home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.next.advocacy
- Path: sparky!uunet!microsoft!hexnut!edwardj
- From: edwardj@microsoft.com (Edward Jung)
- Subject: Re: End of NeXT
- Message-ID: <1992Dec30.090355.25626@microsoft.com>
- Date: 30 Dec 92 09:03:55 GMT
- Organization: Strategy Division, Microsoft Corp.
- References: <1992Dec26.183705.4765@cuug.ab.ca> <1992Dec27.145119.7923@wam.umd.edu> <1992Dec28.025057.6783@news.acns.nwu.edu>
- Lines: 74
-
- [The holiday season gives me time to peruse these great advocacy
- discussions... entertainment for all!]
-
- Ah, we are enamoured of the term "microkernel". Indeed there was a
- great panel session at the last couple of workshops on operating
- system technologies (SOSP and WOOOS) on defining exactly what is
- a microkernel OS (and, for that matter, what is an object-oriented
- one)! Because it has become some kind of "Holy Grail-like" blessed
- techno-marketing term, it is suddenly important to categorize.
-
- Unfortunately, you won't find an unequivocable answer. Sigh. There
- are those who will say that NT is not a microkernel. Cutler said
- this at times because it was so tiring justifying the term, and
- there were many religious zealots of one particular persuasion
- at the microkernal os workshop (where he said this and showed
- supporting slides). Even within Microsoft this is a debatable topic.
-
- Of course, until Mach 3.0, Mach isn't a microkernel implementation
- either, although you could debate that it was a "microkernel-
- inspired architecture".
-
- Anyhow, if you are really interested in getting informed, Custer's
- book on NT is a reasonably good one that outlines the basic NT
- structure. It is heavily client-server, and has an architecture
- that separates core services (in the NT Executive) from user-level
- subsystems (much like Mach). It has a kernel within the Executive
- that manages kernel objects. Read the book -- it's reasonable. The
- NT architecture is different from Mach. In some ways there were
- compromises made (as always in shipping software). Rick Rashid,
- the head of the Mach project at CMU now heading the research
- group at Microsoft, admits that in some places NT is very nice,
- and in some places it is not as nice when compared to Mach.
-
- Ultimately, however, those level of services aren't even very
- interesting. For the users, applications, server services,
- and user interface are interesting. For purchasers, a variety
- of support and pricing issues are important. For developers,
- the API and tools are important. The quality of these is not
- as constrained by Mach vs. NT as other factors. The main point
- for Microsoft is that NT is good enough to build upon (i.e.
- it is a "real" full featured OS infrastructure), not that
- it is "God's gift" to OS technology. Infrastructure is very
- important -- look, for example, how much mileage NeXT gets
- from Mach ports or typed streams, or Microsoft has from
- DLLs or window handles.
-
- A final point. Microsoft has traditionally focused on customer
- needs. NeXT, and Jobs, have focused on anticipating the needs
- that customers don't have yet, but probably will. As the
- industry has become more market driven due to competitive
- and pricing pressures, direct focus on customer needs has
- been very profitable since the purchaser tends to think in
- those terms (esp. for corporations). There is a drawback to
- this -- when paradigm shifts occur, you need to make the
- long term big investments. GUI is one. Networking. Groupware.
- Early investments made Apple, Novell, and Lotus Notes key
- products today. In the future, pen, multimedia, consumer
- information services, and other things like that may be
- key. Microsoft and Bill Gates have focused their efforts
- in innovation in a different direction with a different
- vision from other companies. This is the perogative of a
- company -- to find their own direction. If and when you can
- obtain a system that gives you ubiquitous access to services
- and information from a variety of devices including mass
- market consumer ones like your TV, who is to say that
- company made the wrong choice by not focusing on making
- the most "insanely great" platform for object development?
-
- After all, no company can solve all the problems; so we
- pick the ones we think we have a good chance at solving...
-
- --
- Edward Jung, Software Architect edwardj@microsoft.com
- Advanced Systems, Microsoft Corp.
-