Re: XKB's keymaps for the Hurd console

[Diego Nieto Cid]

Hi,

On Sun, Mar 27, 2011 at 07:28:26PM +0200, Samuel Thibault wrote:
> 
> Since it's now using pc_kbd as a base, I guess key repeater is now
> supported (xkb/README still says it's not)?  Also, README needs to be
> updated to stop saying that you need to replace -d pc_kbd with -d xkb.
> 

Yes. Also the TODO file may need some updates too. I'll check them.

> 
> Do you have assigned your copyrights on Hurd changes to the FSF?  This
> is needed before your changes can be commited upstream.
>

No, I haven't. What should I do?

> 
> The xkb/Compose file is compatible with usual Xorg Compose files, right?
> I guess we don't want to ship it but rather use e.g.
> /usr/share/X11/locale/<thelocale>/Compose
>

Hm, I suspected these files were hidden somewhere. And given they are
locale specific the right thing to do would be to look at the
environment and load the appropiate one. 

Should the command line switch be left for the user to be able to
override this choice?

>
> I also guess xkb-data files could be used instead of shipping
> default.xkb?
> 

The data files are used. default.xkb was loaded when the user didn't set
a keymap file.

Hm, but now that a default file is always set that code is no
longer reachable.

>
> It also seems to me like xkb/xkbdefaults.c could be generated at build
> instead of shipped?
>

Yes. It was generated from default.xkb. Anyway, I'd just remove them.
Having a hardcoded fallback keymap doesn't add much value anymore. The
old pc-kbd code could be used instead.

>
> Is there a reason not to use keysymdef.h installed by the Xorg x11proto
> package rather than copying it?
>

Other than avoiding the package dependency... there's probably none.

>
> Rather than copy/pasting ks_table.h, kstoucs.c and symname.c, I'd rather
> see the Hurd console simply depend on libX11, actually.
> 

In that case I'll need some assistance with autoconf magic to check for
it :)