Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.1 6/24/83; site mcgill-vision.UUCP Path: utzoo!linus!philabs!micomvax!musocs!mcgill-vision!mouse From: mouse@mcgill-vision.UUCP (der Mouse) Newsgroups: net.flame Subject: Re: UNIX, UNIXpeople, USENIX ravings Message-ID: <317@mcgill-vision.UUCP> Date: Mon, 14-Oct-85 05:09:16 EDT Article-I.D.: mcgill-v.317 Posted: Mon Oct 14 05:09:16 1985 Date-Received: Tue, 15-Oct-85 07:33:30 EDT Organization: McGill University, Montreal Lines: 210 FLAMETHROWER SET ON "GOLDEN BROWN" > Well, I haven't had any choice. For the past couple of years, I've > been forced to use Unix to get my job done - the choice was made for me. > And, despite repeated reassurances from Unix-people that "you'll really > like it once you get up to speed", it's still at best an uneasy truce. It's not for everyone (what is?). > There's no question that it has its good points. Structured directories. > Pipes. History mechanism (yes, I'm sure everyone's yawning). You missed one of the greatest. Modifiability. You don't like the way (say) the shell handles `cd'? Fix it! You don't have source? Well, if it REALLY bugs you, rewrite the shell. It's not THAT bad. > But, the documentation... I'm really tired of illiterate ramblings and cute > little social commentaries and other trash (which seems to be particularly > endemic to Berkeley "documentation") - especially when what real information > is therein is so sketchy that one ends up having to struggle for days to > figure out how do things with Unix that could be determined in a few minutes > with a DEC or IBM manual. One simply can't do anything very sophisticated This is one of UNIX's weakest points. The manuals are written as reference works, not tutorial works (gee, now, what are the arguments to sendmsg() again...). However, they are not tutorial works at ALL. > with a DEC or IBM manual. One simply can't do anything very sophisticated > with Unix without (a) the source code, and (b) a (shudder) "UNIX-person", Usually you have the source. It's dirt cheap for universities, and not all that bad for industry, considering the usual R&D budget. And UNIX is NOT a production environment! UNIX is a development environment; that's how it was created and it still shows. If you are "just a user", UNIX is not the right system for you to do *anything* except run a program someone else has already written for you. > with Unix without (a) the source code, and (b) a (shudder) "UNIX-person", > which has already spent the better part of its adolescence blundering across > all the stupid little quirks which users end up fighting on their way to trying That should be "or", not "and". Anyone who can read and write C and who knows data structures, algorithms, and software engineering (as opposed to most of the "computer science" graduates nowadays, but that's another story) can, or should be able to, take the source code and do useful things with it. And anyone who's doing anything requiring source code twiddling certainly ought to be such a competent person (or have such a person to do the actual work). Now just wait a minute! I am deep into UNIX internals, having been exposed to other several OSes. "Better part of adolescence"...hm. I used VMS for about two years before encoutering Eunice, a UNIX emulator. After being hopelessly confused by Eunice for about a year, we got true UNIX. This was about two years ago and I'd say the past six months or so have been pretty static as far as my skill with UNIX goes. Is a year and a half the "better part of adolescence"? Especially when my (physical) adolescence was several years ago? > to accomplish in 2 weeks what would take 3 days with a good commercial > operating system. Let's try a simple experiment. Find a VAX running VMS. Now try to write something which finds all processes belonging to a certain user (say they were logged in over a modem and got hung). Automatically, that is; this happens a lot and you're getting fed up with doing it by hand every time. Now try it on UNIX. Put the following alias into your .cshrc and then "finduser " does it. Or if you get really picky, write a program (about ten lines) which takes a username and prints a uid and put it in backquotes in the alias; then you can use usernames. % alias finduser 'ps alx | awk '\''{if $2 == \!:1 { print; } }'\' > It works just fine for sending mail around, or for editing (as long as you > don't try to do anything very sophisticated with any of the plethora of > editors). I notice it has at long last learned about some little things like Emacs can do some VERY impressive things. Or are you too timid (or afraid to have your carefully built preconceptions shattered) to have bothered to try Emacs? Remember, Emacs Makes All Computing Simple. Yes, Simple. > editors). I notice it has at long last learned about some little things like > memory management (or has it, really?) and task-to-task communication > (barely) and... how 'bout shared resident memory, and, and, and... And it's > hardly possible for anyone to apply what little Unix has in the way of such > "sophisticated" features (they're fundamental to most other O.S.'s), without > having to become a "Unix-wizard" - the term itself being testimony to the > infantile mentality of Unix-people. Memory management? You mean like virtual memory? That's in the hardware on machines which support it; on machines which don't you can't get it with other OSs. What do you mean "barely"? Certainly no worse than (say) VMS. Shared resident memory - see what I said above about modifiability? Well, I wanted shared memory, so I put it in! Work: About one week, an hour or three each night. A UNIX wizard is defined as someone who's good at using UNIX. So your statement suddenly is revealed as tautological. Except for the last 11 (12?) words. Why is this a testimony to the infantile mentality of anyone? Someone who can wrap VMS around his little finger and make it tap-dance is called a VMS wizard (with reason, VMS is pretty ugly if you want to do something fancy); why shouldn't the UNIX analogue be called a UNIX wizard? It does imply the person is substantially better than most; again, UNIX is no different from the rest. > Unfortunately, despite its undesirability in other respects, there's > considerable incentive to use Unix due to its portability. When an O.S. is > needed for a new system, Unix can be brought up quickly, since most of it is > written in C. What gets overlooked by the naive management which allows the > thing into the company, of course, is that (1) they're going to be forever > tweaking and grooming and hassling and hacking in an effort to get it to run > efficiently - which is hopeless, since it will never be as efficient as a > completely native O.S. no matter how long one fiddles with it - and (2) as > long as they keep attempting to use it, they're going to have to put up with > Unix-people... When you get an operating system, you trade off certain things against certain others. One of these is portability. If you need efficiency, UNIX is not the way to go. UNIX is NOT A PRODUCTION SYSTEM! It is a DEVELOPEMENT system. By the by, in what sense is UNIX not a "native" OS for, let us say, a VAX? Or a SUN, for which I know of nothing else available? Why say "put up with" UNIX-people? They are good at what they do, you generally need only one, and UNIX plus wizard usually costs less than your average vendor OS licence plus software support plus updates. Besides, with a vendor OS you have to rely on the vendor for bug fixes etc. If they decide to listen to you at all, you tend to get "we intend to fix this in release 4.5, due out in March 1987". With UNIX, your support is just down the hall and is not likely to take more than a few nights. > (The commercial mainframe manufacturers could take a lesson from this... > If a package is portable, people will buy it even though it's trash - > and that situation is not going to change. It's a big selling point.) Agreed. Would that consumers would refuse portable trash.... > Of course, just as often, it gets used for little or no reason: for instance, > because a gang of Unix compunerds, again characteristically from Berkeley > or some similarly virulent seedbed, infiltrated a computer-center dragging Unix > in its wake, snowed the appropriate set of ignorant bureaucrats, and then > proceeded to inflict Unix on the resident mainframe and its unfortunate users. > After all, it's inexpensive, compared to a real commercial product (you get > what you pay for, of course) - and the Unix-people are more than happy to > sit up all night eating Twinkies and hacking yet another fully-customized > installation into existence, all the while congratulating each other > upon their wizardliness... If it is truly "fully-customized" they you shouldn't have any complaints! If it's not what you wanted, there's more customization to be done. > A large proportion of the people one finds "supporting" Unix systems grew up > with Unix and have never used anything else - Unix is their religion, and they > have no perspective at all on operating systems or even software in general. > As an experiment, try discussing another O.S. with one of them - and observe > the scandalized, intolerant looks you get, as if to say, "How DARE you even > even SUGGEST that any O.S. other than Unix even exists!!!" They generally You get much the same attitude out of (say) Digital with respect to VMS. Ask any VAX/UNIX installation how many times the local hardware service engineers have hassled them about hardware faults "well, if you'd just run VMS...". Besides, what's wrong with this - in a UNIX support person? This is very likely the best person around for supporting UNIX. If they're narrow-minded, that's a personal issue and has nothing to do with whether or not they should be out there making money for keeping a UNIX system up and running. You can pay a UNIX wizard to maintain UNIX or you can pay a vendor to maintain their OS. > even SUGGEST that any O.S. other than Unix even exists!!!" They generally > have a very limited skill set - very few of them can be described as So do the people who maintain vendor OSs. > have a very limited skill set - very few of them can be described as > software engineers or computer scientists. If anyone ever markets a really This is also true of practically everyone out there working on computers. Software engineering is NOT a common skill; its absence is not restricted to UNIX support persons. > software engineers or computer scientists. If anyone ever markets a really > well documented Unix which doesn't require babysitting by a phalanx of > provincial Unix clones, there'll be a lot of unemployable, Twinky-braindamaged > misfits out deservedly pounding the pavement. Nope, they'll go to work for the vendor as support staff. Or are you under the impression that there is ANY OS that doesn't need a plethora of support? > For a real eye-opener, check out a Usenix convention. I went to the last Right - the really good people are too busy to go; when the fact that they are really good is found out, they tend to get lots more to do, for some strange reason.... Well, maybe that's not quite fair. You do hear of DMR or Fair or those people attending USENIX conferences sometimes. -- der Mouse {ihnp4,decvax,akgua,etc}!utcsri!mcgill-vision!mouse philabs!micomvax!musocs!mcgill-vision!mouse Hacker: One responsible for destroying / Wizard: One responsible for recovering it afterward