Emacs Control Characters

I prefer to use the Dvorak keyboard layout. Unfortunately, Emacs control keys were designed for a Qwerty layout. This is a frustrating problem. Try typing C-x b (switch buffers) or C-x C-f (open file) in Dvorak. Your fingers will be all over the keyboard. And the physical layout of my board is also Qwerty; so, when I look, I get really confused. I tried to solve the problem by reverse mapping the control keys back to Qwerty. Thus, normal typing would happen in Dvorak, but C-x b would become C-q x. And C-x C-f would be C-q C-u. Sounds good, right?

Not quite. At first it seemed there was no way to remap just the control characters. Someone suggested using psetf with global-key-binding. I tried this and it seemed to work with some control sequences but not others. I'll explain why later. It also only worked with a single control character. That is, C-t would work, but C-x b and C-x C-f would not.

The next suggestion was to try messing with key-translation-map. It works like this

(define-key key-translation-map "\C-t" "\C-s")

That will take C-t and turn it into C-s. This obviously won't work if you type two control keys or a control key followed by a letter. (I'm going to ignore more keys than that since I almost never use more than two.) But luckily that last parameter can be a procedure. It must have one argument (which always seems to be nil) and must return a string with the new control sequence. So once again it's looking like this is possible. I'll just use read-char-exclusive to read in the next character and construct the new control sequence. And this almost worked, but there's a gotcha.

The first thing I needed to do was create a hash table to map keys back to Qwerty. An example of how to do this is shown below. I thought I could use the same hash table to remap the control keys as well. But it turns out you can't break apart "\C-t". Emacs considers that a single character string. So trying to concat "\C-" to "t" doesn't work. I had to make another hash for control characters. It wasn't until later I learned you can use an alternative form like (kbd "C-t"). And this will let you concat "C-" to "t".

(setq dvorak->qwerty (make-hash-table :test 'equal))
(puthash "t" "y" dvorak->qwerty)

Here's the trouble. I mentioned before that certain control characters didn't seem to work. It turns out, if you're not in Xwindows (console Emacs only), it's not possible to get certain control sequences. Check here. There are 32 control characters in ASCII. That's how many control characters are available to Emacs. But there are more than 32 characters on the keyboard. Go ahead. Count. It's OK. Notice that C-. is not in there. In order to map C-. back to its Qwerty equivalent, C-e, you would have to know when it was pressed. Try the following code snippet Type C-x C-e after the right parenthesis of each to execute.

(equal ?\^. ?\^n)
(+ 0 ?\^.)

The ?\^x syntax is for characters. Do you see how they both map to 14? So "\C-." was being seen by emacs as "\C-n" which explains why it wasn't working. In addition, some other keys, like tab and backspace, were mapping to other control characters I needed which really goofed things up. At this point I've somewhat gotten used to the control keys in Dvorak; although, it is an ugly dance my fingers have to do on the keyboard to get there. My options at this point are to stick with Dvorak or to try and figure out how to capture all the control keys in Xwindows and use Dvorak in Xwindows and Qwerty in a console. When my Xwindows is running and stable, this is normally what I would use. For now I'm going to leave it as is. You can find the code in mapback.el.txt on my free software page in my backup files. It's named with .txt so emacs doesn't try and load it. If you learn more please contact me.