Path: utzoo!utgpu!attcan!uunet!lll-winken!lll-lcc!ames!pasteur!ucbvax!RICHTER.MIT.EDU!krowitz
From: krowitz@RICHTER.MIT.EDU (David Krowitz)
Newsgroups: comp.sys.apollo
Subject: Re: why no native unix??
Message-ID: <8901061437.AA06662@richter.mit.edu>
Date: 6 Jan 89 14:37:12 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 44

ACL's don't give you that much of a penalty in speed, and they allow
the rest of us a great deal of flexibility that doesn't exist in
BSD4.3/SYS V. Our network consists of a conglomoration of several
loosely related research groups. People in one group frequently
colaborate with people in other groups on certain projects but not
on others. We also run a number of seminars for undergraduate students
who are associated with different professors. If we had to rely soley
on the Unix owner-group-world protections we'd have to have a
completely open network. There are too many inter-group projects
going on all the time in a constantly shifting environment. Since the
SR10 ACL's exist as an *extension* to, not as a replacement for, the
usual Unix protections they cause the devoted Unix purist little, if
any, problem.

Here is my real gripe with Unix purists -- as shown by this discussion ...
anytime anyone (Apollo, DEC, HP ...) makes an extension to improve the
capabilities of their system they are criticized for not being "standard".
It doesn't matter that the extension can be ignored -- they not only
want their programs to run exactly as they do on a VAX 11/780 running
BSD4.3, they want them to *fail* exactly the same way.

We bought Apollo machines precisely because we did not want our programs
to fail the way they do with a Unix system. We did not want to live with
the limitations imposed by a fixed paging area, lack of memory-mapped I/O,
lack of typed files, lack of network-wide login registries, lack of true
transparent file system, and an inflexible file protection scheme. Much
has been learned about operating systems in the 15 years since Unix was
conceived, and very little of it has been incorporated into either
BSD or SYS V up to now. Carnigie Mellon, with their work on MACH, are
the only people who are actively trying to shove a standard, non-vendor
specific, version on Unix forward into the 1990's. If Apollo were to
abandon their kernel in favor of a native BSD or SYS V port (and I don't
see them doing that), we would have to abandon them in favor of Steven
Job's NEXT machine running MACH. Our research can't afford to wait for
BSD / SYS V to catch up to the capabilities of our current software.


 -- David Krowitz

krowitz@richter.mit.edu   (18.83.0.109)
krowitz%richter@eddie.mit.edu
krowitz%richter@athena.mit.edu
krowitz%richter.mit.edu@mitvma.bitnet
(in order of decreasing preference)