Sorry if this has already been brought up before, but this forum doesn't seem to have any decent way of searching within a topic...

Has anyone else had an issue where enabling the `misc:extend` option disables the Ctrl+Alt+Fx shortcuts for switching to console? I really like the extend layer but as a relative newbie to the Unix game, the console is super important for when (and not if) I break things.

I imagine I can define my own shortcuts to replicate what the built-in shortcuts are doing, but if this is a known issue with a relatively easy (or at least cleaner fix), I'd much prefer to go about it that way.

Thanks for clearing that up! Annoying that the Ctrl+Alt shortcuts are messed up by the extend file... I'd definitely like to dig into it at some stage, but I have quite a number of things on my plate at the minute (I suppose we all do...)

Also big kudos for putting this mod out there, it makes things a lot easier than having to write all of these from scratch! :D

Hey all, seeing some issues with the mod after an update to xkeyboard-config (from 2.29 to 2.30) on Manjaro 20.0.3 (based on Linux 5.4). Originally had the layout hard enabled in /etc/X11/xorg.conf.d/00-keyboard.conf, like so:

Section "InputClass"
	Identifier "system-keyboard"
        MatchIsKeyboard "on"
        Option "XkbModel" "pc104curl-z"
        Option "XkbLayout" "us(cmk_ed_us)"
        Option "XkbOptions" "misc:cmk_curl_dh,misc:extend,lv5:caps_switch_lock"
EndSection

This worked fine until my next boot after that upgrade (there were a couple of other upgraded packages but I can only presume this is the one causing the issue?)

Anyone else who upgraded xkeyboard-config seeing this issue?

I am very much enjoying this Extend functionality! But I'm having one issue. When I'm using Google Spreadsheets, whenever I hit Caps Lock (which is mapped to ISO_Level5_Shift) it clears the current spreadsheet cell. I have to use these spreadsheets for collaboration at work and I'm getting a reputation as the person who accidentally deletes other people's data.

Is there some other modifier I could map Caps to that would avoid that side effect?

Thanks!

Yes, I have the same problem - and I'm not even using Caps. My Extend key is Left Alt, but sure enough, if I hit Left Alt in Google Spreadsheets, it does clear the current cell. However, I am remapping modifiers by keycode (in keycodes/evdev), so my LeftAlt key is still sending the CapsLock keycode to the OS. Therefore I conclude that Google Spreadsheets is listening for the Caps keycode, and that's what's causing us the problem.

This isn't often a problem, as most software doesn't use the hardware keycodes in this way, but I must admit I have come across it occasionally enough to be an irritant.

It further leads me to think there might be a ready solution. Perhaps we could have Extend mapped to an unused keycode so ensure it doesn't conflict with anything else, and bind the Extend feature to that keycode. Would be interested to know DreymaR's thoughts?

Yes, I am talking about Linux/XKB here. So, I have experimented a bit with this... but not very successfully. Something I don't really understand is going on.

I was thinking along the lines of using the function keys too. In fact, I did this in my AHK/Windows scripts, where I used F22 and F24 as my base mappings for Extend and AltGr layers, to make it easy to remap those functions to any arbitrary key. So doing something similar for Extend in XKB would be quite nice. To this end, my first attempt was just to see if Extend can be mapped OK to a function key. Taking this:

xkb_symbols "caps_switch_lock" {
  include "level5(lock)"
  key <CAPS> {
    type[Group1] = "EIGHT_LEVEL",
    symbols[Group1] = [ ISO_Level5_Shift, ISO_Level5_Shift, ISO_Level5_Shift, ISO_Level5_Shift,
                        ISO_Level5_Lock, ISO_Level5_Lock, ISO_Level5_Lock, ISO_Level5_Lock ]
  };
};

What I found is that you can use have like LALT and RWIN in place of CAPS, but for some reason it won't accept other keys. I tried FK12 for example, didn't work. Perhaps only special keys / modifiers are allowed there?

I then decided to just go with using RWIN as my Extend key, and then rebinding my physical LeftAlt key to be RWIN.

xkb_symbols "rwin_switch_lock" {
  include "level5(lock)"
  key <RWIN> {
    type[Group1] = "EIGHT_LEVEL",
    symbols[Group1] = [ ISO_Level5_Shift, ISO_Level5_Shift, ISO_Level5_Shift, ISO_Level5_Shift,
                        ISO_Level5_Lock, ISO_Level5_Lock, ISO_Level5_Lock, ISO_Level5_Lock ]
  };
};

As you'd expect, Extend works perfectly well. But if I go to Google Sheets, I can still reproduce the cell clearing problem. So it seems it doesn't really matter what the binding is, but I'm a bit confused about how the same effect is seen with both different physical keys AND different evdev mappings. Maybe it's something inherent with ISO_Level5_Lock?

As far as I can see, there is only a setting called "Enable compatible spreadsheet shortcuts". Enabling or disabling this has not effect on the bug.
I also confirmed the bug in both Firefox and Chrome, so it's not browser specific.
Problem is, ISO_Level5_Shift is so obscure and rarely used, that even reporting this bug would be difficult, let alone getting them to fix it.

Using xmodmap in the past (before I discovered Extend etc), I have set caps = Mode_switch, and that works as a modifier that doesn't bother Google Spreadsheets ... is that possible with xkb?

OR, is there a way to set the Caps key to work as a modifier in combination with another key, but when it's pressed on its own, not to send anything? I see that xkb has something called VoidSymbol but I don't know enough about the workings.

I think it should be possible to implement it using kmonad

Maybe it's a different scenario, but in this article it looks like they set Caps as a modifier and use VoidSymbol to prevent side effects: https://www.linux.com/news/extending-x- … d-map-xkb/

Really? That's in the "caps_switch" definition which I don't use myself, as I've used "caps_switch_lock". Furthermore, I read that as giving a ONE_LEVEL key two shift levels of which the first one is VoidSymbol and the second/shifted one is the ISO_Level5_Shift. I don't see how that'd work?

The level5 file is still a mystery to me. It contains the "modifier_mapping" and "lock" behavior which are included by the others, and map the modifier to a "real" modifier (the <MDSW> and <HYPR> keys). How exactly this works is still beyond my pay grade I'm afraid.  (p〃д〃q)

Just a quick note on here to say that the above problem (Extend key clears cell) doesn't happen using an Extend implemented with kmonad (mentioned here), which further suggests it's related to ISO_Level5_Shift.