[PATCH:xf86-input-keyboard 21/21] Add some NetBSD feature detection code.

[Thomas Klausner]

On Sun, Jul 28, 2013 at 04:18:40PM +0200, Thomas Klausner wrote:
> On Sat, Jul 27, 2013 at 01:47:10PM +0200, Mark Kettenis wrote:
> > > From: Thomas Klausner <wiz at NetBSD.org>
> > > Date: Fri, 26 Jul 2013 23:24:16 +0200
> > > 
> > > Signed-off-by: Thomas Klausner <wiz at NetBSD.org>
> > 
> > There should be no need to add -DPCVT_SUPPORT and -DWSCONS_SUPPORT.
> > Those defines should come from <xorg-server.h>.
> 
> Oh, good point. I've reduced the diff.
> 
> > Can you explain why the "standard" protocol doesn't work on
> > powerpc/sparc/sparc64/vax in NetBSD?  It works fine on OpenBSD/macppc
> > and OpenBSD/sparc64 for me.
> 
> It's also for evbarm, sgimips, shark.
> 
> I think the idea on NetBSD is to DEFAULT_TO_WSKBD everywhere, however,
> it's work in progress. There is a comment in the Makefile in xsrc:
> 
> 1.12         (macallan 28-Apr-11): # turns out we can't use wskbd everywhere without a couple more translation
> 1.12         (macallan 28-Apr-11): # tables in the X driver so make it the default only where we know it will work
> 
> I don't know what the issue with "standard" is, sorry. I'll ask on
> tech-x11 at NetBSD to find out.

Ok, I asked, and how I understand the answer, it is:

OpenBSD MD kernel keyboard drivers provide "rawmap"s which return PC
XT scancodes for X servers.

NetBSD's MD kernel keyboard drivers only return "native" keycodes, and
src/bsd_KbdMap.c has many MD maps that translate from the "native" map
to PC XT code -- which is why wskbd must be used where it is
supported, and manual xmodmaps where not (some less-used ports don't
have maps yet).
 Thomas