Re: Is Elisp really that slow?

[Ergus]

On Sun, May 19, 2019 at 07:35:58PM +0900, 조성빈 wrote:
> 2019. 5. 19. 오후 5:16, Van L <van@scratch.space> 작성:
>
> > Emanuel Berg writes:
> >
> > > Ergus wrote:
> > > I don't consider myself an Emacs expert -
> > > far from it. But I've been here for 10+
> > > years, so I'm happy with my Emacs and my
> > > skill level.  But this place still
> > > doesn't feel like home!  That is strange.
> >
> > Is it possible to have the best of all possible
> > worlds?
> >
> > For conservatives, a winter release of old gold keybindings.
> > For the free radicals, a spring release with modernizations.
>
> What if having a compatibility-mode that can be activated by something like:
> ```elisp
> (classic-keybindings-mode 1)
> ```
> and refine the default keybindings to be more consistent/mnemonic? People who 
> miss the old keybindings will be elisp-proficient; Adding 1 s-exp to the init 
> file won’t be a barrier.
> For the refined keybindings, Spacemacs can provide a good starting point.
This is interesting. (of course spacemacs not in evil mode).

I will just tell an idea I had (before knowing how things work in emacs,
I actually thought it was implemented in that way)

If we create a single file with the full list of general bindings as
variables for all the main modes like

;; Basics

C-a begin-line-binding
C-e end-line-binding 
C-b go-backward-line-binding 
C-d delete-forward-binding
...

;; Prog mode
C-c C-c whatever-you-decide-binding

;; Minibuffer mode
bla bla bla
...

And so on.

All the derived modes can be changed to use the variables instead of
hard-code the bindings in multiple files.

this:
(define-key c-mode-base-map "\C-d" 'c-delete-forward)
will be:
(define-key c-mode-base-map delete-forward-binding 'c-delete-forward)

This is similar to use remap, but with some advantages:

1) The bindings will be organized in a single file and the collisions
will be exposed easily. And fixed for all of them.

2) In case a mode adds a new functionality and it needs a binding, it t
in the global map, it will be clear if some others already have a
similar one and will bind to the same using the same criteria. If the
binding has never been used, it will be added to the list for future
modes that wants the same functionality.

3) Old users will be minimally affected because the starting criteria
are the actual bindings.

4) Implementation of test modes (like ergoemacs or evil-mode) and future
modifications will require less effort and the final experience will be
less hacky than now..

5) If a user personalize a binding (changing the value of one of these
variables) the changes will apply to all the packages and modes she
uses in his section.

6) If the user wants to use remap or hard-code the binding (as now) the
actual behavior will be unchanged.

7) Packages like elpy, irony, or company will also use the "binding-list",
so the user will only learn the most specific bindings (if any) so
external packages will be standardized with minimal effort.

8) Future bindings changes that the community agree will not require
changes in the files of all the modes. With the potential side effects.

I know that proposing this here is a crazy idea, but the intention is
not to start a crazy war again, If it is crazy (for technical reasons)
just ignore it. Actually I just mention it because it seems to me like
something extremely simple-advantageous and maintainable. With minimal
affections to the user space.

But remember that we have a limited number of bindings so managing them
better is important if we want to find place for new functionalities.

In any case I will leave this threads as it is not going anywhere (as
usual with this topics) and what is obvious for me seems to be heretic
for the rest. But you are free to consider the idea if it has anything
useful for emacs..


> > When I use a long M-x sequence, a shortcut suggestion appears. It disappears 
> > before I can catch it. Can it stay for 30 seconds? Can there be an instant 
> > interactive override to set it whatever you like?
>
> I would like a semi-AI that suggests interactive functions based on key presses 
> or actions the user performs... `You can use C-e (goto-end-line) to perform 12 
> keystokes you performed.'
> Saying about discoverability, I would like a context-sensitive right-click 
> mouse menu, something like Microsoft Office. Most newcomers are familiar with 
> finding functionality with the mouse; and it isn’t intuitive to find new 
> keybindings/functions that Emacs provide to boost productivity. (Actually, 
> that’s one of my problems; how should I find new functions...?)
>
> > Evolutionary programming of popular custom keybindings collected at upstream 
> > and put thru obstacle course competition is one way of composing a spring 
> > release.
> >
> > --
> > © 2019 Van L
> > gpg using EEF2 37E9 3840 0D5D 9183  251E 9830 384E 9683 B835
> >                             "The interface is a nightmare." - Brendan Schaub