Hi Alfonz!

• 'Colemak' is spelled with a 'k' at the end. Thank you. ;-)
• The images are obsolete with a vengeance. I never got around to figuring out the perl+something+browser thing to make new ones.
• Ctrl+Shift+1–4 work fine for me (1 shows the image, 2 changes layout etc). Maybe you have something else remapping these?
• You can change those hotkeys to something that works for you in the pkl.ini file.
• In my Colemak[eD] PKL layout, Shift+End does select text whether I use the physical Shift or S with Caps+Y for End.
• When using the physical End key with the Colemak[eD] you're right; it doesn't work. This is likely because I've made VirtualKey mappings for the physical End key so I can use it with Caps for keyboard mousing. I never use it as End anymore since I have the better Caps+Y for that. If you need it back, comment out the mapping of 'SC14f = END' in the layout.ini file and possibly in pkl.ini too.
• For macros, you'd have to either edit the PKL source code or run some other program. I'd prefer the latter.
• I'd like to have ligatures though (sequences of characters mapped to one key press) but they're only available for Extend I think.
• You have to add the OS dead keys of your keyboard to pkl.ini in the 'systemsDeadkeys = `´¨^~' line to avoid the space. Only those.
• Then, when you press the 'deadkey5' key for instance the layout.ini has a [deadkey5] section describing what happens next.
• If you press the deadkey5 key twice or with space, it should release the '0 =' symbol (in this case the ~ glyph); but that's a bit buggy...
• There's not much by way of background reading other than what I've linked to in my sig posts. Did you find the AHK forums post on PKL?

Hi DreymaR,

sorry for the late answer, I did not have a time to test it. I'll reply on few things:
- Colemak it is, sorry, I've overlooked it.
- about that deadkeys: SEE LAST EDIT!! I probably did not understand your answer. Using colemak-eD_ISO layout and pkl.ini containing "systemsDeadkeys = `´¨^~", when I press any physical key which relates to listed deadkey, then dead key is not dead, it's get printed instead. I would expect, that when I press any listed dead key, no output will be printed. Then when I press following key, these two symbols will be somehow combined if possible. And when I press dead key followed by space, then dead key followed by space is printed. But that's not happening. When I press dead key, this deadkey is immediately pressed followed by space. Example: I've got no opportunity to type let's say í using dead key ´ followed by i, because when I press ´ it will be printed immediately followed by a space which I haven't pressed. Where's error? In my thinking/expectations or somewhere in configuration?

EDIT: I do believe, that configuration that ´~ and others are dead keys is lost somewhere. Since when I remove them from systemDeadKeys the space suffix is lost when I type them, so systemDeadKeys controlls only that space should be added when I type ´ or ~ etc. Which is wrong anyway I think (in programming there's lot of places when I want to type them without that extra space). But the main problem is, that ´~ etc. isn't recognised as a deadkey.

DEAD KEYS WORK JUST FINE. I've created my own help image on paper, and there are for example two ~ symbols. One IS dead key, the second is not. This is where my confusement come from.

Thanks for the praise, Cevgar! You probably know I'm a sucker for praise and consequently would make your ANSI Wide(28) non-angled layout for you. Check the latest upload. ;) There are lines in the layout.ini to convert it to a Z-Angle layout should anyone want that.

In the pkl.ini you may cut and paste the Angle and Wide sections as necessary from the "repository" sections further down. I'd start with the Wide pkl.ini and replace the Angle section as that's simplest and most consistent.

I followed your advice regarding the systemdeadkeys setting. As for the layout= line, the ones included are a) the ones I use and b) only meant as examples. I don't agree that your suggestion is the most intuitive one: Rather, I like to start in Colemak[eD] (which is what the Colemak user would autorun PKL for) and have a VirtualKey QWERTY on backup so you can switch into that when someone without Colemak skills try to use your keyboard (and still have Extend mappings). More frequently I'll just disable PKL altogether at that point, granted. But the main point is that with this you can make anything you like yourself.

Thanks for the Wide ANSI! I can't actually use the Wide mods at home at the moment (I've rotated the keys for Plover (Plover Wiki), see  this post if you are curious about the keyboard), but PKL is made for travel use anyways. As for the other bit, I ... uh... didn't mean you should change the deadkey or layout settings for everyone. That was literally a note to myself, so when I download the file again in the future, and probably run across the same problems (such as mistaking vk_qwerty-No[rwegian]_ISO for vk_qwerty-No[t]_ISO, because I am apparently terrible at reading documentation), I'll hopefully look at the thread, read my own note and know how to fix the problem.

In defense of the qwerty start; qwerty is more practical for me (not necessarily for you or anyone else) on a "Hey you, check this out! This program/script is awesome!" angle. The extended caps layer already has a learning curve. I try not to complicate things for converts with having to deal with fiddling with the PKL program as well. If I need it just for me, I have the time to swap the layouts, or even edit the ini file.

Out of curiosity, does anyone know if the extended caps layer can be done through the registry? Or MSKLC? Or if something like this could be done through an Ergodox's firmware, or the Arduino Hardware Remapper?

Actually, speaking of suggestions, I do have one, though I haven't found a way to do it in PKL myself:

A key that will insert a horizontal tab character. Since, you know, the tab key's primary function these days has almost nothing to do with tabs or tabstops. Unless you are in a text editing application.

How, Windows: You can bypass Tab key behavior by inserting the Tab character directly with Alt Codes. The Alt Code for Horizontal Tab is 0009. To use it, you press and hold Alt, then type 0009 on the Numberpad (though sometimes you can get away with 009, 09 or 9). If your keyboard does not have a numberpad, you can't use Alt codes, because Windows is stupid like that. (perhaps the numberbar keys on the non shifted layout could be swapped for numberpad keys, leaving symbols on shift?)

How, AHK: An inelegant ahk solution might look something like this:

[hotkey]::
    Send, {AltDown}{Numpad0}{Numpad0}{Numpad0}{Numpad9}{AltUp}

How, PKL: ...I have no idea. Well, no idea aside from maybe adding a line to run an external script. I can get it to spit out 0009, and I can get it to use Alt+key, but I can't seem to get it to hold down the Alt while hitting multiple keys.

If it were working. I'd suggest the Tab Character for something like Caps+Shift+Space. (Of course the next problem would be getting a universal implementation of Elastic Tabstops)

Brilliant!! Just what I was looking for.

BTW, if you ever wondered why I didn't make a Wide Mod PKL setup when I first posted the idea... well, having seen my attempts at ahk code, now you know. ;)

Uhhh, you're mad. :D

Good call on generic {key} entries in layout.ini files. I had forgotten that (although I seem to have used it at some point).

The Extend layer I have includes mouse functions (GD/<> moves, BKM presses). Right?

I did include a MirrorExtend file (in the 'other' folder for PKL), but as PKL can only handle one Extend layer I'm not using it.

Mad, but brilliant. That's cevgar all right.

I don't think you could map mouse *movement* that way, but we might look into the mouse buttons – effectively doubling the number of buttons available. Interesting thought.

The question is: What key events do you miss when mousing? Back/Forward and Cut/Copy/Paste do spring to mind, reducing the need for repositioning of pointer and/or hands. Triple-clicking can be annoying so one might map Caps+Bt1 to that (which would also allow you to "rapid-fire cheat" in some browser games, hehe). Ctrl+Tab and Ctrl+Shift+Tab might actually be very handy, for switching tabs in browsers and elsewhere (although some use Ctrl+PgUp/Dn instead). Same with Alt+Tab.

I believe the MouseWheel functions are actually button presses of sorts as well. So with Extend you could wheel scroll horizontally (not always supported, but neat when it is). In X.Org the standard Y axis wheel "buttons" are numbered 4 and 5, and the X axis ones 6/7. Another fun thought is VolumeUp/Dn; maybe WheelClick should mute/unmute sound then. The main buttons could even be song back/forward!?

I've had mice with extra buttons that were used for BrowserBack and BrowserForward. I liked that, although I've grown very accustomed to Caps+F/P for it now. Might be redundant to map Caps+MBt presses to back/forward since that works so well with the left hand. I've also seen Cut/Copy/Paste mapped to mouse buttons. Even if the ZXCV Extend mappings cover those functions they're still a bit awkward to press, and when mousing the focus is on the mouse.

In *nix there's this handy tradition of middle-clicking to paste selected text. We could probably make that work by making Caps+MBt3 produce a Copy,MBt1,Paste chain of key events? That'd be quite awesome I feel. (Not sure how to do that in XKB actually, but in PKL it should be feasible.) Since you'd have to hold down Caps and middle-click to make it happen, there's not any big chance that you'd trigger it by accident. (I'm wary of using Cut instead of Copy for any such mappings though!)

Mapping a mouse click to a new Extend-type modifier isn't supported by PKL today because of the single-Extend-layer limitation. But it's a cool idea. Given a long mouse cord or a USB extender cable you might even use a mouse as a new kind of elegant ghetto foot switch? A mouse button might be easier to use with your foot than the keyboard buttons are. The catch to that is that you'd have to disable the mouse movement for that mouse (I guess?); a piece of black tape over the laser aperture should do the trick. Also, the computer can't see one USB (HID) mouse from another so you'd shanghai the whole mouse button functionality that way – maybe too high a price after all.

[Edit:] This really got me thinking. What if we could manage to use the shift states together with Caps+MBts? That'd give us a plethora of mappings! Example:
Caps+LBt: Alt+Tab; w/ Shift (or S) Alt+Shift+Tab (switch programs); w/ Ctrl ???; w/ Shift+Ctrl ???; w/ AltGr(LCtrl+RAlt) hotstring???; w/ Shift+AltGr hotstring???
Caps+RBt: Ctrl+Tab; ––"–– Ctrl+Shift+Tab (switch tabs); w/ Ctrl Mute; ???
Caps+MBt: PasteSelected (or clipboard if no selection – that should work?); w/ Shift Copy; w/ Ctrl Paste; w/ Ctrl+Shift Cut; ???
Caps+Wheel: ScrollX; w/ Shift Vol+/-; ???

Just throwing out ideas; the potential really is quite huge if we can do this.

[Edit2:] According to the AHK docs, this is the list of buttons we may remap by their AHK names:

Mouse (mouse hotkeys require Windows NT/2000/XP or later)
LButton - the left mouse button 
RButton - the right mouse button
MButton - the middle or wheel mouse button

WheelDown - this is equivalent to rotating the mouse wheel down (toward you)
WheelUp - the opposite of the above
WheelLeft and WheelRight [v1.0.48+] - these two require a mouse with left/right scrolling capability, but they have no effect on operating systems older than Windows Vista.
(See mouse wheel hotkeys to detect how far the wheel has been turned.)

Supported only in Windows 2000/XP or later:
XButton1 - a button that appears only on certain mice 
XButton2 - a button that appears only on certain mice

In this terminology (taken from XKB), buttons 4/5 and 6/7 are actually the Wheel Scroll Y and X axes which are handled as button presses! So yes, you could assign Enter to one of them ... but I feel that it might be a bit unintuitive? I think that for consistency wheel should primarily be kept to flipping between things – PgUp/Dn for instance.

I disagree with having an Extend button and modifiers on the mouse. What would you be using the left hand for (and you porn browsers out there need not answer that...!)? So the question is one of ergonomic benefits while using one hand on the mouse and one on the keyboard. Indeed, one-handed typing (mirrored?) combined with Extended mousing is a possibility – but I fear it'd be too tiresome and not efficient enough after all, like riding two horses. Anyone's free to try it out and tell me I'm wrong of course.

But back to ergonomy: Button chords on the mouse sounds like a bad thing to me, since the mouse hand gets enough flak as it is. The other bad thing when mousing is contorting that poor left hand to do stuff like Ctrl+Z, Ctrl+Tab, Ctrl+Q while mousing, and that's where a better way to reach such functionality that's useful while mousing could come in!

I realize from this that it's very much about the Ctrl button. The Extend chord Caps+T feels quite relaxed to me as a long-time user of it – but it can't be used for Ctrl+W for instance since the W key isn't accessible on the Extend layer. Also, hitting Tab with it is possible but far from nice. Maybe it's as simple as adding a Tab button on a Caps+MBt? Then Caps+A+MBt# flips between programs, Caps+A+S+MBt# reverses etc. The left hand, only making these Extend+mods chords, stays quite rested and free of twists but still active.

On the other hand, mapping a Caps+letter+MBt combo to Ctrl+Q or Ctrl+W is bound to create annoyance as windows are closed inadvertently. Better keep those hard to make.

Yes, the dreaded SwissGermanCaps. Apparently, they're not much used. As you say, MSKLC lacks the power of PKL to map practically everything around as you see fit. I still toyed with the idea of making a MSKLC layout with an extra AltGr set as it were – until PKL came along and caught my interest.

It was supposed to support "ligatures", that is, hotstrings up to 4 characters long. I don't think it does though. That's something I'd like to see PKL do better too: I'd like to have such ligatures or even strings of any length as dead key releases.

Okay, so I tried button-mapping with PKL out. Doesn't work so far.

In the pkl.ini [extend] section:

LButton = }!{Tab

This did nothing, but then again I've noted that keys (like the function keys) need to be mapped in the layout for them to work with Extend.

In the layout.ini [layout] section:

LButton = LButton    VirtualKey

This didn't enable the LBt to be Extend mapped, but disabled it completely! The other VK remappings are by scancode and I know no scancode/virtualcode equivalent for the mouse buttons.

There is actually a separate mouse hook (PKL, of course uses a keyboard hook), but I'm not sure that matters here. You can affect mouse buttons with the keyboard hook as we just witnessed.

I should check around for some AHK examples that may shed light upon the situation, I guess. But we may have overstretched here and reached a place where only editing of PKL source code would help...

That looks nice! I'll try it out.

Yes, PKL sends strings just fine (to get NumPad1, you'd need to use ={Numpad1}  [edited - I forgot the '=' ]). But not as dead key releases, which I needed. It'd be powerful.

The '2Ctrl' field is an inheritance from MSKLC, used to map Ctrl+keys which normally shouldn't be done as these are used for hotkeys. Ctrl+[ = Esc is an exception.

The CapStat field likewise. It denotes whether shifted keys and shifted AltGr keys are capitalized versions or not. If the shifted key is a capital version, add 1; if a shifted AltGr key is a capital version, add 4. So the D key where I have 'd D ð Ð' would have a CapStat value of 5. Not sure if it's important (probably not).

You can accelerate by adding a number after the key, e.g., {WheelUp 3}. Of course, it'll only be a multiplier and not true acceleration.

One thing: Send {Blind}{Key} may be good for mapping shift states as it preserves the modifier state. In Extend mappings, that's 'Blind}{Key' instead of just 'key' (not sure if you've guessed it but Extend mode adds a Send {} around your mapping so to cancel those braces I use a }{ pair - a trick I discovered a while back).