the idea in an understandable manner :)
>
> Maybe I missed it!
>
>> I think that would be the best approach, but there’s no infrastructure
>> that I know of that does this ATM (but I haven’t had a lot of time lately,
>> so my knowledge on this issue is limited!).
>
> Can't we do this with tempo? It will have to be "handmade", not
> automatic, but:
>
> '(("s" "Source Code")
>   ("se" "Elisp" "src elisp")
>   ("sp" "Python" "src python")
>   ("TAB" "Empty" "src ")
>   ...etc
>   ("e" "Export Block")
>   ("eh" "HTML" "export html")
>   ("el" "LaTeX" "export latex")
>   ("TAB" "Empty" "export ")
>   ...etc
>   ("v" "Verbatim" "verbatim")
>   ("q" "Quote" "quote")
>   ("E" "Example" "example")
>   ...etc
>   )

Ahem, should have actually tried that first:

(org-mks
 '(("s" "Source Code")
   ("se" "Elisp" "src elisp")
   ("sp" "Python" "src python")
   ("s\t" "Empty" "src ")
   ("e" "Export Block")
   ("eh" "HTML" "export html")
   ("el" "LaTeX" "export latex")
   ("s\t" "Empty" "export ")
   ("v" "Verbatim" "verbatim")
   ("q" "Quote" "quote")
   ("E" "Example" "example"))
 "Insert Block" "Block: ")

It's a bit ugly, but it works...

Re: Poll: new keybinding for org-insert-structure-template?

[Rasmus]

Eric Abrahamsen <eric@ericabrahamsen.net> writes:

> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>
>> Rasmus <rasmus@gmx.us> writes:
>>
>>> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>>>
>>>> Eric Abrahamsen <eric@ericabrahamsen.net> writes:
>>>>
>>>>> Rasmus <rasmus@gmx.us> writes:
>>>>>
>>>>>> Kaushal Modi <kaushal.modi@gmail.com> writes:
>>>>>>
>>>>>>> On Fri, Dec 15, 2017 at 6:23 AM Rasmus <rasmus@gmx.us> wrote:
>>>>>>>
>>>>>>>> The only way it’s "bad" is in the sense it limits the flexibility of
>>>>>>>> snippets, like "<s".  As you point out, if I add "<sr" as an "src R" block
>>>>>>>> I can no longer have "<s" as plain "src" block.
>>>>>
>>>>> I don't see any way around that. Any system that allows string keys of
>>>>> arbitrary length is going to run into that problem.
>>>>
>>>> One possible fix, a bit arbitrary: in the default value, provide <TAB>
>>>> as an artificial "stop key" in the sub-menus. So "s" starts the "source
>>>> code" sub-menu, and a <TAB> after that simply inserts "#+begin_src", and
>>>> leaves point after that.
>>>
>>> Yeah, I tried to suggest that earlier (unless I didn’t say it), but I
>>> might not have expressed the idea in an understandable manner :)
>>
>> Maybe I missed it!
>>
>>> I think that would be the best approach, but there’s no infrastructure
>>> that I know of that does this ATM (but I haven’t had a lot of time lately,
>>> so my knowledge on this issue is limited!).
>>
>> Can't we do this with tempo? It will have to be "handmade", not
>> automatic, but:
>>
>> '(("s" "Source Code")
>>   ("se" "Elisp" "src elisp")
>>   ("sp" "Python" "src python")
>>   ("TAB" "Empty" "src ")
>>   ...etc
>>   ("e" "Export Block")
>>   ("eh" "HTML" "export html")
>>   ("el" "LaTeX" "export latex")
>>   ("TAB" "Empty" "export ")
>>   ...etc
>>   ("v" "Verbatim" "verbatim")
>>   ("q" "Quote" "quote")
>>   ("E" "Example" "example")
>>   ...etc
>>   )
>
> Ahem, should have actually tried that first:
>
> (org-mks
>  '(("s" "Source Code")
>    ("se" "Elisp" "src elisp")
>    ("sp" "Python" "src python")
>    ("s\t" "Empty" "src ")
>    ("e" "Export Block")
>    ("eh" "HTML" "export html")
>    ("el" "LaTeX" "export latex")
>    ("s\t" "Empty" "export ")
>    ("v" "Verbatim" "verbatim")
>    ("q" "Quote" "quote")
>    ("E" "Example" "example"))
>  "Insert Block" "Block: ")
>
> It's a bit ugly, but it works...

Great find; I didn’t realize we can use tab here!

I’ll try to build the mks list automatically.  It will be a bit annoying,
as we’ll have to figure out valid keys for things like "prop". 

Rasmus

-- 
Lasciate ogni speranza o voi che entrate: siete nella mani di'machellaio