From: James Simmons <jsimmons@edgeglobal.com>
  To  : ggi-develop@eskimo.com
  Date: Thu, 15 Apr 1999 16:12:10 -0400 (EDT)

Re: current state of kernel input work?

On Thu, 15 Apr 1999, Steffen Seeger wrote:

> > Thats what I have been doing very recently. In the process of seperating
> > /dev/fb from fbcon. I soon realized that the console code was to
> > interwoven with VT switching.
> 
> That's why I rewrote it from scratch. So I am not the only one
> to realize that. :-) Basically, the notion of "head" is equivalent 
> to KII's notion of a focus. 

So thats what a focus is. I have looked at your code. Sometimes what you
call your structures give no clue to what they are for exactly. 

> However, we need a 'workplace' id to be
> assigend to processes (just as user and group id's). KII currently
> uses the controlling tty (or that of the parent, if a process has none)
> to determine the workplace a process belongs to. Not perfect, but working.
> 

Thats the way I doing it. Since I use /dev/ttyX to represent a head. 

> This will always be a pain if you refuse to accept that a _central_ resource
> management instance (like KGI/KII) is neccessary. Yes, you can avoid it,
> but it will be far easier if you do accept it. You can believe me that
> I took a very long thought before deciding to throw the kernel console
> system away and replace it with code written from scratch.

Actually I came up with a idea. I tried to avoid thinking of a head and a
virtual head as being different. So I think of the true heads as the
active ones. The virtual heads are the inactive heads. When you do avt
switch the current VT on that head is no long consider active. It switches
to the other VT on that head. 
 
> Take a look at the KGI-0.9 code. All that's missing to have graphical
> multihead is to enable the dev_graphic_init() to have /dev/graphics 
> (or /dev/fb, /dev/gfx) and to write a driver for your card.

Trust me. I have been studing your code.