To: rms@prep.ai.mit.edu
Cc: don@brillig.umd.edu
In-reply-to: rms@prep.ai.mit.edu's message of Mon, 15 Dec 86 20:40:37 EST
Subject: So whadda ya thinka NeWS?
--text follows this line--
   Date: Mon, 15 Dec 86 20:40:37 EST
   From: rms@prep.ai.mit.edu (Richard M. Stallman)

   I don't know whether the NeWS protocol is technically better than X,
   but if we don't limit ourselves artificially to technical advantages
   alone, X is better than NeWS because X is free and NeWS is not.  I
   urge people who support the GNU system to do everything they can to
   encourage the use of X rather than NeWS--unless they would rather
   write a free implementation of NeWS.

Please take the time to read this, as its length is due to the fact
that I am trying to be helpful, rather than mindlessly flaming at you. 

From what I know of X, PostScript, and NeWS, the technical advantage
that NeWS has got over X is a very real one that is profoundly
important and useful. I'm sure you fully realize the importance of
extensibility. This advantage alone makes a NeWS server a much more
useful and powerful machine than an X server. The fact that you can
load a library into a NeWS server than emulates an X server, without
even having to modify the X clients, demonstrates this point.
PostScript is a beautiful, powerful, complete language; It's not just
another set of drawing commands, as the term "page description
language" seems to imply to some people.

On top of that, the whole thing is incredibly portable. I have seen it
running on an Atari ST! It is not dependent upon any operating system,
like X is. It will eventually be running on many different machines,
low and high end, from the Mac to the Iris. On the outside, the user
interfaces of other window systems can be emulated. On the other
outside, the programmer interfaces to window systems and graphics
packages can be emulated, so that programs written for other systems
can be relinked to use NeWS. On the inside, modules of each interface
can be modified or replaced by the user. Think about what it would
mean to software developers for a system like this to become widely
used. 

A free implementation of NeWS would be quite a lot of work, but
absolutely worthwhile, and very important. The idea of writing a
PostScript engine on top of X is upside-down thinking, though. In
expressing all of your imaging operations in terms of the X protocol,
you could not take advantage of hardware that performs operations
other than those that X describes. NeWS is designed to take advantage
of special hardware, which speeds things up considerably, without
requiring the PostScript code to have any knowledge of what's going on
behind the scenes. It's wrong that high level PostScript graphics
operations that the hardware can directly support should have to be
broken down into lower level operations that X supports.  And if the
PostScript engine were an X client, instead of being built into the X
server, you can say goodbye to all the bandwidth and context switches
you would otherwise have saved by sending the server postscript
programs, instead of individual commands. It is clear which can do a
more elegant and efficient job at emulating the other.

I see X as a very powerful system, that's got a lot of advantages over
many other window systems. I got disgusted with SunStools some time
ago, and installed X. I've been very happy with it, and have been
maintaining, using, and programming with it ever since. It's much more
flexible, a lot easier to program, and does many things that SunStools
can't approach. But NeWS is even more so. In many ways, NeWS is to X
like Unix is to MS-DOS: it's extensible, portable, multitasking, has
modular components, and allows for a much wider range of user and
programmer interfaces.

X may be free, but DEC certainly has a lot of money invested in it. I
don't know at this point exactly how Sun is going to distribute NeWS,
except that they are going to handle it similar to how they're
handling NFS. They would like for it to become an industry standard.
It will be distributed "free" with version 4.0 of Sun Unix. I'm sure
that in your opinion, this method of distribution is not politically
correct, but I think the ideas upon which NeWS is based are very
sound, and that it represents the cutting edge of window system
technology. It will take some time for many of the wrinkles to be
ironed out, but once people have gotten some idea of what the
capabilities and limitations of this amazing machine are, it will be
time to start thinking about writing a public domain version.

I don't think that discouraging people from using NeWS is a productive
thing to do at all. I am sure there will be a lot of top quality free
software written for NeWS, before there is ever a free version of NeWS
itself. And when a free version finally comes out, all that free
software will work with it just peachy keen. Gnu Emacs takes advantage
of programs written in MockLisp with a compatibility package. But the
NeWS "protocol" specification is on the public domain, and PostScript
is a rich, stable language, that is becoming increasingly popular. A
public domain NeWS server should be completely compatible with the
commercial one. If you had a public domain MS-DOS right now, would you
tell people to write free programs for that, instead of Unix? You
certainly wouldn't want to discouraging people from writing programs
for Gnu, under commercial versions of Unix, instead of waiting for Gnu
to be finished. By the same token, it makes good sense not to
discourage people from supporting NeWS just because a free version
does not exist yet.  Writing a NeWS implementation would be a hell of
a lot of work, but discouraging its support just because that work has
not yet been done, in spite of NeWS's technical superiority, would
only discourage the work from ever getting done.

	-Don