When using smarter-backward to jump defined number of words the behavior is not correct on the current master.

Testcase
One Two Three Four
Five Six Seven Eight

Jumping words in the above with 2 followed by smarter-forward (C-u 2 o) jumps two words at a time as expected. For the equivalent 2 followed by smarter-backward jumps are however not correct. In the above testcase jumps are both one and 2 words in length, while all these jumps should be two words long.

In case it helps: error-message displayed: "Looking-back: Invalid search bound (wrong side of point)"

Sometimes, I want to chenge the meaning of, say, boon-smarter-downward in specific buffers. For example, it makes more sense for me to use org-next-visible-heading binded to the same key as boon-smarter-downward.

Is it possible to set the keymap changes only for the current buffer?

Like d operator but with kill-ring-save instead of kill-region maybe using y in qwerty.

When chinese-py is selecteed as input method, moving commands (i.e. "o") in boon command mode trigger inserting characters, which is not expected behavior.

Steps to reproduce:

1. emacs -Q
2. Load the following config:

 (require 'package)
  (setq package-archives '(("MELPA" . "https://melpa.org/packages/") ("ELPA" . "http://tromey.com/elpa/") ("gnu" . "http://elpa.gnu.org/packages/") ("org" . "http://orgmode.org/elpa/"))
	load-prefer-newer t
	package-user-dir "~/.emacs.d/elpa/"
	package--init-file-ensured t
	package-enable-at-startup nil)
  (unless (file-directory-p package-user-dir)
    (make-directory package-user-dir t))
  (setq load-path (append '("~/.emacs.d/site-lisp/") load-path))
  (setq load-path (append (directory-files "~/.emacs.d/site-lisp/" t "^[^.]" t) load-path))
  (setq load-path (append (directory-files package-user-dir t "^[^.]" t) load-path))
  (add-hook 'after-init-hook #'package-initialize)
  (package-initialize 'NO-ACTIVATE)
(require 'use-package)
(use-package diminish :ensure t)
(require 'bind-key)

(use-package boon :ensure t
:config
(use-package boon-qwerty))

3. M-x turn-on-boon-mode
4. C-\ chinese-py
5. o ;; the chinese input is triggered instead of moving next line

I saw in #25 and #31 that there's some interest in having boon state keybindings in a per mode basis. I'd like to share one way of how I've implemented it. If there's enough interest in the full functionality of this, I can make a PR. Or maybe it could be put in the boon wiki.

The idea

The following buffer local symbols are added to emulation-mode-map-alists:

• One for the major-mode: boon-major-mode-map-alist
  • The format of its value is the same as boon-mode-map-alist.
  • It updates to point to the keymap alist for the active major mode (e.g., boon-org-mode-map-alist) or nil.
• One for minor modes: boon-minor-mode-map-alist
  • The format of its value is the same as minor-mode-map-alist.
  • When there's a change of boon state, it updates to point to a corresponding keymap alist:
    • boon-minor-mode-command-map-alist
    • boon-minor-mode-special-map-alist
    • boon-minor-mode-insert-map-alist

The functions boon-boonify-major-name and boon-boonify-minor-name setup the extra keymaps and alists for a mode.

Sample implementation

To keep things simple, I've done it only for command state:

boon-boonify.el

Sample usage

With org-mode, I'd like to follow links with the Enter key when I'm in command state, but always insert newlines otherwise.

First, I run this once:

(boon-boonify-major-name 'org-mode)

It will create a keymap named boon-org-mode-command-map. Then:

(define-key boon-org-mode-command-map (kbd "RET") 'org-return)
(define-key org-mode-map              (kbd "RET") 'boon-newline-dwim)

(edit: moved code to gist; improved the idea a bit)

If I select a region then go into insert mode the region is unselected, is there an option to stop this from happening ?

This is quite annoying if you use expand-region in command mode then want to enter insert mode for example.

I've come across a mode that binds something useful to C-u C-c C-<something>. I could work around this with some custom configuration, but in the spirit of being as compatible with existing modes as possible, I'm wondering if it would be worth adding a binding that resolves to C-u C-c C-.

I can suggest using C for this.

Make boon motions and operations more discoverable by using whichkey.

So while modes in the Boon Special Mode List start in SPC mode, they still enter CMD mode on ESC, requiring vto return to SPC mode. (I'm using a CapsLck-as-both-Ctrl-and-ESC thing, so it's easy to accidentally turn on CMD mode in a SPC mode buffer.) Can't this just be blocked?

Selection of the current line with SPC SPC doesn't work on the last line of the buffer (it has an entertaining behaviour on the very first character of the last line though).

The special buffers sections last sentence says:

You can get the original behavior of x keys by prefixing it
with (').

But this doesn't seem to be true, because x works, the minibuffer shows: x-
and for example: x o switches to another window.

Pressing: ' x
in for example a help buffer, shows:
Buffer is read-only: #<buffer *Help*>

Maybe the last line in the special buffers section could be removed.

Everything gets highlighted when searching for a: . (period)

one.
two.
one.
two.

When e SPC searching for: .
before pressing RET then the cursor jumps to the next occurrence of a ., and all periods are highlighted, but when RET is pressed, then all characters becomes highlighted.

This is most likely because the . is detected as a regexp "any" character.

In this state, the region is below the highlight, this makes it difficult to see what's selected.

The workarounds are to either search for a short word, or to un-highlighting with:
C-x w r RET or M-s h u RET

But it might be better if everything didn't become highlighted when one is trying to jump to a period by searching for it.

boon-20181014.927
GNU Emacs 26.1 (build 1, x86_64-w64-mingw32) of 2018-05-30
Windows 10

There used to be two prefixes (e and w) which lead to a number of search commands. E.g. e was bound to isearch.

After an update to the latest build, today, they appear to have been replaced by boon-navigate-forward (resp backward). And I can't locate bindings for isearch, for instance.

Should I be able to use boon in special buffers like messages buffers, currently its active for most buffers but not for buffers like this.

Also similar question for things like helm magit and diff should it be easy enough to make boon work with these perhaps, trying to replace shortcuts so everything is consistant across all buffers.

I use projectile, which defines a keymap, I bound this keymap to C-x p like this:

(use-package projectile
  :ensure t
  :bind-keymap
  ("C-x p" . projectile-command-map)

However, if I use x p in command mode, emacs complain that (from *Messages*):

use-package-autoload-keymap: Key sequence x p starts with non-prefix key x

I'm not sure what I should do to make this work properly.

I'll transform the answer in this thread into a documentation PR.

Deletion appears broken when doing multicursor replace. How to replicate

1. In command mode "r @ w d" on a word.
2. Write some chars
3. Try to delete using Backspace (backward-delete-char-untabify)

Only the last occurrence of the word will have chars deleted, all other instances will be untouched by the deletion.

I've been looking at Boon as an alternative to Evil as the keybindings make a lot more sense on Colemak, but the big thing missing seems to be the "repeat last command" and in your todo.org you say you don't know how to implement it.

Have you considered using the dot-mode minor mode to get this to work? See: https://github.com/wyrickre/dot-mode

Based on the description it looks like it should get most of the functionality that is needed (if not all).

When I am in a buffer using command mode and switch to insert mode if the buffer is read only it causes issues.

Basically it should stay in command mode currently it seems to leave me in a mode that is neither insert or command mode so my keybindings no longer work.

Perhaps check current buffers status before changing mode no sense even trying to use insert mode if the buffer is read only.

I noticed that you go through some effort to bind escape to helm-keyboard-quit. Why is that? It looks like someone else has encountered (possibly) the same problem and I've pointed them to your config here. Maybe you could elaborate why it had to be done that way and couldn't be done with a simple:

(define-key helm-map [escape] 'helm-keyboard-quit)

For me personally, having a special mode for special buffers like dired causes confusion. In ansi-term for example, I have to quote x or call boon-insert before typing. Is there a built in way to use Insert mode for special buffers? I am brand new to elisp, but I am willing to do the work if you could point me in the right direction.

Thanks!

https://github.com/jyp/boon#comparison-with-other-modal-layers-for-emacs

The last words In the RYO modal mode sections second sentence:

Compared to Modalka, it provides support for repeating a coentirely deactivated.mmand.

Boon-c-god does not allow to bind lambda functons.
For example, (bind-key "C-c C-b" (lambda () (interactive) (org-edit-special)) org-mode-map) and c b in command mode does not execute the lambda.

The issue in the code is indicated below:

(defun boon-c-god (arg)
...
    (while (and binding
                (or (eq binding 'mode-specific-command-prefix)
                    ;; if using universal prefix, the above will happen.
                    (not (symbolp binding)))) ;; <<--- this line only considers symbol bindings

So while eshell and shell allow for toggling between CMD and INS, term/ansi-term start in SPC mode, which hijacks x which means no shell command with the letter x can be used.

e s jumps to occurrences that are part of words.

one
two
bone
two

With the cursor on the first line.
pressing: e s
highlights the first word one, but it also highlights and jumps to one in the word bone

boon-20181014.927
GNU Emacs 26.1 (build 1, x86_64-w64-mingw32) of 2018-05-30
Windows 10

I want my LaTeX documents to start in insert mode rather than command mode, but adding

'boon-set-insert-like-state
or
'boon-set-insert-state

to the hook does nothing. Documents are still opened in CMD mode.

Is there a possibility for the cursor face of insert mode to be different from the one of command mode?

feature request

When making a selection by pressing "SPC" in Command mode, some strange behavior appears:

Replicate:

1. Press "SPC". Minibuffer shows "mark 0"
2. Press any arrow key ex. "down", mini buffer displays "Wrong type argument: number-or-marker-p, down"

In my setup the down key is bound to " 'next-line". So is "k", however if I press "k" rather than down, the next line down is selected as expected. Also if I press "k" or any other non-arrow key bound to moving between lines or chars first, I can freely use the arrow keys to control the selection afterwards. There is only a problem if an arrow key is the first key pressed after "SPC". The problem is not present if I envoke 'set-mark-command

To reproduce:

1. Enable boon everywhere with boon-mode.
2. Open a file tracked by git, then M-x magit-blame. I'd expect that mode line to show "SPC" at this point but its shows "CMD" instead. The keys from boon-command-map block the keys of the magit-blame keymaps.
3. In the same buffer, run M-x boon-local-mode twice to turn the mode off, then back on. Mode-line now shows "SPC" and the special map is active. Can use the magit-blame keys now.
4. However, after the magit-blame-mode quits, the buffer stays in the special state instead of going back to the state before magit-blame started.

Running master at commit 0c25219.

’M-x turn-off-boon-mode’ doesn't disable the keybindings.

The only difference seems to be that the modeline indicator is removed: Boon:

boon-20181014.927
GNU Emacs 26.1 (build 1, x86_64-w64-mingw32) of 2018-05-30
Windows 10

Hi! Just a small thing, but the documentation mentions that RYO modal doesn't work with the Emacs repeat command. Since version 0.4 (February 2018) this is no longer true.

When using org-goto (traditionally bound to C-c C-j) to navigate an Org document's outline, Boon immediately goes into command mode. However, the org-goto mode rebinds keys, in particular the arrows.

In order for the navigation mode to work well, I need to drop to insert mode. But really, it should open in special mode. Is there a way to make this outline navigation mode use the special mode ? If not, is there at least a way to make it start in insert mode?

When the cursor is on whitespace, and there is a whitespace character before the cursor, ex:

one  two

Place the cursor on the whitespace before the 't'.
Press: d v w

The minibuffer shows:
Wrong type argument: integer-or-marker-p, nil

It should probably have a better error message.

Possibly something like:
Word deletion failed, cursor not on a word.

boon-20181014.927
GNU Emacs 26.1 (build 1, x86_64-w64-mingw32) of 2018-05-30
Windows 10

c is bound to boon-c-god but it doesn't work as god mode so c x maps to C-c C-SPC x instead of C-c x

boon-smarter-forward raises the following error when point is on a closing quotation mark:

Wrong type argument: integer-or-marker-p, nil

The same happens with boon-smarter-backward when point is right after an opening quotation mark. Also, this error doesn't happen when major-mode is text-mode.

Test case:

;;  ,---- boon-smarter-forward from here
;;  v      
"foo" "bar"
;;     ^
;;     '---- boon-smarter-backward from here

One potential fix is using reimplementations of er--move-point-backward-out-of-string and er--move-point-forward-out-of-string, like these:

(defun boon-backward-out-of-string ()
  "Move point backward until it exits the current quoted string."
  (when-let ((char (nth 8 (syntax-ppss))))
    (goto-char char)))

(defun boon-forward-out-of-string ()
  "Move point forward until it exits the current quoted string."
  (and (boon-backward-out-of-string) (forward-sexp)))

So that they don't call goto-char when (nth 8 (syntax-ppss)) is nil, which is the cause of the error message mentioned above.

Tested with:

• emacs 26.2
• boon 20190529.849
• expand-region 20190416.538

When the cursor is on the last line of a buffer, boon-open-next-line-and-insert fails with « End of buffer » (an, interestingly, moves the cursor to the end of the line).

First of all, thank you for the great package, I used xah-fly-keys and ryo-modal mode before but I like boon better, because it's so simple and small!

I often edit latex files with auctex and tried the command for multiple cursors on this code

\begin{labeling}[:]{xxxxxxxxxx}
  \item[vgl. \LR{loeffler1} -- \LR{loeffler2}] Viele übergewichtige Kinder
    und Jugendliche durch ungesundes Essen und Bewegungsmangel
  \item[vgl. \LR{loeffler3} -- \LR{loeffler4}] Ursachen des Übergewichts
  \item[vgl.\LR{loeffler5} -- \LR{loeffler6}] Psychische und physische
    Folgen des Übergewichts
  \item[vgl. \LR{loeffler7} -- \LR{loeffler8}] Diagnose durch den Arzt
  \item[vgl. \LR{loeffler9} -- \LR{loeffler10}] Eltern und Kinder arbeiten
    bei Therapie zusammen
  \item[vgl. \LR{loeffler11} -- \LR{loeffler12}] Vorbeugendes Verhalten
    gegen Übergewicht bei Kindern \bewp{}
\end{labeling}

So I'm on the word labeling and hit
s @ s g

and only the first appearance of labeling is matched. When I save the file as *.txt it works as expected. When I try to select the whole paragraph with SPC g it's the same. I works for *.txt but not for a latex buffer.

Can this be fixed?

Regards
Thorsten

Frontends should store mnemonics in the keymap, as if it were the name of a menu item. Example:

(define-key boon-command-map "a" '("around" . boon-enclose))

Then the tutorial will be able to lookup those mnemonics.

Following #40, it becomes much harder to use the magit-blame buffer: it's a read only buffer with some keys remapped. If I type q q, to quit, it will not use the remapping but try to write q in the buffer instead (which fails, of course). (as it happens, we can still quit the magit-blame buffer with c q, but enter will not work, for instance).

It may be related to #34 .

I don't have time to investigate at the moment. I'll try to get back to this in a few weeks.

First I set with (global-set-key (kbd "C-c f") 'ggtags-find-file).
If I press c followed by f in command mode, I get an error that C-c C-f is not bound. (No command bound to C-c C-f)
My examples:
C-c i -> No command bound to C-c TAB
C-c o -> No command bound to C-c C-o
C-c p -> No command bound to C-c C-p

I would like to understand the logic behind C-God. If I know correct, in the original God mode, c directly followed by i would mean C-c C-i whereas c followed by space and then by i would be C-c i.

I keep hitting issues which are likely not boon specific but effect boon and likely others will hit this type of issue.

Basically when using boon shortcuts I have defined something happens in the buffer which remaps the keys to different behaviors, the issue is I don't know how to find the cause

Any tips / suggestions on how you can find the source of this could you hook into lisp in some way when a boon key has been changed and what triggered the change perhaps ?

An example is in insert mode I can no longer type q this changes my buffer and spc goes to the bottom of the buffer instead of inserting a space.

Thanks for writing boon-mode. I enjoy it a lot and have spend some time setting up my own QWERTY mode. I have found an annoying bug. It can be reproduced like this:

1. Run emacs in daemon mode
2. Open a file
3. Close emacs using "C-x C-c"
4. Change the file outside of emacs
5. Open the file again.

Emacs will show the "Revert buffer from file " minibuffer, but it will not be active.

Looks like this may be the plugin I am after I wish to define my own modes and keys however.

Are there any existing examples of this the readme just talks about the boon keymap but it also mentions you could use boon-core to do your own.

What are the functions for defining a new mode and attaching key combinations to it perhaps something in the readme on this could be useful for others ? or have i completely mis understood and you cant define new modes and key combinations for the modes ?

It seems that you mention several times in the comparison that several packages (modalka and ryo-modal) are similar to boon-core.el but pale in comparison in several ways. As someone interested in generating their own modal editing mode based on their own editing habits who's looked at modalka and ryo-modal before and found them lacking and poorly integrated at times, this leaves me very interested in working with boon-core.el.

Are there any plans to document its usage (other than the existing code already using it) or, even better, separate it out and consider boon to be a modal editing package built on top of it?

This is a minimal config for boon, I use a local mirror of melpa but it's up to date.
If I type M-x I can't go out by typing esc, and a warning is shown "Wrong type argument: commandp minibuffer-keyboard-quit"

       (require 'package)       
       (setq package-enable-at-startup nil)

       (add-to-list 'package-archives  '("melpa" . "~/elpa-mirror/melpa"))
       (add-to-list 'package-archives  '("org"   . "~/elpa-mirror/org/"))
       (add-to-list 'package-archives  '("gnu"   . "~/elpa-mirror/gnu/"))
       (package-initialize)
       
       (unless (package-installed-p 'use-package)
         (package-refresh-contents)
         (package-install 'use-package))
       (use-package boon
         :ensure t
         :init
         (require 'boon-qwerty)
         :config
         (boon-mode)
       )

http://kakoune.org/why-kakoune/why-kakoune.html

Cheat sheet colors, hard to read black text on blue background

The cheat sheets that are linked in the https://github.com/jyp/boon#usage section:
https://github.com/jyp/boon/blob/master/colemak.pdf
https://github.com/jyp/boon/blob/master/qwerty.pdf

The blue color probably needs to be changed, because it's hard to read the black text on the
blue (#0000ff) background color.

w s doesn't jump to the next occurrence of a word under the cursor.

one
two
one
two

With the cursor on the letter o on the first line.
pressing: e s
highlights both occurrences of one, and the cursor jumps to the second occurrence (as expected).

But if the cursor is moved to the t on the last line.
pressing: w s
highlights both occurrences of two, but the cursor doesn't jump to the previous occurrence.

boon-20181014.927
GNU Emacs 26.1 (build 1, x86_64-w64-mingw32) of 2018-05-30
Windows 10