On Tue, Apr 07, 2020 at 10:38:16PM +0200, Ludovic Courtès wrote:
> I think the approach is OK, even as-is.  We can then improve it by
> (hopefully) replacing the hard-coded list of non-Latin layouts (ar, jp,
> etc.) with code that determines whether the main layout is Latin or not.
>

I am still doing final testing, but I believe I can push the attached
patch.


> > +    (match first-layout
> > +      (("ar" "azerty") (list "ar,fr" "azerty," "grp:alt_shift_toggle"))
>                                                ^
> Remove comma?

The comma is needed, for example the Japanese Kana ("jp" "kana")
layout breaks otherwise.  I believe the reason is that the "kana"
variant gets applied to "us" layout.

Regards,
Florian