Multiseat Support in HAL

[David Zeuthen]

On Fri, 2006-09-22 at 18:20 -0700, Artem Kachitchkine wrote:
> > Quick question. Are these exported in the Solaris HAL bits, and if they
> > are, will the new "edit partition table" code we're adding [1] going to
> > work? Because that would be awesome, I mean, for SunRay users.
> 
> Absolutely. From the application standpoint disks connected to Sun Rays are real 
> disks, only not located under /dev (each terminal has its own device root).

Sounds good.

> Two things to keep in mind...
> 
> One, we might choose to use partitioning implementation other than libparted on 
> Solaris, compatible with the original at the D-Bus level.

Sounds OK to me; libparted has, IMHO, quite a few short-comings, if the
code you have to do this is available under a suitable license, I'd be
happy to use this on Linux too (I already wrote my own disk label
scanning code as it wasn't a good idea to use libparted for this).

I'll make sure to send the D-Bus interfaces for this to the list for
review before committing them.

> Two, there will be a separate backend for Sun Ray, because Sun Ray software is a 
> separate product from Solaris and not (yet) opensourced. We expect that the 
> backend will ship as part of the Sun Ray software bundle, and dlopen'ed by the 
> Solaris backend.

This sounds OK to me, the core hald code is dual AFL 2.1 / GPL so it's
indeed possible and something I've wanted to be possible from day 1. 

(for other readers, note that some of the Linux hald code is GPL only
and thus forces distributors of this code (e.g. Linux distributions) to
give back all their changes to hald. A decision that is quite
delibirate.)

    David