AT&T/Sun merged UNIX

[Barry Shein]

From: Doug Gwyn  <gwyn at brl-smoke.arpa>
>Of course it would be pretty stupid for the AT&T/Sun merged OS
>project to not track the standards, but they haven't been immune
>from stupidity in the past.

I believe the following comment is beyond idle chatter but I'd
rather not get into quoting names etc.

Basically the standards committees have spent most if not all their
time standardizing Unix as it existed in 1978 with a few new items
thrown in here and there (some of which were very important, I don't
mean that to be disparaging, just that the vast majority of the
efforts are devoted to relatively old stuff.) There's nothing wrong
with this, but there's nothing particularly inspired or useful about
it either.

Very little to none of the effort has been dealing with issues like
networking, remote file systems, windowing etc, aka "modern needs".

This is not a damnation, it is simply a statement of fact that on the
one hand standards committees tend to focus on old, well-trod ground
while something as dynamic as the Unix industry desperately needs
acceptable industry standards in these new areas fast, even if they're
only based on widely accepted de-facto standards (eg. TCP/IP, NFS, X,
NeWS), or they're dead.

I believe some folks at Sun and ATT recognized this fact and decided
to plow ahead with a bold super-set of what the standards committees
were working on, determined to present to the industry these badly
needed standards so we could move on to other things quickly.

Sometimes this works, sometimes it doesn't (Unix itself could be
considered such a bold venture for its time, I'm quite sure any
standards committee of its time would never have approved C as an
implementation language, for example, they only have eyes in the backs
of their heads, as they probably should, and surely would have
considered the issue of a SIL wide open, Bliss, Algol variants, PL/I
etc, as many of you are right now saying to yourselves "but, but, but
whaddabout...", yet to many of us who were around then Unix was
obviously a much needed standard the first day we got it out of the
box and got past Irma Biren's nice cover letter.)

I have no doubt that many are frantically clinging to the concept that
many of these issues such as networking protocols are still wide-open
and shouldn't be standardized on something like TCP/IP. Be that as it
may, but nothing else really exists (eg. you can't say ISO exists,
whether it ever will exist remains an open question) and the rest of
the world needs to get on with things and be able to assume that these
are standards so those "nuisance little things" called applications
can be written. Their adoption by Sun and ATT (not to mention de facto
adoption by dozens of other vendors) ensures that they cannot be way
off the mark, reality is its own excuse. Let's codify them and get on
with the show, I say.

So the issue is not entirely whether or not the ATT/Sun merge tracks
standards as it purports to standardize areas that no current
standards area seems to even be addressing. Obviously this could cause
some incompatibility with standards as they are being written unless
the standards are powerful enough to support these expanded needs. I
suspect the tendency will be to try to do no worse than superset
(there's a fine line between supersets and incompatibility.)

Personally, I consider the ATT/Sun merger a welcome and long overdue
kick in the ass.

	-Barry Shein, Boston University

P.S. Again, this is not anyone's official policy, just the situation
as I understand it.