[RFC] Preliminary XI 2 feature list

[Jim Gettys]

On Mon, 2008-09-15 at 21:34 +0300, Daniel Stone wrote:
> On Mon, Sep 15, 2008 at 12:41:29PM -0400, Jim Gettys wrote:
> > On Mon, 2008-09-15 at 19:33 +0300, Daniel Stone wrote:
> > > As covered with Peter, I think making the distinction between users at
> > > the input protocol level is unnecessary and pointless, and will only
> > > cost us in the long run.
> > 
> > We'll likely need some way to tie back to some sort of authentication
> > context, for security.  Viewing a 16 bit value as a single item with
> > structure may be all we need for that though...
> 
> Of course, and this isn't something we can mandate sufficiently
> flexibly in the protocol.  We already have properties, so any security
> framework could just expose read-only 'user n' properties, or whatever,
> and we don't have to worry about details like this in the protocol.

It it is to fit with Eamon's work, some minimal hook is needed.  Not
much, but some place to stand....  As little as possible, as much as
necessary...  I agree that as little as possible should be hardwired, as
the old XSecurity extension failed on "one size fits all, and it better
be MLS...." disease.
                                  - Jim


> 
> Cheers,
> Daniel
-- 
Jim Gettys <jg at laptop.org>
One Laptop Per Child