wsr@lmi-angel.UUCP (Wolfgang Rupprecht) (03/16/87)
From 2/19/87 "Electronics" p.17:
"Carnegie Mellon University [...] has reached an important milesone in
its three-year effort to develop Mach, an operating-system support
environment that will run Unix applications on a variety of
multiprocessing computers. CMU is releasing the first version of Mach
into the *public domain* after completing tests on a network equipped
with DEC Vax multiprocessors and more than 100 workstations from Sun
Microsystems and IBM Corp." [emphasis mine -wsr].
Does anyone know if this stuff about Public Domain is correct?
--
Wolfgang Rupprecht {harvard|decvax!cca|mit-eddie}!lmi-angel!wsrsteve@mimsy.UUCP (Steve D. Miller) (03/18/87)
Well... I haven't heard anything specific, but this is what I know from
looking at the Mach sources we have: the Mach kernel proper (nifty IPC,
nice virtual memory, tasks/threads, and that's it) is apparently written
from scratch, and, if so, is something that the Mach people *could* put in
the public domain if they so choose. However, the comment about Unix
compatibility is a little misleading. Sure, Mach is 4.3BSD binary
compatible, but it's not like they rewrote all of 4.3 to make it so. They
basically sawed off the bottom layers of the Berkeley kernel (i.e., anything
that touches the virtual memory stuff is changed), but left the upper layers
intact. And *that* they can't just give away. We had to show source
licenses for SunOS and 4.3BSD before we could get a copy of the Mach kernel
plus its Unix-lookalike wrappers for both Suns and vaxen.
As an interesting aside, the Sun version of Mach is mostly 4.3-based; for
example, the Mach group didn't put NFS back in (they have their own), and I
don't think they put pixrects back in (I do know that the last time I heard,
SunView didn't work and they were telling people to use X). They did grab a
lot of device drivers, though, and probably some other stuff.
-Steve
--
Spoken: Steve Miller ARPA: steve@mimsy.umd.edu Phone: +1-301-454-4251
CSNet: steve@mimsy.umd.edu UUCP: {seismo,allegra}!mimsy!steve
USPS: Computer Science Dept., University of Maryland, College Park, MD 20742dld@theory.cs.cmu.edu (David Detlefs) (03/18/87)
Mach is not in the public domain, but is distributed for free. They maintain a copyright, and control over redistribution. In particular, no one is allowed to modify and sell it.
Karl.Kleinpaste@cbstr1.att.com (03/19/87)
Posting-Front-End: GNU Emacs 18.37.1 of Wed Feb 25 1987 on cbstr1 (usg-unix-v) A friend at CMU sent me a copy of the copyright notice attached to all MACH source files. (I'd include it here, but I don't know what that would mean legally.) It says quite clearly that [a] MACH comes from CMU, [b] it has certain identified authors, [c] it is derived from other sources, and [d] appropriate licenses are required both to obtain it and to redistribute it. No, MACH will not go public domain. It can't.
steve@mimsy.UUCP (Steve D. Miller) (03/20/87)
The copyright notice I looked at (on a Mach kernel file) states
that some of the code distributed with Mach is derived from other
code, and that any previous copyright notices are preserved below
the CMU copyright. In the case of an obviously 4.3-derived (for
example) module like netinet/ip_input.c, it's clear that CMU doesn't
own it. In the case of their Mach-specific kernel stuff, though,
I think (but don't know for sure, I admit) that it's theirs to do with
as they will. That doesn't mean that it will go into the public
domain, become freely redistributable, or whatever...
-Steve
--
Spoken: Steve Miller ARPA: steve@mimsy.umd.edu Phone: +1-301-454-4251
CSNet: steve@mimsy.umd.edu UUCP: {seismo,allegra}!mimsy!steve
USPS: Computer Science Dept., University of Maryland, College Park, MD 20742