UNIXCPM.WS4
-----------

"Unix Feedback"
in "Letters" column, BYTE, August 1982, p.20

(Retyped by Emmanuel ROCHE.)


     I find that I grow tired of the Unix-versus-CP/M argument,
particularly as it is phrased by people like John Lynn Roseman
(April 1982 BYTE, "Letters", page 22): "Unix is a full-featured
operating system which is widely regarded as the finest ever
written, while CP/M is little more than a program loader." Really?
I defy anybody to take a competent secretary and make him or her a
useful word-processing person on the Unix EX/VI in less time that
it takes to get your work done on the CP/M WordStar system.

     And I don't like the crystal-ball predictions and dogma-
before-the-fact apparent in Mr.Roseman's statement: "... we can be
sure that the commercial software which will eventually be
available under Unix will be of higher quality than that found in
the CP/M market." We can?

     I direct your attention to an article by Donald Norman that
appeared in DATAMATION magazine all the way back in November 1981
(page 139 and following). It is titled "The Trouble with Unix",
and it hits a number of nails on the head. Although I am fluent in
a number of dialects of a number of languages and in a number of
operating systems, I still haven't found the ultimate anything.
CP/M has a number of serious limitations, but so does Unix (and so
does anything else that I have ever used).

     Allow me to paraphrase Norman's conclusions, in which he
states his three most important concepts for system desing: be
consistent, provide the users with a clear idea of what is going
on at all times, and provide mnemonics as aids to us poor humans.
I would add a final imperative: remember the users' context. In
other words, decide what you want to have a given system do, and
for what audience. CP/M is a tremendous environment for single
users doing word processing and data acquisition; BASIC is a
wonderful tool for a wide range of (generally small and one-of-a-
kind) programming tasks; Unix is an amazing tool for some of the
data-intensive work that I sometimes need to do.

     But, please, give us a break from the search for a perfect
system for all people for all time. Provide me with information,
tell me (as objectively as possible) about the tools that are
available, and then leave me alone so that I can get my work done.

     Jeffrey L. Star


     I have some bad news for John Lynn Roseman and the recent
crop of university-type Unix supporters. Unix has been running on
16-bit computers called Digital Equipment Corporation (DEC) PDP-
11s for many years. There have been some other operating systems
for the same machines. Guess which operating system is NOT at the
top of the popularity list?

     The most popular operating system on PDP-11 computer large
enough to run Unix is RSTS/E. The primary language used with
RSTS/E is BASIC PLUS, not the "powerful C language". While RSTS/E
is used in the commercial PDP-11 environment, RSX11M is more
popular on the scientific systems. When DEC introduced the VAX11
superminicomputer, it did not select Unix but rather upgraded
RSX11M. I have never even seen an advertisement for a programmer
with Unix or C experience.

     This is not meant as a criticism of Unix or C, nor is it
meant to endorse RSTS/E or RSX11M. I would be tempted to write off
RSTS/E as a primitive, crude system except that it is enormously
popular and its users extremely enthusiastic. The marketplace is
different from the university classroom. The needs of the end user
are different from those of the system software developer.

     CP/M is a rinky-dink kind of operating system. It does,
however, do most of what most microcomputer users want, with a
minimum of fuss. I have used various operating systems on IBM,
DEC, and Control Data Corporation machines, and I don't feel
neglected or abused by CP/M.

     Unix, C, and Pascal may be excellent teaching and development
tools, but they may not be so good for commercial production work.
While we, Old Timers, must be open to new ideas, the new crop of
computer science graduates must keep in mind the difference
between theory and practice. (By the way, what ever happened to
ALGOL?)

     Mike Draper