[LWN Logo]

Date: Thu, 02 Apr 1998 17:55:15 -0500
From: Shaleh <shaleh@livenet.net>
To: "debian-private@lists.debian.org" <debian-devel@lists.debian.org>
Subject: X statement

This was posted to another mailing list I am on.

See http://www.camb.opengroup.org/tech/desktop/onlineorder/xdownload.htm
for The Open Group's *free* download page and the license.  It is no
more onerous, IMO, than Netscape's or Qt's licenses for source code.

IMO... background:

The X Window System has historically been funded by the system vendors,
such as Sun, HP, DEC, IBM, etc, a few companies interested in graphical
technologies, such as Adobe, Kodak, etc, and finally by small amounts
from end user companies and small companies such as ourselves.  There
was a noticeable decay in funding the X Consortium from the system
vendors, after COSE/CDE and at the point where Microsoft was perceived
to have owned the mass market desktop.

Core technologies for graphical interoperability have been deployed. 
The X Window System protocols have been well defined and well
respected.  This was aided by test suites and by virtue of not having
changed the fundamental protocols for a long time.  That, however, was a
combined decision, because the major funders did not want to have to
help their customer base change graphical applications, and because
there was a feeling that the core protocol had been sufficiently
developed.

The changes at the Server level were so small that there was almost
nothing extra in the protocol (extensions, perhaps, but not in the core)
between X11R4 and X11R6.1.  There were some big changes in X11R6.2 and
6.3, so that some extensions could work (security, for example), but the
core protocols remained just about identical and certainly
interoperable.  Between X11R4 and X11R6.3, the major changes have been
in the libraries and some core and contributed clients.  For example,
completely new libraries and mechanisms have been developed, such as
Session Management and Interprocess Communication, but these have barely
begun to be seriously used until very recently.

For about the last six years, XFree86 has been a member of the X
Consortium, at the $5,000 subscription level.  The funds have been paid
by other members of the X Consortium.  This is a minute fraction of the
funds needed for continuing development of the test suites, and for
development and enhancements.  Several million dollars are needed for
that effort.  Now that X is part of The Open Group, it competes for
funding with DCE, "internet dialtone", Java and other activities, many
of which the system vendors (still the main funders) see as more
important than improving X, because they've yielded the desktop fight to
Microsoft.

The cash haemorrhage is made worse by the non-fee paying XFree86. 
Although it may be able to claim several million users, the net
contribution to the development of the X Window System, as embodied by
specs, protocols or cash is a flat zero.  No contributed specs, no
contributed standards and no cash... but a lot of individuals and
companies undoubtedly benefit.  The reason for the system vendors to
fund X was so that each of them increased their own chances of being
involved in a networked graphical environment, and the source code was
made freely available as a way to foster the interoperability and
because the project came from an academic background, which requires the
free flow of information.  However, the model breaks down if one member
has a lot of customers and makes no contribution.  At that point, the
other members are funding the freeloader.

In its' way, the fees are similar to the "Linux tithe", where commercial
organisations benefitting from Linux, are supposed to provide cash for
Linux development activities.  The Open Group is similarly a non-profit
organisation.  As with Linux, anyone is free to contribute, and to
generate profitable products (the GPL does not restrict efforts to make
a profit, contrary to the belief of many users).

Research use (non-commercial use) of X11R6.4 is also free.  The folk
affected are us, MetroLink, RedHat, Caldera, SlackWare, Walnut Creek
CD-ROM, Delix, SuSE, and so on.  XFree86 should remain unaffected. 
Enlightenment will be unaffected.  KDE will be unaffected.  GNOME will
be unaffected.

Under the terms of the license, nothing prevents XFree86 from continuing
to port X11R6.4 and to add Servers.  Fee-based distributions will need
to license X11R6.4 or buy already licensed binaries.  Pretty simple. 
Should be non-threatening.  Those who paid for their Linux before, will
now include a proportion for the X fees.  Those who didn't, probably
won't have to pay anything now, either.

IMO... prediction:

The X protocols in X11R6.3 are interoperable with those of X11R4. 
Extensions have been implemented in such a way that clients and servers
can negotiate about the new features.  I expect that if there is a move
to stay with X11R6.3, it should have no impact, whatsoever, other than
restricting access to whatever X11R6.4 features are offered by X libs or
X Servers.  

Many of the features of X11R6.2 and 6.3 are still not well implemented
by either us, or XFree86.  For example, the XInputExtension was defined
in X11R6, and is still unrefined (though improving in the most recent
releases).  An hiatus to catch up to the specs in X11R6.3 should be
welcome, I'd think ;-)

BUT... There is a large and vocal fraction of the Linux community who
equate open access to source and zero cost pricing.  Had their been a
significant contribution to X Window System technologies, I'd have more
sympathy with this view.  The measurable, practical effect of XFree86 on
X is zero... I'm not referring to the *uses* of X, but to the technology
basis.  Anyway, this vocal fraction is likely to be incensed by any
charges for open access source code.  I'm expecting a flurry of incensed
news postings and email, with calls for a new standard to be developed. 
Given the comprehensive failure of XFree86 to do anything at all (in
developing new technology, not in implementing Servers and some trivial
support programs using the technology) in the last seven years, I'm not
seriously concerned that enough effort could be devoted to produce
anything significantly different from X to be a cause for concern.  To
be as blunt as possible, XFree86 can't even conform to the current specs
with the assistance of the X Consortium, so I can't see how they could
possibly generate anything new.

IMO, it is unlikely that there will be a new revision of X based on
"free" development.  It is very likely that the core technologies will
continue unchanged for as many years in the future as they have in the
past.  This will give a clear seven to ten year period when everything
based on X11R6.3 will continue to work with new revs of X specs.  This
is infinite in computer software terms.  No one can sensibly predict
software for more than about two years, and even then they might be
confounded by something new (e.g. who in 1992 was seriously predicting
that the Web would emerge as a significant force in 1994?  Most people
would probably have picked gopher...)

Since the impact is on RedHat, Caldera, and other profitable
enterprises, I think it is as fair as the Linux tithe that these same
organisations should be paying already.

So, my inclination is to remain completely calm.  If I was interested in
the future of XFree86, I'd be concerned that they fully and correctly
implement *any* set of X specs, before I'd be worried about the impact
of introducing charges for the commercial distribution of X11R6.4 on the
development of window managers.

What might be more worryng for XFree86 is that the X Consortium might
now have a case for forcing them to withdraw the usage of "X Window
System" from their works, until such point as it is conformant with the
freely available specs and standards.

As for us... we've been looking at X11R6.4 for ages.  We've followed the
development of the specs.  We've our own opinions as to which of them
*we* will implement, and when.  As I've made clear in the past, we pick
and choose amongst the specs and sample implementations to produce
something that we feel is robust, and where we think customers have an
interest.  We're not planning on, and could not (without voiding the
whole intent of interoperability, a fundamental idea in X), break
interoperability with X11R6.3 or X11R6.4.  Indeed, the X11R4 based X
Test Suite still executes with perfect safety on our Servers :-)

The change should be less significant than, say, the introduction of
glibc2 in RedHat ;-)

Cheers, JeremyC.
-- 
Jeremy Chatfield, Xi Graphics  mailto:jdc@xig.com  tel:+44(0)1234.710030 
-- 
-----------------------------------------------
How can you see, when your mind is not open?
How can you think, when your eyes are closed?
- Jason Bonham Band, "Ordinary Black and White"
-----------------------------------------------


--
To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org