OSF, etc. -- some facts, some rumors
Rich Salz
rsalz at bbn.com
Fri May 27 01:24:47 AEST 1988
Last time I posted something about OSF, I closed with an offer to forward
along any facts or rumors people heard that they didn't want associated
with their name. This article is a summary -- I no longer have the mail I
got.
It was kind of interesting, being net.gossip-columnist. Almost everything
I quote below I heard from at least two different sources, and I tend to
believe it all. I don't advise you to do that, tho: just treat this as
possibly-accurate rumor. There are some facts at the end.
Sorry this is so long; hope folks find it useful and/or entertaining.
Anyhow...
OSF will be not be rewriting all of ATT code because it's too hard to
prove otherwise.
ATT was invited to join, and knew about OSF a few weeks before the
announcement. A possible reason for the one-week delay in the
announcement was that ATT asked for more time. OSF gave up waiting after
the Wall Street Journal article about Sun and ATT came out.
The first time Sun heard about OSF was either when Scott McNealy got a
FedEx envelope on the day of the press conference, or when East Coast Sun
folks saw articles in the Boston Globe. (Some strategic alliance between
ATT and Sun, eh?)
The OSF's "level zero" spec is out. The basis for the kernel is AIX, but
it's not the current one -- i.e., the port done by Interactive -- but is
instead a future release of AIX. I don't know if they're throwing out the
Interactive work or not.
Apparently each company brought something to the OSF, and engineers from
the other companies reviewed code. ("Hey June, at the code review next
week HP will want to see how your login program handles 8-bit characters."
I can hear morale dropping all over the place... :-)
Educational firms can join the OSF for $5k, companies for $25K, and it's
something like $1Meg or $5Meg for a seat on the board, which gets you a
vote. Microsoft will be joining as a regular member.
The idea for OSF came about when someone (Mashey?) tossed off the idea
during a Usenix discussion in a hallway.
There will be between 50 and 75 people on the OSF technical staff.
The press contact for OSF is Deborah Siegel, Cohn & Wolfe, phone number
1-212-951-8300.
A few other things. These are facts.
ATT has said that future versions of Unix will be based on SPARC. What
does this mean? Not necessarily much. It just means that as it comes off
the tape it'll compile and run on a particular machine. Previously the
porting base was a PDP-11, then it was the Vax, then it was a 3B
something. The only thing that will need to be modified by other vendors
are things like device drivers and memory management. On the other hand,
it COULD mean a GREAT deal if as Unix comes off the tape it just happens
to contain device drivers and memory management code for a machine that
Sun sells.
The intent of ABI, the Application Binary Interface, was always to define
one for each different CPU. It basically specifies what an a.out file
looks like, what number each system call should be, and where their
parameters go (which registers, on the stack, etc.) It's hard see this as
a bad thing, except that ATT gets to pick who makes the standard for each
CPU.
DEC lost a $90Million contract because it specified SVID compliance. They
contested, saying that POSIX is the only real vendor-independent standard,
and were overruled when someone declared that SVID could be used as a
standard for government RFP's. I believe this went to the courts, but an
oddball one, like Commerce Court, or such.
Roughly half of the VAXen that DEC sells these days come with Ultrix on
them. (Someone on the net quoted a 15:1 VMS:Ultrix ratio; the only
possible way this can be true is if you count dollar value of the systems
sold.) When you count VAX workstations, more than half of DEC's machines
go out the door with Ultrix.
If you've got any more anonymous facts, send it to me and I'll post
another of these articles...
/rich $alz
NONE OF THIS IS IN ANY WAY RELATED TO THE ORGANIZATION I WORK FOR.
--
Please send comp.sources.unix-related mail to rsalz at uunet.uu.net.
More information about the Comp.unix.questions
mailing list