[Previous] [Next] [Index] [Thread]

UNIX in Perspective (was: Re: DOS and Macro Virus Discussion)

Phill, I don't want to get into a flame war with you -- I have way too much 
respect for your technical competence.  However, permit me to violently 
agree with you...

>But MSDOS is not alone in this. UNIX has failed to develop significantly
>so far as the user interface is concerned since 1970. It is still based
>on a smug and complacent happy hacker attitude in which it is the
>responsibility of the user to adapt to the machine, where to expect the
>machine to be documented is to be a looser. In fact the more obscure
>the O/S is the happier the elite are - it protects their status and
>power. Like MVS, UNIX represented job security for tens of thousands
>of sysops to whom it gave power over lesser mortals.

UNIX was developed by programmers, for programmers, to make good use of the 
few available machine resources (speaking as a UNIX user since the Version 6 
days (1978)).  At that time for most of us, multiuser TTY access at 9600 
baud was close to the state of the art.  256K was regarded as a pretty 
sizeable machine memory size.  Given these constraints, UNIX performed its 
job admirably.  Much of the kernel API paradigm survives to this day, 
because they got it right on the issues they covered. Even today, there are 
times that a powerful textual-command shell can be of enormous help; I keep 
a Korn Shell (courtesy of MKS) up on my NT desktop continuously. Those days 
really did require separate system administrators for computers, as 
computers were just too expensive to dedicate to individual users.

Some issues not covered by the original UNIX design:
* Threads -- Although most current versions incorporate adequate support.
* Security -- UNIX required a ground-up rewrite of the kernel to incorporate 
security properly.  (This rewrite is called Windows NT :).  As UNIX was 
developed in a near-academic environment, security (apart from not stepping 
on each other's files) was seen as a hindrance to information sharing. 
 Obviously, the computing philosophy of security has advanced since then.
* Reliability -- I may be mistaken, but the great advances in mainframe 
reliability occurred during the same period that UNIX was under initial 
development.  Again, a complete rewrite of the kernel would be required. 
 Probably the kernel API would then only need a very few changes, as 
reliability should be transparent to the user and the programmer.
* Graphical User Interface -- GUIs were experimental during the early UNIX 
development period.  The main fault UNIX systems now have is the lack of 
adequate GUI system administration tools, which is perhaps a limit of the 
UNIX culture -- it is certainly not a technical limitation, as there is no 
reason that a UNIX system could not be completely graphically administered.
* Networking -- The lower level (socket C API) survives to this day, even in 
Windows and Windows NT.  We are all still figuring out what the higher 
levels need to look like.
* Documentation -- UNIX documentation (like the UNIX kernel and shell) was 
written by programmers for programmers.  Generally, I have found UNIX 
systems to be much better documented (at the programmer's level) than MS-DOS 
and Windows systems, which require the purchase of multiple "*Secrets*" 
books just to get the systems to perform the simplest tasks!  To answer 
_your_ point, UNIX system documentation for non-programmers still leaves a 
great amount to be desired even at this late date.

Although I have certainly run into system administrators who would be best 
suited to running otherwise unattended computer rooms in the Antarctic, most 
of us try our best to supply our users with the computing resources they 
need to do their jobs.  Unfortunately, no matter what the profession, a few 
will regard it as a badge of honor to bamboozle their customers by hiding 
behind a wall of jargon in order to keep the profession pure.

UNIX was written by programmers for programmers -- but that is not the 
majority of computer users now.  Unfortunately, the major UNIX vendors have 
not been willing to commit the resources to move to UNIX++ or VOJY :).  From 
my perspective, it may yet be that Windows NT will be the rightful heir to 
UNIX, as it incorporates the good points of UNIX with the more modern 
computing paradigms of effective but not annoying security, reasonable 
reliability, and the graphical user interface.  Because of UNIX's power, 
which (at the this stage of computer science) comes at the cost of 
complexity, a priesthood was able to grow up under UNIX.  This is not the 
fault of UNIX -- it is the fault of modern computing science, which still 
produces systems that are just too hard to use!
</rant>

We now return you to your regularly scheduled Web security issues 
discussion, already in progress...
======================================================================
Mark Leighton Fisher                   Thomson Consumer Electronics
fisherm@indy.tce.com                   Indianapolis, IN
Follow-Ups: