Re: [Linuxconsole-dev] Re: Localization of Linux

[Jim P. <ji...@ag...>]

James Simmons wrote:
> > What is his objection?
> 
> Bloat. He doesn't even like the idea of posix termios in the kernel. 
> He feels that userland should define any other type of console.

That is what I was thinking, but thinking about what Eric said convinced me
otherwise.  There are definite advantages to having all the terminal
emulations hidden behind the /dev/ttyN that we're used to using.

If the solution *has* to exist in user-land, then kernel changes will be
needed to redirect tty input/output on a per-tty basis from the back end of
the /dev/tty device (after stty stuff) to a selected user-space console
server, which will then in turn plug into the input event stream, and some
kind of display (fbdev or GGI?).

Maybe have some kind of /dev/??? device that the console server connects to,
which allows it to feed input bytes to the /dev/tty* ttys it's been assigned
to, and receive output bytes sent to those ttys.  Perhaps a stream of:

	{ tty-number length-in-bytes data-bytes } x ???

The kernel code would act as a kind of multiplexor, gathering/distributing
data between the list of current ttys and current console servers through
this interface.  The standard `linux' console-code would remain as a default
and fall-back within the kernel.

I can see advantages of having a console server outside the kernel, in that
it can use interfaces that are not designed to be used from within the
kernel, if it wishes, such as GGI.  Will fbdev acceleration also require
cooperation from user-space libraries ?  In which case the same applies
here.

Jim

-- 
 Jim Peters         /             __   |  \              Aguazul
                   /   /| /| )| /| / )||   \
 jim@aguazul.      \  (_|(_|(_|(_| )(_|I   /        www.aguazul.
  demon.co.uk       \    ._)     _/       /          demon.co.uk