Hello DreymaR,
Thank you for your interest. I didn't look at your layout before placing the accents on mine, but l've just been looking at it now (Colemak[eD]) and I see that you had the same idea, sort of, which comfort me because I was't sure at all this was a good idea.
When I designed my variant of Dvorak, I placed the accentuated vowels on Alt. It was very convenient because they were all under the left hand, and I used my right ⌘-key as Alt. But I started using emacs since, and I needed my Alt key back... The problem with OSX is that we don't have AltGr and I try to stick with .keylayout files to be able to port my layout whithout needing additional software such as karabiner. So I decided to go with deadkeys instead. The middle columns were elected so that I can use both hands to reach them (To type both `à` and `è`).

I think that swapping Hk or HkLm is a great idea. I'm not sure whether I should fall back to your Hk or my HkLm. I like the idea of minimizing changes to the standard Colemak layout, but on the other hand swapping L and M seemed like a good idea at the time.

Good evening stevep99 :-)

Your answer makes me notice that I wrote "Q" instead of "W". My original idea was to swap "M" and "W". The SFU is increased by the digram "rm" on the left ring-finger as explained in the first post. I assumed it was a typo when DreymaR wrote "Q" but it turns out that the typo's mine...

Concerning M, H, and L, I'm still hesitating. Here's why:

Regarding letter frequency, I found two very good sources that I recommend, they used a very huge corpus from google and have detailed statistics:
- http://mdickens.me/typing/theory-of-let … uency.html (note: his programming corpus is not awesome, but better than nothing)
- http://mdickens.me/typing/letter_frequency.html
- http://norvig.com/mayzner.html

Now, the letter frequency of what I actually type is not the letter frequency of casual or formal english (even these two actually differ according to the first link).

I would weight what I write like this:
Programming: 70%
English: 20% (google search ; university assignments ; code documentation ; and very seldom: questions on StackOverflow or forums)
French: 10% (mainly chats and e-mails)

The letter H is not frequent in either programming or french:

e t a o i n s r h l d c u m g y f p w b v k x j q z     (English Casual | mdickens.me)
e t a r i s n o l c d p u f m g h b v x y w k q j z     (Programming | mdickens.me)
e a i s n r t o l u d c m p é g b v h f q y x j è à k w z     (French | wikipedia.fr)

Now, those are statistics found on the internet, which I only trust so much. But my experience tells me the same: H and W are not very frequent in my average typing while L is. And in fact, it's one of the reasons that drove me to quit Dvorak, because H is under the right index finger's resting position which is a waste for me (and also the very bad placement of R and L which are overused in programming or in the terminal e.g. "ls").

Despite not having a proper very large corpus, I draw the heatmap of the most frequent letters using Patorjk's keyboard layout analyzer with the source code of my last two programs (an android note taking app in Java (around 10k lines) and a compiler in Scala (between 2 and 5k lines)) as input. And the L key is a lot more frequent than either M or H. (H isn't even very frequent, as expected).

Here is the heatmap:

So, here's why I made those choices.

What do you think?

--------------------------------------------------------------------------------

Edit: looking at all this data now, I think that the HkLm cycle is pretty good for my usage, even better than my original Lmh

Edit 2: I started keylogging what I type to build a personal corpus, even though I know that when my job changes, the frequencies will too.

Edit 3: programming is a complicated beast because there are many different programming languages... But some letters will be overused in most of them because some keywords (such as "int", "float") are universally used.

Hi Julien

You've certainly done your research and even experimented with working out your own letter frequencies, which is very good! You've clearly thought about it quite a lot!

I wonder though - even as a programmer, you may spend more time typing natural language than you think. I seem to recall reading somewhere (sorry, don't remember where!) than programmers only spend about half their time typing code, and the rest is emails / instant messages / documentation / comments / filling in issue trackers etc etc.

Peter Norvig's data is good, I've looked at his stuff before. He also has H more frequent than L but the gap is narrower compared to the wikipedia data. But I accept of course that people's typing experiences varies, and so it might be possible that for some, L could be more common than H. I'm not entirely convinced it's worth switching, but if you're data shows that L is a lot more common for you, then having L on the bottom row and H on the top row should be OK.

I concur with DreymaR that changing M to a new finger is not a good idea. The costs are unlikely to outweigh the benefits, especially as M is not that common anyway. If you think the Qwerty-N position is better than Qwerty-H, you could switch M and K so that K occupies the Qwerty-H position, like DreymaR.

In summary: I think it's OK to switch around H,L,M,K amongst themselves if you really need to, but I'd advise against switching with keys on other fingers.

Thanks for your advice
For now, I'm trying HkLm and it seems ok. I can always switch H and M later.

btw, switching from Dvorak, the hardest to retrain for now is T, S and E.