Keyboard_layout: deprecation of xkbmap caused regression with layout/variant

[cortices]

Hello, my previously working setup with xkbmap driver for keyboard_layout block stopped working in the last few months with an update and I finally got around to investigating.

It seems that github.com//pull/2133 must be responsible.

I switch between a layout with no variant ("us (N/A)") and one with the same layout but different variant set ("us (workman)").

I can successfully switch keyboard layout with setxxkbmap us and setxkbmap us workman just fine, but the i3status-rust block doesn't show that change, including when I force i3 to restart.

i commented out the substitutions section completely to test and what I get is the block always showing "us workman" for the format "$layout $variant".

I'm wondering if something perhaps isn't registering the variant being unset, or set to null, or empty string etc...

Running 0.34.0 from nixpkgs.

Thanks