Re: Which language pays most 17457 -- C++ vs. Java?

["James Giles" <jamesgiles@worldnet.att.net>]

Lawrence Kirby wrote in message <882757510snz@genesis.demon.co.uk>...
>In article <67et6o$dql@bgtnsc03.worldnet.att.net>
>           jamesgiles@worldnet.att.net "James Giles" writes:
...
>>Exactly what I said.  Bill Gates (Slick Willie West) has learned his
>>lessons well.  He'd never have got away with it if UNIX hadn't set the
>>stage.
>
>The "stage" at the time for the development of DOS was CP/M. Windows was
>an attempt to counter the Mac. NT is greatly influenced by VMS. Unix
>has very little relevance here.

On the contrary.  The stage was the choice between, first CP/M
and then DOS on the one hand, or to buy a mini (or later, a workstation)
and run UNIX (probably with several terminals and users per UNIX machine
if you bought the mini).  That was the percieved choice of business users
circa 1980.  DOS only succeeded because people percieved UNIX as the
other choice.  They did not see that UNIX provided better performance (for
the money), or that is was more reliable - which it wasn't at the time.
Indeed, one of the problems is that most businesses had to hire additional
personel to support UNIX but not DOS.

I remember this because I was there.  I was rather amused by the
debate at the time because it didn't have much to do with me: at
work I used mainframes (which no one in those days would have
suggested using UNIX on) and at home I was a hobbyist without
the means to buy either a mini or a high-cost workstation.  A few
years later I had a SUN on my desk at work though (can't say I was
impressed, but it did have a really big screen).

>Whether you believe it is adequate or not Unix has a security system
>built in, it is a significant issue in the OS. If BG learnt anything
>from Unix it would have been that security is an issue. Clearly he
>didn't. I apologise for getting drawn into this troll.

To be sure, UNIX has better security than DOS and Windows (which have
none at all) but it's not really very secure.  In any case, when I mentioned
security as a desirable asset, I had in mind comparing the system to
other multi-user systems.  Only one UNIX implementation I've ever heard
of met Orange Book category B requirements (at the lowest level of
category B and only with enormous changes to how it did security).
Systems exist (or used to, before the UNIX revolution) which meet
category B and impose less of a burden on users in terms of confusion
and maintenence than UNIX security does.

And hell, I'm only in this discussion because I got drawn in by someone's
*defence* of UNIX which said that I should somehow expect systems
software to be *less* reliable than other kinds of software.  What kind of
a defense is that?  Certainly any reasonable person disagrees with that
assessment (including the person who originally said it).

...
> [...]  The ratio of installed systems to system
>>maintenence personel should be about the same as the ratio
>>of vehicles on the road to the number of auto mechanics: 500 to
>>1000 (or higher).
>
>What systems are you aware of that have this sort of ratio (and I mean
>systems that service user applications, not things like embedded systems)?

CTSS (Cray, at DOE labs) used to support several thousand users
with a system maintenence staff of about a half a dozen.  And, that
was not a mass-produced system, but an internally produced one
(you'd expect an in-house system to have *more* maintenence
problems, not less).  UNIX installations supporting this number of
users invariably require many times the support staff.  (In spite of
the fact that they're now running a vendor-supplied and vendor-supported
UNIX, I understand the labs have *more* of their own personel doing
system support than before.)  Furthermore, the users in such an
environment tend to be more demanding (full contact programming:
their applications are always right on the edge of what's actually
computationally feasible on the fastest hardware available) and
UNIX doesn't provide the same performance and reliability as CTSS
did.

>When security becomes a burden to the users it becomes counter-productive.

Yes, and UNIX manages to be both a burden and non-secure at the
same time.  For example, having each file independently specify access
permissions *sounds* like a convenient thing, but it increases the burden.
Most people forget to set or check their settings and/or set all their files
with the same privilages.  Doing so fails to use even what security UNIX
has.

Of course, the access privilages of files are a way in which the users
themselves can contribute to system insecurity, but they aren't in
themselves they main causes of UNIX being insecure.  That rests
in a lot of other bad design decisions.

I'm not saying that security never introduces a burden on the user,
but it is possible to design a security system that is less burdensome
than what UNIX has and yet is more secure.

Finally, I don't understand why everyone is in such vehement opposition to
me on this.  Even if you *don't* believe that there have been better systems
than UNIX in the past, you must admit that UNIX is *NOT* perfect; that ideas
for improved systems ought to be pursued; that among these ideas, what I've
described *might* have merit.  The only people I can think of who'd actually
oppose a more reliable and more secure system are those whose jobs are
to maintain the existing ones.

Well, argue it out among yourselves and Happy Holidays.  I'm
leaving and probably will not see another round of this "discussion".

--
J. Giles
Ricercar Software