Re: Desktop Kernel Stuff

[<Jim Gettys hp com>]

> Sender: gnome-private-members-admin gnome org
> From: Alan Cox <alan redhat com>
> Date: Tue, 15 Jul 2003 10:32:52 -0400 (EDT)
> To: Jim Gettys hp com
> Cc: danw ximian com (Dan Winship), sander vesik sun com (Sander Vesik),
>        Murray Cumming Comneon com, jdub perkypants org,
>        gnome-hackers gnome org
> Subject: Re: Desktop Kernel Stuff
> -----
> > Commerical UNIX's now have to worry about being compatible with Linux,
> > rather than the other way around.  And most facilities we would want
> > are usefull for other things, in any case.
> >
> > So I no longer believe that getting kernel changes for the desktop
> > is a futile exercise anymore, the way it was a few years ago.
>
> It isnt the kernel changes that are the big problem. Linux and *BSD will
> adopt sane changes and even work together on them sometimes. I'm sure
> the Sun guys are taking notes too.
>
> The problem is legacy infrastructure - NFS has no EA's for example or
> file notifications

Yes, this is the crux of the real issue.  As Xft1 vs. Xft2 shows, if you
can code something to be able to work at all on old network infrastructure,
it's uptake into the community can be very much faster than if you can't
get the new facilities at all.

A reality of life is that some of what we do will have to have provisions
to deal with legacy infrastructure; but in the long run, we must drive
the base systems forward or we'll be toast.

Not getting some of the new features some of the time is better than
saying we have to live with the constraints of the past forever.

                               - JIM

--
Jim Gettys
Cambridge Research Laboratory
HP Labs, Hewlett-Packard Company
Jim Gettys hp com