Subject: Re: Research questions WRT SCO's complaint
From: DV Henkel-Wallace <>
Date: Sat, 8 Mar 2003 15:45:55 -0800

On Saturday, Mar 8, 2003, at 15:06 America/Los_Angeles, Karsten M. Self 
> on Sat, Mar 08, 2003 at 02:14:14PM -0800, DV Henkel-Wallace 
> ( wrote:
>> On Saturday, Mar 8, 2003, at 13:26 America/Los_Angeles, Karsten M. 
>> Self
>> wrote:
>>>  - Early GNU/Linux development -- I know Maddog (then with DEC) got
>>>    involved in 1991/92.  Any other Unix vendor types on the bandwagon
>>>    early?  Cygnus also had ties to Sun IIRC.
>> I don't know what you mean by "ties": John was the earliest Sun
>> customer and our first machine was one of the very first production
>> sparcstations.  Michael was consulting for Sun until the company got
>> going.
> I was aware of some of the Cygnus history, in part from recent
> discussion on this list.  By ties -- close associations with Sun
> personel, contracts, HW contributions, etc.
> What I'm trying to refute (from several directions) is SCO's claim 
> that:
>     (Paragraph 82)
>     Virtually none of these software developers and hobbyists had 
> access
>     to enterprise-scale equipment and testing facilities for GNU/Linux
>     development.
> Direction 1 is clearly to demonstrate that these "hobbyists" had 
> access to
> enterprise-scale equipment and testing facilities (often masquerading
> as, at your option, Univresity labs, and the whole fscking Internet).
> Direction 2 is to point out that many of the better bits of Unix came
> from a markedly similar environment:  LSD, er, UCD, in the 1970s.

I don't know how much "unixness" (fork, suid, no revisons in the FS, 
etc) matters; we didn't do this sort of development at Cygnus.

We had access to Cray and SGI hardware from the get go with our first 
customer, NASA.  Michael developed g++ on Sun workstations at MCC, 
Stanford and later Sun (remember, these were 68K sun workstations).  
However, despite these being "enterprise class" as the term was 
probably known at the time, I don't think that helps much.

As far as testing goes we had lots of gear at various times over the 
years: Suns of all sorts, HP, IBM, SGI...Sconix was one of the few we 
rarely encountered.

The Hurd, and most of the initial other GNU development done at MIT was 
done on a TI nu machine.  That was an enterprise workstation too.

Better to focus on the Vax: that was an "enterprise" machine.  At MIT, 
because of how we worked, there was only one vax in the AI lab, and 
nobody used it (even it's name, prep, was an obscure joke).  It was 
only a 750, but that might be good enough for your purposes.  RMS used 
it because nobody else wanted it (the rest of us used "real machines" 
like PDP-10s and lispms.  There were other vaxes at MIT of course).  At 
UCB the vax was the machine of choice and there were lots of them, 
including the biggies.  I first installed a BSD distribution on a Vax 
11/780 in 1982 in Paris.

>>>  - Unix marketshare.  Does anyone have values for breakdown of
>>>    marketshare through the 1990s of the major Unix flavors?  My
>>>    understanding is that SCO was never much more than a bit player in
>>>    the x86 market.  Possibly a fair number of deployments, but low
>>>    overall valuation.
>> Any of the usuals (Gartner, etc) can give you these numbers as well as
>> the magazines (Unix Toady etc) since they were eagerly followed by the
>> kinds of people who cared about these sort of things.  A corporate
>> library may have a collection of the trade rags.
> Little access to any of these ATM.  I've been able to dig some
> rudimentary stats via Google searches.  Anyone who can assist, again,
> this is appreciated.

Remember that this all predates the commercial use of the Internet, so 
you have to go to dead trees.