New manufacturing data flags for keyboards (2nd draft).

[Bernardo Innocenti]

Bert Freudenberg wrote:

> I disagree. It's a major hassle if the keys you press do not do  
> what's printed on them. Simply assuming a US layout when the machine  
> can know quite well what keyboard is attached is unacceptable. It's a  
> shortcoming of external keyboards that they do not communicate their  
> physical layout to the host (*), but for the built-in keyboard in a  
> laptop we should not have to live with this.

We're talking about the recovery consoles and firmware prompt...

In our case, it's clearly not a matter of detecting the correct
keyboard, but rather including a potentially large amount of
different keymaps in low-level components of the system such as OFW
and kernel.

These components shouldn't need updating every time a new keyboard
come out.


> (*) the OS should let you choose the layout for every keyboard newly  
> attached, and then remember that setting. For some reason that I  
> don't understand, no OS allows to use a German and English keyboard  
> in parallel. They all can only have one keymap active at a time.

An unfortunate leftover from when there was a 1-1 relation between
keyboards and terminals or consoles.

I think X could now do the right thing now with evdev: each driver
instance gets its xkb settings in xorg.conf.  Oddly, there's no
way to specify the device from setxkbmap, so we'll once again loose
this feature when xorg.conf gets phased out.

-- 
 \___/
 |___|   Bernardo Innocenti - http://www.codewiz.org/
  \___\  One Laptop Per Child - http://www.laptop.org/