home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!pipex!warwick!uknet!gdt!aber!aberfa!pcg
- From: pcg@aber.ac.uk (Piercarlo Grandi)
- Newsgroups: comp.arch
- Subject: Re: OS/MFT and Macintosh (was: ..History)
- Message-ID: <PCG.92Sep5132252@aberdb.aber.ac.uk>
- Date: 5 Sep 92 13:22:52 GMT
- References: <PCG.92Aug27125826@aberdb.aber.ac.uk> <9209031044.AA00288@iecc.cambridge.ma.us>
- Sender: news@aber.ac.uk (USENET news service)
- Reply-To: pcg@aber.ac.uk (Piercarlo Grandi)
- Organization: Prifysgol Cymru, Aberystwyth
- Lines: 79
- In-Reply-To: johnl@iecc.cambridge.ma.us's message of 3 Sep 92 14: 44:36 GMT
- Nntp-Posting-Host: aberdb
-
- On 3 Sep 92 14:44:36 GMT, johnl@iecc.cambridge.ma.us (John R. Levine) said:
-
- smith> OS/MFT provided a "flat" file system.
-
- pcg> Really? That's news.
-
- johnl> It's true, mostly. [ ... VTOC was just a flat table of file
- johnl> names, somewhat like the ilist under Unix ... ] The catalog, on
- johnl> the other hand, was indeed tree structured [ ... ] But it
- johnl> appears that in most installations, people were very sloppy about
- johnl> using the catalog [ ... ] and there were a lot of strange flat
- johnl> names.
-
- Well know, unfortunately. A possible reason for Smith's mistake.
-
- johnl> PS: Followups to comp.arch.history, if it ever exists.
-
- Actually there is something more architecturally relevant to be said.
-
- Making a comparison between a microprocessor OS os the eighties with one
- of a mainframe in the sixties has to be taken with a wide margin of
- imprecision. Smith's nitpicking remark that MacOS is more like MVT than
- MFT looks rather silly in this respect; I could even defend the notion
- that MacOS is actually more like MFT than MVT, by pointing at some
- particular features in MVT that are not yet in MacOS.
-
- Actually it can be argued that MacOS (in its three incarnations, without
- multifinder, with multifinder, with single address space VM) is under
- certain respects more like the original PCP. While MVT had a variable
- number of partitions, and MFT had a fixed number, they also had a lot of
- things, like tasks, memory protection, scheduling, volume management,
- generalized locking, that are yet to be seen in MacOS.
-
- But the main gist of my observation was that, broadly, in the various
- architectural generations that operating system design has gone through
- (very roughly: none in the 40s, single program loaders in the 50s,
- multiple programs in the 60s, timesharing in the 70s, modularized in the
- 80s), MacOS belongs to the same generation as OS/360 or 370, in its
- various incarnations as PCP (first few releases of MacOS), MFT/MVT
- (MacOS 6) or SVS (MacOS 7).
-
- Indeed, as another poster observed, the original MacOS, and even today's
- MSDOS, was much more like the original OS/PCP, or maybe, let me say,
- more like the 7090 monitor/program loader.
-
- One sad thing about all this is that the machines on which MacOS and
- MSDOS run are some orders of magnitude more powerful and sophisticated,
- under many respects, than those on which PCP/MFT/MVT/SVS used to run.
- Hey, the GE645 used to develop Multics did not have the power required
- today to run Windows 3 or MacOS 7! And the original Xerox Alto or Star,
- which had far neater GUIs than the Windows or MacOS ones, could not
- possibly run either!
-
- Another one is that not only most major microprocessor operating systems
- are still stuck with technology of the sixties, but also with the least
- agreeable sixties technology. I mean, in the sixties one had CTSS, for
- example, which is still more advanced then either MacOS or MSDOS, and
- its successor, Multics, which is still far more advanced than things
- like OS/2 or "Unix" (e.g. in its 4.3BSD/SVR4/SunOS incarnations -- Mach
- is another thing really).
-
- I could even go as far as saying that most/a lot of the technology in
- currently popular OS architectures can be directly traced back to the
- fifties (my favourite example is 'ld' on many "Unix" implementations).
-
- All this is not encouraging. I seem to have noticed that as the computer
- market has expanded, the lead time for new technology has increased. I
- think that as things currently are, the time between successful
- demonstration of an architectural technology and its diffusion in the
- marketplace now stands at around 20 years (for example Accent,
- reincarnated as Mach3, was conceived about 15 years ago, and its
- underlying technology is only now becoming popular), if ever.
-
- This surely has (dire) implications for architecture in all branches of
- computer science.
- --
- Piercarlo Grandi | JNET: pcg@uk.ac.aber
- Dept of CS, University of Wales | UUCP: ...!mcsun!ukc!aber-cs!pcg
- Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@aber.ac.uk
-