Subject: Re: Why not track our xsrc with X11R6.6 from
To: NetBSD-current Discussion List <current-users@NetBSD.ORG>
From: Andrew van der Stock <>
List: current-users
Date: 07/14/2001 19:20:49
XFree86 tracks releases relatively quickly. I would not be surprised
to see X11R6.6 being pulled in time for 4.2, due roughly November.

Anything in the xc/programs/Xserver/Xfree86 directory is Xfree86's.
Everything outside of there is not; I've noticed in the last five years of
watching the xfree86 devel list, that there is a great reluctance to change
a single line of code outside there. And I feel that's a good thing.
Generally, when bugs were spotted and resolved, the patch went back to
rather than staying within the xfree86 tree.

I don't recall who re-wrote or changed the Xinerama extension, but the
reason was probably to support the multi-head stuff in XFree86 4.0 and
later. A quick perusal of the changelog would probably dig this out. I don't
think it's important, as realistically, the XFree86 version is likely to be
less buggy and have more features.

XFree86 as PC Linux weenie platform vs NetBSD as "workstation" snobbery

The two major early lights of X, Jim Gettys and Keith Packard are active and
frequent XFree86 contributors. Keith Packard in particular is very active
with his role with SuSE Linux, doing stuff like the first major revision of
how text is rendered under X (from incredibly badly to more than just
passable) in over fifteen years. Many other brilliant and active developers,
such as the hardworking Matthieu Herrb, Brian Paul and Renderman (well,
possibly a bad example, but his intentions are for good demos :-) work on
the XFree86 source base, making X more useful for games and general use.

XFree86 was the only active X development for many years. It's the most
widely deployed version of X. The only reason that XFree86 became
Linux-centric is because NetBSD for the longest time had its own
(unnecessary and hindering) lagging version of XFree86. With no engagement,
no or little bug, no backporting, and seemingly no feedback, it's like we
didn't exist. And in true open source style, NetBSD appeared pushed to the
side because we appeared to not contribute. If we had spent the time since
late 1997 porting all our unique servers into the new architecture, like
FreeBSD, we would have DRI today.

I am glad that the guys looking after current's X11 are moving to closer
tracking of the real X source tree. I would like ot see it closer still.
There's nothing to be gained by holding out a seperate branch, and much to
be lost, particularly in volunteer organisations like this, where time is at
a premium.

Realistically, I'd like it if people stopped getting on their hobby horses
about XFree86 being PC-centric. If someone implemented the Xsun server as a
module under XFree86, you'd not only get things like anti-aliased text for
free, you'd use less memory, and your Xserver would be faster due to the
modern server architecture. The only thing that XFree86 can be accused of is
that it does seem to assume PCI or PCI-like (ie AGP) bus space if
implementors don't override the init stuff. As 99%+ of all post-1995 video
cards are indeed PCI or AGP based (by manufacturer and model; if by
marketshare it is closer to ~100%), this is not an unreasonable default
position to take.

Lastly, cards that have not been actively supported in XFree86's devel list
such as the grand plus ancien Paradise card were dropped for a reason - no
one came forward to look after them. Similarly, Amoeba, Minix and non-ANSI C
compiler support will soon be dropped. It's estimated that the Minix port
hasn't been built in seven years. Dropping old and barely used features are
good for combating bloat and improving reliability and readability of code.

If you like or use your card, build a recent XFree86 tree and see if it
still works. If it does - great; if not, patches welcome at

4.2 is due around November; plenty of time.


----- Original Message -----
From: "Greg A. Woods" <>
To: "NetBSD-current Discussion List" <current-users@NetBSD.ORG>
Sent: Saturday, July 14, 2001 12:25 PM
Subject: Re: Why not track our xsrc with X11R6.6 from

> I think it kinda goes the other way around.  X11R6.6 is from's
> ongoing X Window project, and XFree picks up the stuff and makes
> it more pc-like...   [0.5 :-)]
> It seems XFree-4.1.0 is based on X11R6.5.1, but I'm not sure just how
> much they integrated and what's unique to XFree -- their release notes
> aren't exactly clear on these issues.
> (all I see is that they re-wrote Xinerama)