ethernet, unix
ucbvax!works
Ethernet:
People are comparing apples and oranges here. Ethernet has
had a long period of testing in the field. If you want to
buy a technology that's known to work, buy Ethernet.
Broadband data nets (not broadcast cable video) now being
"announced" by venture companies, and even Wang, are all
promise... I'd get a contract that let me sue if they don't
satisfy my needs if I were to risk my company on those
technologies (they may be "better" eventually, but that is
only one aspect of considering them--would you buy a BART?)
Most of the technical "differences" advertised by manufacturers
fall into the "brand differentiation" area here -- when
all is said and done, looking at the whole system, the approaches
probably will all work about equally well. whose snake
oil do you buy?
UNIX:
It sure is nice not to have to build an operating system for
each machine you build. Thus hardware makers will love UNIX.
On the other hand, the concept of an "operating system" is obsolete.
What is needed for workstations or any other computer is a set
of easy to use primitives for manipulating the kinds of abstractions
that are of interest in building applications. For many of the
things I used to do, MACLISP (interlisp as well) is the programming
system of choice. Many users had built a huge library of wonderful
primitives for data structuring, debugging, screen managment, ...
If it runs on UNIX, fine. But I'd never use UNIX raw for those things.
UNIX is good at streaming-file-oriented processing (in fact it makes I/O
devices look like files so that the file-oriented tools can work on
character streams from other sources). For objects not structured as
byte streams (databases, knowledge rep systems, window management systems,
simulation modelling, video game construction, robot control, ...)
UNIX provides no help. In fact, since UNIX provides rather clumsy
processes and access to I/O, UNIX can't easily be perverted to do these
things.
At times it becomes useful to attack established dogma. UNIX is
an embodiment of a lot of clever ideas that comprise a model of what
computing is. I assert that computing for the 80's is best treated
by a radically different model or set of models. Little or nothing
is contributed by UNIX and C other than a programming system that might
be used with the intent of throwing it away when the better ideas
jell into practical systems.