[TUHS] Source code abundance?

Joerg Schilling schily at schily.net
Sat Mar 4 20:04:04 AEST 2017


Clem Cole <clemc at ccc.com> wrote:

> I've been given two different interpretations so I'm not sure who to
> believe.  I really would like to hear a lawyer from Oracle (ney Sun) for
> Micro Focus (ney - At&t -> Novell) make a statement.
>
> I believe the issue is that Sun was given something called "complete
> rights", similar to what IBM had( which is how OSF was licensed - from the
> IBM one).   This was interpreted to believe they could anything with it
> with anything >>they<< did.   That is to say, if they hacked on the kernel
> and published there kernel, then the parts that came from AT&T could be
> also.

Aprox. half of the code from Svr4 was written by Sun anyway. AT&T still 
believed that they owneed all of the code ans required Sunto pay royalties.

Sun for this reason started to negotiate a buy out with AT&T. They agreed that 
Sun pays 3 years of royalties at once and then gets the same rights as AT&T.
Shoirt before the contract was signed, AT&T handed over the code to Novell for 
less than what Sun should pay to AT&T.

Sun then continued the negotiation with Novell and finally did the deal that 
included the right to sub-license the code to an arbitrary number of customers.

Thia was around 1994/1995 and it took Sun Lawyers upt to y2000 to decide that 
the right to sub-licence includess the right to make it OpenSource.

In y2000, Sun published aprox. 95%% of the Solaris-ON code but this did not 
include the source ls.c because it contained 3 lines from Microsoft/Xenix.
It also missed the kernel networking code.

> The question is what happens to the code that got from AT&T but did not
> use.  I'm going to be hypothetical here, Larry correct me to the specifics
> please as I never saw Solaris sources, but SVR4 had Streams Networking in
> it.   Let's say the Solaris pulled that out like we did at Stellar with
> SVR3 and put a BBN or BSD style stack back in and never shipped the streams
> code.   The Network stack they did publish would be available, but what
> about the AT&T version?
>
> I have heard different legal folks say it was both still "closed" and
> others say, it was now opened.

AFAIK, there never was a streams based networking code in Svr4 from AT&T.
The "original" code was from Lachman but it was slow so Sun bought new code 
from Mentat inc for Solaris 2.3 (IIRC). Sun then soon hired all important 
people from Mentat and enhanced the code, but missed to buy Mentat itself.

As a result, Sun could not OpenSource the Mentat networking code. Fortunately, 
this code was not fast enough for 1/10 GB ETH and needed a major rewrite 
anyway. So Sun management did give the OK for the rewrite that finally made it 
possible to Opensource 99% of Solaris-on in June 2005.

Let me finally give some information about SCO....

Sun bought a license from SCO for two reasons:

-	The permission to use the NIC drivers from SCO.
	Then it tourned out that that SCO did use something similar but
	incompatible to the Solaris GLC NIC abstraction layer.

-	The permission to use "lxrun", but then a group of people inside
	Sun wrote a new subsystem from scratch and caused the other group
	to become the looser of an internal dispute.

A friend is one of the people who did work for the SCO kernel group before SCO 
was bought by Caldera Linux and later renamed to SCO... The lawsuit was 
initiated by the Caldera Linux people and not by the "former" SCO people.

He is still working for the company that now owns sco.com and that now 
redirects to the new company name xinuos.com. They still have the ELF standard 
documents and AT&T UNIX documentation online. Their business is now FreeBSD 
based as they did not have the momentum to make UnixWare a 64 bit system. While 
I implemented Joliet and ISO-9660:1999 support for them, I noticed that the 
code was not very 64 bit clean in general....


Finally something about the way the code way sold:

Novell bought not only the source code and the license but the building and the 
USL people in New Jersey as well. When they later sold UNIX to SCO, they  of
course sold the USL location and people as well. IIRC, SCO did even present 
a contract for that fact but the judge did rather believe the oral claims from 
the Novell people.

While I was implementing Joliet and ISOI-9660:1999 I had to discusss things 
with the former USL people from New Jersey.

Jörg

-- 
 EMail:joerg at schily.net                  (home) Jörg Schilling D-13353 Berlin
       joerg.schilling at fokus.fraunhofer.de (work) Blog: http://schily.blogspot.com/
 URL:  http://cdrecord.org/private/ http://sourceforge.net/projects/schilytools/files/


More information about the TUHS mailing list