Christian wrote about servers:
>Well, I just (20 minutes ago) finished installing a Linux server
(file
>print, authentification and Xapps server) for a lab here at our
dept. It
>took me 15 minutes to do the base install of the RH51 off CDROM,
another
>15 minutes to dump the user database from the old WinNT server
and stuck
>it in the smbpasswd, and another 30' to transfer over the net all
the user
>files from the old server to the new one. Guess what? The guy who
will
>take care of the routine admin of this server left for his
vacations with
>a free spirit because he hasn't anymore to be afraid of the
damned GPF's
>of the WinNT server he used to get twice a day during the last
week. Guess
>again? His users will never have a clue about a new server
(unless they
>will realize their network now works around the clock).
I expect and insist that any suitably-trained person can install
an OS and make it reilable and secure. You too are obviously
well-trained and I salute you as a fellow-server builder. There
are of course those that have the Linux chromosome enabling them
to install Linux from inside the womb by telnet - but that gene is
very rare I think. One of my points is that in the future, such
staff and us lesser people who need to be trained will be too
expensive. Similarly at the present time, to make a Unix box (or
any box) secure from hacking, is non-trivial and requires constant
tinkering as weaknesses are found and fixed. Hence the need in the
future for an OS which does not require these chromosomes or
training.
It is possible that I am wrong about the next, say 20, years but
let me point out two things: 1. The demand for Secure
Internet-proof servers will explode in the future and there just
ain't the trained people to go around. 2. The general population
can not, and is not interested in training up to this sort of
level. I belong and contribute to another mailing list which is
still grappling with the concept of making a new folder in Win95.
Christian wrote about client workstations:
There was OS/2 (killed by back-kniffing) there is BeOS (attempts
to squash
>it still to come), there was Rhapsody (killed during pregnancy by
its own
>mother) there was NeXTStep (the best and the greatest, so
"greatest" that
>was lapidated to death like all the outstanding "mostruosities"
known to
>the history). Still, I can give you 5 to 10 (personal) live
>(generalizable) examples where Linux (with KDE and XEmacs) is THE
ONLY
>viable workstation (scientific computation and programming, which
you have
>to agree HOLDS importance, in any way more than the Joe SixPack's
Flight
>Simulator "training" or pornographic sites "surfing").
>
I can think of a few other instances where specific machines and
OSs were very sucessful. The IBM 525x series was a great
entry-level floppy disk system for small businesses. The Commodore
Pet with its IEE488 interface was a huge success in labs and
Universities and was the machine of choice instead of the costly
HP alternatives. A very handy little box. The IBM Series/1 was
also used in a specactular range of applications. The good old
PDP8 was also very sucessfull.
But where are they now? (Mostly in my garage, along with my Valve
(tube) computer and 5-bit teletype <g>)
As of today, the economies of commercial software writing means
that when I write a marketable application, I want it to reach the
widest possible customer base. In the current climate, I can write
something under Win9x/NT which I know will reach the most
customers first. Such economies of scale means that I can spread
the costs of development over a large number of customers, leeping
the selling price down so I can go on buying antiquated
computers<g>.
Whatever the history, we are in a situation where the Win9x and NT
client is predominant. Clearly, that OS-type will stay predominant
until some alternative comes along. To go forward from here,
people need a good reason to switch and a way of found of enabling
software vendors to port their applications en-mass. Not a trivial
task.
Perhaps others would care to comment on how this could be done.
One way could be to write a Win9x/NT clone under GNU public
license and then begin a gradual strategy of weaning. But there is
a danger of there being a Rad Hat version of the clone, a Calderer
version and so on. That will defeat the original purpose. In
addition it will not address the problems of the basic limitations
of the way the Win95/NT OS works - and any attempt to produce a
widely used alternative OS will still have to face the problem of
changing people to the new system.
Kind regards
Tony Wells
Phenomenal Books
"I have made this letter longer than usual because I lack the
time to make it shorter" - Blaise Pascal.
[EMAIL PROTECTED]
bookstuff: www.phenomenal-books.com
anyotherstuff: [EMAIL PROTECTED]
Intnl tel/fax: +44 1524 845559
UK tel/fax: 01524845559
Mobile: (+44) (0) 370 963410
-----Original Message-----
From: Cristian Tibirna <[EMAIL PROTECTED]>
To: Tony Wells <[EMAIL PROTECTED]>
Cc: [EMAIL PROTECTED] <[EMAIL PROTECTED]>
Date: Friday, June 26, 1998 3:54 AM
Subject: Re: Should we be pushing Linux over Windows 95?
>On Fri, 26 Jun 1998, Tony Wells wrote:
>
>> For the client workstation ease of use and penetration of
>> application software is the driver. Win9x and NT are the clear
>> winners here. Those client OSs will stay with us until another
>> supplier provides a viable alternative. I'm not holding my
breath.
>
>There was OS/2 (killed by back-kniffing) there is BeOS (attempts
to squash
>it still to come), there was Rhapsody (killed during pregnancy by
its own
>mother) there was NeXTStep (the best and the greatest, so
"greatest" that
>was lapidated to death like all the outstanding "mostruosities"
known to
>the history). Still, I can give you 5 to 10 (personal) live
>(generalizable) examples where Linux (with KDE and XEmacs) is THE
ONLY
>viable workstation (scientific computation and programming, which
you have
>to agree HOLDS importance, in any way more than the Joe SixPack's
Flight
>Simulator "training" or pornographic sites "surfing").
>
>Guess what? Ease to use has nothing to do here. Marketing has.
Windows
>is easy to use because every idiot got suffocated with it from
>kindergarten to university. One gets used even with a Boeing 747
cockpit
>if enough time is available. But ask my colleague which is now
crying here
>next to me because his WinNT just "BSD"-it with all his code
files 15'
>work worth, ask him, what ease of use he got?
>
>>
>> Back to servers. I realise that what I am going to say next
could
>> be considered contraversial. But lets be realistic. Looking
into
>> the next century neither Unix or Linux is suitable for the
needs
>> of that century. Considerable time and training (as evidenced
by
>> the number and variety of questions on this list) is needed to
set
>> up and maintain a secure Unix/Linux server on a corporate site.
>> Growing pressure on limiting support costs, staff costs and
plain
>> business needs for resilience will result in a demand for
simple,
>> self installing and self-maintaining servers that require
mimimum
>> training of staff, support and downtime. Clearly at the moment
all
>> current OSs do not fit those requirements.
>
>
>And about mailing lists. Have you ever wandered on an NT list? I
did (for
>necessity) I have twice the white hair I should have because of
the dumb
>shitty questions I had to float on there. Here the people get
stuck
>because of their laziness. They don't want to read. There the
people get
>stuck because MS learned them that they have a stupidity-proof OS
(which
>isn't true) so they act in consequense (I mean they act stupid)
and
>because if WinNT doesn't want to recognize your ether card, it
wont, no
>matter what you'll do. And there's no Torvalds, Cox, Baker to
save you
>overnight.
>
> Cristian
>
--
PLEASE read the Red Hat FAQ, Tips, Errata and the MAILING LIST ARCHIVES!
http://www.redhat.com/RedHat-FAQ /RedHat-Errata /RedHat-Tips /mailing-lists
To unsubscribe: mail [EMAIL PROTECTED] with
"unsubscribe" as the Subject.
