To the Editors, Unix-Haters Handbook,
Have just received by EMAIL a copy of the preface to The Unix-Haters Handbook.
For interest, I append my own contribution to this literature, which recently
appeared in the “Starlink Bulletin” (the bulletin of the UK Starlink project,
which supplies computing facilities to astronomers). It got a pretty good
reception — also a lot of hate-mail on Usenet uk.org.starlink.misc It was
good to find that I Was Not Alone.
I look forward to a copy of the book as soon as my bookstore can get it.
Regards
Rachael Padman
or
VMS disappears (disappeared?) from the Cambridge Starlink node on the 1st of
April 1994. This is not because no one wants to use it, and it is not because
of demonstrable failings. It is because Unix is thought (by some) to be a
great improvement. In some ways it may be; in others it falls far short of what
is required in a so-called standard. Of course it is too late to
stop this happening, and the trend to Unix extends way beyond astronomy, but
neither of these facts means that it is necessarily correct to abandon VMS, nor
that we should not bemoan its passing.
These views generally result in my being labeled a Luddite by the Unix
priesthood. But Ned Ludd was right — the new machines did not make
conditions better for ordinary working folk: the people whose conditions were
immediately improved were those who owned the machinery, and whose profits
thereby increased. Similarly, the change to Unix will, in the short term
at least, do more to provide employment for our computers minders than to
improve the average astronomer’s relationship with her computer. Unix does have
its advantages, but few of these are intrinsic — in the sense that they could
not be implemented within other operating systems — and to my mind they
are far outweighed by the disadvantages.
Alright, I have to admit it. I don’t like Unix (or is it “unix”?) Of course
one can get used to almost anything. But is “rm” (pronounced “remove”)
really synonymous with “delete”. Do I call the deleters when I want to move
house. How would my first cousins once-deleted feel? And any real
language is at least pronounceable. I have only been able to come up with one
algorithm for creating Unix command names: think of a good English word to
describe what you want to do, then think of an obscure near- or
partial-synonym, throw away all the vowels, arbitrarily shorten what’s left,
and then, finally, as a sop to the literate programmer, maybe reinsert one of
the missing vowels. Mnemosyne, goddess of memory, must be rolling in her grave.
(She was, incidentally, the mother of the muses, including Urania, the muse of
astronomy.) It is notable that all the machines I have used interactively in
the past spoke to me in plain English. OK, they didn’t necessarily have a big
vocabulary, but typing HELP was always likely to return a menu of commands that
was almost identical from machine to machine and whose function was immediately
obvious. Although there are some really powerful features in Unix — such as
the pipe — what I see (and interact with) is a user interface designed
by adolescents and constructed deliberately so as to make access to the machine
as difficult as possible for a novice. Is this really a suitable foundation for
an international standard?
In reply to this criticism one is normally reminded that Unix was invented for
a PDP7, which had about as much memory as an earthworm and CPU power to match.
That is, the user interface was designed to minimize the time required to
decode a command. That meant short, fixed-case, commands, and no minimum
matching. But 1969 is a long time ago, and both memory and speed have improved
a thousandfold since then. This response thus goes directly to the core of the
case against Unix. It is old-fashioned. Why is the world now, 25 years later,
adopting as a long-term standard a primitive operating system designed when the
capabilities of today’s machines were then undreamed of? Actually, the answer
is easy. Operating systems cost hardware manufacturers, and therefore consumers
(us), money. Unix is free, and whatever its defects at least no one blames
the hardware manufacturers. So in a world that thinks megaflops or
Specmarks are the only indicator of computer power, Gresham’s law holds
— bad software drives out the good.
[This lemming-like rush to adopt Unix just now seems slightly curious. It will
be ironic if it eventuates that Starlink has felt it necessary to adopt Unix
just as Unix is itself superseded. Yet that appears to be a real possibility.
Windows NT has been designed with distributed processing and interoperability
as the major drivers. Here are two quotes from IEEE Spectrum, December 1993:
“… Windows NT, by covering almost all the latest RISC processors, opens
the way to a consistent computing environment. Delighted early users are
already contemplating making it their new standard.” And: “Thus Unix will go
on evolving, but largely in order to keep up with Windows NT”.]
Even the commands that are pronounceable are idiosyncratic to say the
least. How can one understand “biff”?. As people never tire of telling me, it
is the name of a programmer’s dog (but you’ll note, just in case you thought
this might help you to memorize the command, this name starts with a lower case
letter). It comes coupled with a twee American icon, that is meaningless in the
UK context. And there are other examples. Grep suggests to me that the author
of this one had been reading too much Robert Heinlein (you grok?), or possibly
— and this is in fact quite likely — was under the influence of
psychotropic substances at the time. Unix seems to consist largely of
arcanae which be learnt only by taking instruction direct from a priesthood who
seem largely to be stuck in the anal-retentive stage. When yesterday’s six-year
olds take over Unix development (in four or five years) I fully expect to see
new commands ldo, nja, mch and rfl. After turtles, dinosaurs. Will the command
tRex devour your competitor’s data?
As intimated above, Unix documentation, or the lack thereof, is a serious
problem. There is no effective way to find out how to use Unix, other
than sitting next to an adept and asking for help (and you know what these
people are like. Don’t you? Haven’t you ever been in a computing centre to pick
up some output late at night? Dirty cups half full of stale machine coffee,
many with cigarette butts floating in them, the floor littered with Twinkies
wrappers, insane giggling over some semi-pornographic picture cleverly printed
out on a line printer using different characters overprinted to get the
greyscale effect… Grab your output and run!). I remember on one occasion
being logged into a Unix machine to send some mail, and, not knowing the
command name, being unable to log out. It was an expensive 8000-mile long
telephone connection, but the machine steadfastly refused to respond to “help”
(or log, logout, logoff or quit). Finally someone suggested I type “man” (hey
man, what the hell do I do now?), and it said “Documentation is held in the
user area in room 932” (or something like that). Which didn’t really help.
Anyway, have you ever tried to use man? It’s fine as long as you know
what you are looking for. How would you ever find out the name of command given
just the function you wanted to execute? You can’t. I just tried to find out
what the librarian was called, and man was silent on this issue. Sure it gives
you pages of information about “ar”, but nowhere can you find out that it is
the pages about “ar” that you need to read. Oh, say the adepts, you just use
“apropos”. Apropos (or even “apropos”) means
“apt, appropriate”, which, even if you knew the command existed, would
not immediately suggest a way to find out information about something. In
addition to their other failings, the authors of this command are illiterate.
On occasion, as in this instance, the makers of Unix have gone out of their way
to avoid using the appropriate word. What was wrong with `help’?
What next? Oh yes, case-dependence. Well, e.e.cummings made a fortune, or at
least a reputation, out of refusing to use capital letters, and I guess Ken
Thompson might very well have regarded him highly back in 1969. But it’s only
convention isn’t it? Long after 1969 most terminals (teletypes we called them
then) only had capital letters anyway, and I suppose that was stifling
creativity or something. So if you want to use lower case letters of course
that should be allowed. But why oh why invent a system where the commands MV,
mv, Mv and mV can all do different things? Here is my question to Unix
apologists. Would you return your old-fashioned sneaker-net mail to the sender
if he or she had forgotten to capitalize your name? Even if it contained a
cheque?
Well, that’s the user interface. As for the applications programs … Even
something fundamental like the SunOS f77 compiler is a disaster. It doesn’t
include the full standard as a subset (look at their OPEN statement for
unformatted direct access if you don’t believe me), and it contains, according
to my Unix-literate colleagues, several major errors. Gresham’s law strikes
again. What about the editors? I don’t have time to go into that one! One can
only conclude that the makers of Unix held, and still hold, the ordinary
computer user in total contempt, and this viewpoint seems to me to be
mirrored in the attitudes of the people who are inflicting this awful system on
the rest of us. Does Unix’s enormously steep learning curve have any function
other than to deter the faint-hearted, those who may want to use
computers without necessarily dedicating their lives to them? It does not seem
fanciful to suggest that Unix is primarily about separating out the elite from
the proletariat, the real programmers from the quiche-eaters. It is about
taking back the mantle of the priesthood, that was lost when manufacturers like
DEC introduced operating systems like VMS. But of course it’s all worthwhile
really, because Unix is a standard, and that means it’s the same everywhere,
and will never change (see what I mean about self-defeating arguments?)
Let me quote again from IEEE Spectrum: “In reality, however, there are
many Unix-based operating systems; the major ones include Apple Computer’s
AUX, Digital Equipment’s Ultrix, Hewlett-Packard’s HP-UX, IBM’s AIX, the
Open Software Foundation’s OSF-1, Sun Microsystems’ Sun OS and Solaris, and
the Santa Cruz Operation’s Xenix and SCO Unix. With such an assortment, users
[who work] with platforms from several manufacturers [are] caught in the
bind of supporting multiple releases of multiple versions of Unix.” This
monoculture is divided by a supposedly common tongue. Unix is not a
standard any more than English is. One of the touted virtues of Unix
is that you can unplug any component and plug in something else
with the same name, so that you can tailor it to your needs (indeed the
Starlink .cshrc does this with your Tex commands!). So while you can go to any
VMS system anywhere in the world and be instantly at home, you can’t depend on
a Unix command to do what you think it should anyway. Sometimes even the
options are different. Take the tar command (please!), which is already a
nightmare where lower-case `a’ means “check first” and upper-case `A’ means
“delete all my disk files without asking” (or something like that — I may
not have the details exactly right). In some versions of tar these meanings
are reversed. This is a virtue?
OK, I admit it, I was joking there. But I’m not joking when I quote from
SUN 145:
- “Ctrl-D End of File”: Used to terminate data or text
entry from the keyboard. If you use it when being prompted for a
command it will log you out!” - “File size is shown in Kbyte on a DECstation or Alpha, but in
units of 512 bytes on a Sun.” - “Starlink’s Unix software can be run from the C-shell or T-C shell,
but not from the Bourne or Korn shells.” - “… but briefly, if the first line of a shell script is
a normal command the Bourne shell is used [but it doesn’t run Starlink
software — see above]; a comment, the C-shell is used, except that
…” Aaaaarrrrghhhhhh!
Douglas Adams may well have had Unix in mind when he described the products of
the Sirius Cybernetics Corporation thus:
“It is very easy to be blinded to the
essential uselessness of them by the sense of achievement you get from getting
them to work at all. In other words — and this is the rock solid principle on
which the whole of [its] Galaxy-wide success is founded — their fundamental
design flaws are completely hidden by their superficial design flaws.”
There is a distressing mindset amongst Unix cognoscenti. Most attacks on Unix
are answered with statements to the effect that make and grep are wonderful,
that redirection and piping are very clever, and that the whole system is much
more “elegant” than VMS. Well, I admit that make is wonderful. Still, it is
only an application, and as such could be implemented on any operating
system. As for grep… A few days ago I thought to search for each occurrence
of the $ character in a piece of code (so that I could move it from its
non-standard position at one end of the FORMAT statement to the alternative
non-standard position at the other end, to make it work on a Sun). Dutifully
I translated “SEARCH [-…]*.FOR”, and typed “grep ‘$’ ../*/*.f | lpr”
(Note the clever redirection of the output to the line printer.) The
following morning I found the entire source code of SPECX in a heap beside the
OKI printer. This does not strike me as very elegant! Neither is it elegant
that the X-terminal driver is so primitive that it reflects characters
immediately rather than waiting until input is requested, nor that everyone has
to include keyboard remappings in their .Xthingy files (in fact this part of it
is worse than a PC), nor that when talking to an applications program you can’t
edit the line you are typing except by using the delete key. It is not elegant
that logical names — sorry, environment variables — are not inherited by
programs, so that each program has to include code to parse input and translate
the logical name before opening a file or whatever. Well, I’m sure I could go
on forever. And the minute you criticize Unix to your Starlink friends they say
“Yes, I agree, but I’m not allowed to say that!”
A Starlink mole has
suggested to me that project staff are not unaware of the sort of Unix
failing detailed here, but that the change to Unix was largely mandated by
external pressures. Canute could not hold back the tide (and knew it), and
obviously neither can Starlink, even it wants to. It’s a pity however that the
tide seems to be largely composed of lemmings — that is, slightly
computer-literate users, who are persuaded by the Unix hype, fancy joining the
grep-awk priesthood, and mistakenly confuse the time spent learning how to use
Unix with productive work. Unfortunately they never have to suffer the
defects of the system they have foisted on the rest of us.
Well, it’s not really funny. The change to Unix has real implications for
astronomers, which have generally not been addressed. Yes, it means that we can
all run Unix mega-packages like IRAF and SAO-Image (name your favourite). Unix
and C however form a powerful deterrent to the average astronomer to write her
or his own code (and the average astronomer’s C is much, much worse than his
Fortran used to be). The powers-that-be in the software world of course have
always felt that “ordinary” users (astronomers in this case) should be using
software and not writing it. The cynic might feel that since those same powers
nearly all make their living by writing software, and get even more pay when
they manage other programmers, then they have a vested interest in bringing
about a state of affairs where the rest of us are reduced to mere supplicants,
dependent on them for all our software needs. It is clear that Unix does not
pose an insuperable barrier — the ever-expanding armies of hackers out there
are evidence enough that the barrier can be scaled given enough time and
enthusiasm for the task. But hacking is not astronomy, and hackers are not
astronomers, and it is astronomy and astronomers I worry about. We shouldn’t
have to scale the Unix barrier, and it is all the sadder because, since the
advent of a VMS-based Starlink, ordinary astronomers have had something denied
to most other scientists in this country — readily accessible, reliable,
user-friendly computing power that can be easily harnessed to a particular
astronomical requirement. Maybe VMS does baby its users. Maybe we have paid
more per Specmark so that we could use the Specmarks we had efficiently. But
along with the rest of the world, we are now losing this nice friendly system.
As with instrumentation and the National Observatories, we are having to teach
our students how to fit their problems to facilities provided by others,
whereas the UK reputation in astronomy was created by fitting the facilities to
the problem.
About the Author
