Colemak UK keyboard layout

Since I managed to reserve a allocation for ColemakUK by mistake, I suppose I should ask for a review on this… ;)

I didn’t assign anyone to review these merge requests, thinking that I could do it later, but I can’t figure out how to do it in gitlab, therefore I’m asking for a review here instead. If anyone knows how to ask for a review properly, I’d be much obliged if you would tell me!

This change request has four parts:

1. https://gitlab.riscosopen.org/RiscOS/Sources/Internat/IntKey/-/merge_requests/2 (Keyboard layout definitions, see here for an image of the layout)
2. https://gitlab.riscosopen.org/RiscOS/Sources/Internat/Inter/-/merge_requests/1 (Necessary definitions)
3. https://gitlab.riscosopen.org/RiscOS/Sources/Internat/Territory/TerritoryManager/-/merge_requests/2 (Mostly relates to documentation I think)
4. https://gitlab.riscosopen.org/RiscOS/Sources/Internat/Territory/TerritoryModule/-/merge_requests/1 (Also mostly relates to documentation I think)

The first two are strictly necessary for this to work, but the last two may be optional. For completeness, however, all four should be merged if any, to avoid future complications. Some of the merge requests also contain some localisation, which I probably should have made separate requests for. Sorry about that.

I have tested this thoroughly on both Pi and RPCEmu, but not on any other hardware.

I didn’t assign anyone to review these merge requests, thinking that I could do it later, but I can’t figure out how to do it in gitlab, therefore I’m asking for a review here instead. If anyone knows how to ask for a review properly, I’d be much obliged if you would tell me!

Go onto each of your merge requests, and in the top right where it says “0 Assignees” select Edit and assign it to ROOL. Hopefully that will bring it to someone’s attention at some point.

With regards to Colemark UK keyboard layout, I’ve been looking for images of the Colemark Uk keyboard layout to add to my Vkeyboard application but could only find US keyboard layouts.

Does anyone know of an image?

Vkeyboard can be downloaded from:

http://kevsoft.co.uk

Julie:
Thanks, I’ll see if I can find my way to that.
Edit: Maybe I’m going blind, but I can’t seem to find it… It lookr like maybe Gitlab merge request don’t load entirely in my browsers (tried Firefox and Chromium) – something keeps blinking. I may have to close them and redo them.

Kevin:

Does anyone know of an image?

My implementation for Risc OS is this one:
https://gitlab.riscosopen.org/RiscOS/Sources/Internat/IntKey/uploads/5ab92ec07acb0e9c516ea5796548925c/ColmakUK.png

On GNU/Linux it looks like this:
https://github.com/skymandr/dotfiles/blob/master/i3/scripts/kbimg/key_gb(colemak).png

There are a few small discrepancies, to do mainly with limitations in the IntKey module without explicit unicode… The GNU/Linux version is, as far as I know, official, but unfortunately the image doesn’t show clearly which are dead keys. You can compare with my figure for that.

Maybe I’m going blind, but I can’t seem to find it…

You can only assign merge requests to people if you’re a member of the project/group. The git cheatsheet suggests that you do this by clicking the “request access” button that should be (somewhere) in the RiscOS group: https://gitlab.riscosopen.org/RiscOS. Once ROOL have granted access you should see the relevant options for setting & changing the assignee.

This will also give you the power to edit other people’s merge requests, which is pretty silly, but that’s how gitlab’s permissions work so we just have to live with it. (All this could be avoided if merge requests could have a default assignee)

All this could be avoided if merge requests could have a default assignee

From work experience of things where a single account is targeted you need to:

1. Ensure it’s a non-personal account so multiple people are given access
2. There’s a specific rota for a nominated person each day to look over everything in there¹
3. Someone feels able to hand out a swift kick to anyone not performing item 2 as required or deal with redistribution of the task in the case of sickness etc.

That’s not inventing new work BTW that’s pointing out the kind of thing you need to do to avoid process failures.

¹ Like for example someone on a rota to playback voicemail and process the requests, where the voicemail box is a set of patients ringing with enquiries about delays to operations (because Covid)

The git cheatsheet suggests that you do this by clicking the “request access” button

Thanks! That makes sense I guess, since Gitlab isn’t really designed for hosting just a single team working on a single project.

@Andreas thanks for that image, it seems that on Colemark each keyhas for different itesm, I will need to add another switch to Vkeyboard.

Hello. Any ideas as to when the colemak layout will be available on the stable release?

Sorry. I’m completely new to Risc os. I was just trying it out last week, and only dvorak was an option. I don’t know how often the releases are launched or how long would it take for a commit to be merged…

By the way, thanks to Andreas, for putting this together.

Hello. Any ideas as to when the colemak layout will be available on the stable release?

Sorry, no. I’ve requested access to the group on gitlab, which is a prerequisite for properly asking for a review. After that, someone has to actually review it as well, which might take some time. If you like, I could try to build an ROM for you and put it somewhere, though it would have to be 5.29 (unstable – current stable is 5.28).

Note that this is Colemak for UK keyboard (ISO), which will work with a US (ANSI) keyboard, but is different in some respects, since the physical keys are in different positions and the US has some other ideas of where to put second layer characters. If there is sufficient interest, implementing US would be easy enough.

There is also a third option called EurKEY Colemak-DH, which as the name implies is derived from a proposed European standard. This is something like a mix of UK and US layout, and is available both as ISO and ANSI (so possibly a fourth option), but then we would be swamping the keyboard options a bit with different Colemak variants, which may not be desirable…

Edit: I have now been granted access and have asked for review. Since this is mostly volunteer driven work, there is no ETA, however.
If you want this layout now rather than later, but don’t want to mess about with the unstable ROM you can also use this Obey script, provided you have KeyMapper ⁰ installed: https://github.com/skymandr/ColemakRiscOS/blob/master/ColemakObey/Colemak%2Cfeb It will put the alphabetic keys where you are used to them, but won’t do anything about second or third layers.

⁰ http://effarig.co.uk/riscos/ This comes pre-installed in for instance Risc OS Direct, so you may already have it. Typing *keymap in a command window will tell you if this is the case.

Note that this is Colemak for UK keyboard (ISO), which will work with a US (ANSI) keyboard,

Thank you for the clarification on this. I forgot all about it. I do have a couple of ISO boards somewhere that I could use to try it out.

I have now been granted access and have asked for review. Since this is mostly volunteer driven work, there is no ETA, however.

Glad to hear this. Finally!

Thank you for the link and needless to say, for the information you just provided to get this rolling.

I need to buy an SD card and get Risc Os on it. I only have one, but have been writing and writing so much on it, I didn’t want to take a chance anymore. It may be a few more days before I try it out.

Thank you again Andreas. I appreciate it.

Thank you for the clarification on this. I forgot all about it. I do have a couple of ISO boards somewhere that I could use to try it out.

I don’t have an ANSI keyboard myself, so I haven’t tried it, but should work on ANSI as well. There may be one or two keys that are mapped weirdly, but most of the difference is with second and third layer keys, which is because ISO/UK and ANSI/US have slightly different ideas on what to put where and not really to do with the mechanical layout as such.

It may be a few more days before I try it out.

I think KeyMapper contains instructions for how to install it. Ask here if you want to try that and have any problems, either with KeyMapper or with running the Obey script, when you get around to it. Good luck!

Hello Andreas. I thought of letting you know either way. Just needed to ask you. Are there any known limitations to install risc os direct on anything larger than 16GB? I know this is a bit off-topic, but I thought it was sort of relevant, since you mentioned keymapper is available there.

I didn’t have a problem by dd‘ing the ro528* img, or by using any of the recommended tools such as etcher, but I couldn’t go past the 64MB partition that the ro527Rdirect.img created. I don’t know. And I’m clueless in that all that is perhaps needed, would be just to create a couple of partitions first and before I try it out once again. But I sorta gave up, and just went ahead with the 528*.img. The latter is only a few hundred megabytes as opposed to 7GB and counting from the direct img. In this regard I would have preferred to have the direct img instead. But perhaps it was an obvious step, that I must have missed altogether, in order to transfer that direct img to the sd card. So I don’t know.

As far as the 2nd and 3rd layers go, I have no idea, if by just looking at the low-level internal key numbers, I’m on the right path either.

Going by the documentation, 108 for example, provides a conversion of Japanese, 6th row, between Space and Kana? Whereas on the implementation of colemak 108 returns keycode 92, wouldn’t it be fairly easy for the devs and/or you, to implement it as such, since 92 is nowhere to be seen. It says it’s keypad 3. What the heck is keypad 3 by the way? (I’m simply going by the scan codes that the documentation provides, and I’m only guessing. But I don’t have the slightest clue how risc os works, really.)

Anyhow, thanks again.

Also, you commented `| The distro I use map right-alt to \, this resets it:
KeyMap 96 96` . Were you referring to a gui in specific by any chance? Anyway. It’s not important.

Are there any known limitations to install risc os direct on anything larger than 16GB? I know this is a bit off-topic, but I thought it was sort of relevant, since you mentioned keymapper is available there.
.
I didn’t have a problem by dd‘ing the ro528* img, or by using any of the recommended tools such as etcher, but I couldn’t go past the 64MB partition that the ro527Rdirect.img created. I don’t know. And I’m clueless in that all that is perhaps needed, would be just to create a couple of partitions first and before I try it out once again. But I sorta gave up, and just went ahead with the 528*.img. The latter is only a few hundred megabytes as opposed to 7GB and counting from the direct img. In this regard I would have preferred to have the direct img instead. But perhaps it was an obvious step, that I must have missed altogether, in order to transfer that direct img to the sd card. So I don’t know.

I’m not the person to answers this, I think, so I would suggest asking this question in a separate thread under Community Support. I’m mostly using the ROD distribution myself, in part because of the larger default image size.

As far as the 2nd and 3rd layers go, I have no idea, if by just looking at the low-level internal key numbers, I’m on the right path either.

Just to be clear, by 2nd and 3rd layers I’m referring to what the Shift and Alt key behaviours respectively. The Obey script doesn’t (and can’t) do anything about these.

Going by the documentation, 108 for example, provides a conversion of Japanese, 6th row, between Space and Kana? Whereas on the implementation of colemak 108 returns keycode 92, wouldn’t it be fairly easy for the devs and/or you, to implement it as such, since 92 is nowhere to be seen.

I’m not sure what you are referring to here. Where do you get the number 92? Since I don’t have a Japanese keyboard I haven’t tried this, but in my merge request above, I don’t touch the 108 (0x6c) key, which means it is treated as default, which should mean that it works as expected. In the Obey-file version of things, I only touch the alphabet keys.

It says it’s keypad 3. What the heck is keypad 3 by the way?

This, I believe, refers to the the number 3 on the numerical keyboard part of a full-size keyboard.

(I’m simply going by the scan codes that the documentation provides, and I’m only guessing. But I don’t have the slightest clue how risc os works, really.)

As you may have discovered, there are (at least) three different key code definitions working in parallel in Risc OS:

• https://www.riscosopen.org/wiki/documentation/show/Low-Level%20Internal%20Key%20Numbers – These are used internally, for example by the IntKey module and by the KeyMap module. They are ordered according to the physical layout on a keyboard.
• https://www.riscosopen.org/wiki/documentation/show/Keyboard%20Scan%20Codes – These are used for, well, scanning keys I guess. They are accessible using SWI:s for use in programming. Their order is sort of a logical map of what keys represent.
• https://www.riscosopen.org/wiki/documentation/show/Character%20Codes – These are used by the WIMP.

There is a program called StrongKey which can be used to sort this out a bit, you can find it here http://www.stronged.iconbar.com/fjg/

Also, you commented
| The distro I use map right-alt to \, this resets it:
KeyMap 96 96
Were you referring to a gui in specific by any chance?

ROD by default (and without telling you) already uses KeyMap to redefine some keys (right-alt/alt-gr and windows keys) on boot. I added this line to override those settings, before I figured out how to simply remove them (this setting was hidden in $.!Boot.Choices.Boot.PreDesk.Keymapper.!Run – the KeyMap module is also loaded on boot from the same directory if you are using ROD).

Going by the documentation, 108 for example, provides a conversion of Japanese, 6th row, between Space and Kana? Whereas on the implementation of colemak 108 returns keycode 92, wouldn’t it be fairly easy for the devs and/or you, to implement it as such, since 92 is nowhere to be seen.

Easy? Hmmm, have a peek at the source for the keyboard handling.

Andreas is still sane, so he’s obviously not gone too deep.
Part of me has been barking for years so there was no harm in me nosing around, but it’s scary down there.

This, I believe, refers to the number 3 on the numerical keyboard part of a full-size keyboard.

It does.
I keep one for plugging in to the laptop for certain test work.
I was thinking of parting with a little cash and getting a wireless one.

ROD by default (and without telling you) already uses KeyMap to redefine some keys (right-alt/alt-gr and windows keys) on boot. I added this line to override those settings, before I figured out how to simply remove them (this setting was hidden in $.Run – the KeyMap module is also loaded on boot from the same directory if you are using ROD).

BTW. It doesn’t just remap keyboard keys, because of the complicated nature of the keyboard handling the mouse keys are included, so you can swap them around.
See the low level key table…

Andreas is still sane, so he’s obviously not gone too deep.

About that…

by 2nd and 3rd layers I’m referring to what the Shift and Alt key behaviours respectively.

Yes. I was thinking of the different layers/levels by which colemak maps the keycodes. So pretty much is in that combination of alt key and alt-shift respectively… Of course, I noticed under risc os, alt key combination already provides a wide range of these characters, save the extra dead key presses of which I’m not very fond of. Frankly. This is only from what I’ve seen on the surface.

Where do you get the number 92?

Deep Thought, Andreas. It’s the answer to life, the universe and everything. :D

Thank you though for the link to StrongKey. This was a valuable info and program I was blissfully unaware of. Had you not provided that link, it would have taken me a while before I were to come across. Very nice.

But take for the example the following: if I were to type on StrongKey internal field, the 108, then the field that corresponds to modifier keys is obviously left blank, or to be more precise, it only shows &-(-), which I presume no modifier keys have been assigned. IF I were to type another keycode that maps modifier keys — whatever these are, then it would show accordingly.

Yes. Perhaps my flaw in reasoning lies on that GNU/Linux mindset. But the beautiful part about it, is that neither one diverges, as much as I would have thought at first. There’s definitely some consistency, notwithstanding the fact both are different systems.

I keep one for plugging in to the laptop for certain test work.

Well. How much testing could be done with just a keypad? I’m not implying is not important though.

Thank you again for the initiative to get this going Andreas.

Deep Thought, Andreas. It’s the answer to life, the universe and everything. :D

I think you’ll find that’s 42. ;)

@Andreas

I have a test Vkeyboard with Colmak UK layout at

http://kevsoft.co.uk/test/test.zip

If you could please test and correct the mistakes I have bound to have made.

Thanks

@Kevin

Great! I’ll have a look. The Colemak UK keyboard is also in the ROM now – many thanks to @ROOL for review etc.!

@Andres

Thanks, it should be easy to update any errors I made, the Add/Edit Board button on the main window allows you to either add or edit existing boards.

The boards live in !Vkeyboard.Boards