garry@TCGOULD.TN.CORNELL.EDU (Garry Wiegand) (01/11/87)
I'm cross-posting this from the "comp.unix.questions" netnews group (no, I know of no way to get a mail subscription to it) because I think the subject might be of general interest... if some votes in favor (email please!) I will cross-post the previous articles as well. .............................................................................. Oh here we go to the wars again! A lot of what Barry says is perfectly true, but allow me to edit a few: In a recent article bzs@bu-cs.BU.EDU (Barry Shein) wrote: >From: alan@herman.UUCP (Alan Kiecker) >>One of our directors is on an advisory board to one of our state colleges. >>Currently the college is using a VAX VMS system, but is considering moving >>over to UNIX... >... > > 1. Business perspective - Unix offers the college "vendor > independence", particularly after the first change. Whereas > VMS is sold only by DEC (and at a hefty cost for both hardware > and software)... Unix licenses and source are derived only from AT&T and there is no "vendor independence" for the software (though there's rumors of new Unixae...) If you mean "it runs on more CPUs", this is assuredly true. "Hefty cost": Last time I had need to inquire, commercial Unix for large machines was priced comparably to commercial VMS. *Both* big $$$$. And I thought AT&T was going to stop giving free Unixes to universities. Can anybody comment? > 2. Industry/University Relations - true > 3. Job Perspectives - I invite you to pick up the Sunday > employment section of a major newspaper (try the Boston Globe) > and make a hash mark every time you see a UNIX position and > others. It should be revealing... possibly true - I don't get the Globe. (I *thought* the most popular was something called "OS/MVS" :-) > > 4. Academic considerations - Unix has been highly praised > for its consistent goals based upon innovative principles > of software engineering and design. VMS is expensive. The Unix kernel is excellently designed. FLAME ON The Unix utilities, from a high-quality software-engineering point of view, are trash. Which what the students have to contend with. FLAME OFF You're repeating yourself about "expensive". > 5. Commonality - Over 90% of all CS departments in the US > use UNIX...[describes Unix-related books]... Try to find VMS books... CS - true (Unix was free for years - excellent marketing move) Books - true. But not very interesting. The Unix utilities that are well-enough written that someone can actually document them seem to be well-enough written that they also run fine under VMS when I try them. > 6. Manuals - The full set of UNIX manuals is readily available > and inexpensive (a very complete set should cost around $100, > a more than adequate set probably less half that.) ... > VMS manual sets are several times more... > expensive and not readily available. I've never seen a tutorial > or self-help type book on VMS in the popular press... Not fair. The Unix manuals are cheap because they: 1) Are cryptic and obscure, 2) Are not kept up-to-date, 3) Contain no examples, 4) Contain no index, 5) Assume that you're a wizard. DEC publishes a number of "tutorial and self-help" books -- there aren't the popular press books partly because the original manufacturer has acted responsibly and published them itself. After all, would you prefer to buy a "Guide to programming in Fortran on XXX" or "Guide to Text Editing" from the manufacturer or from some unknown hacker author? > 7. Personal Development - UNIX is available for the home and > promises to be more so in the very near future, this will > almost certainly never be true for VMS.... true (although personal uVAXes are getting closer and closer...) > 8. Faculty hiring - If 90% of all CS depts use UNIX there must > be a few people out there available to teach it. true. But the professors I've encountered avoid at all costs discussing practicalities like "talking gracefully to the system", so their presumed expertise won't help the students. > 9. Future - I'll make the brash prediction here that VMS has > around 2-5 years (max) left. I don't consider it very responsible > to teach students a system who's days are numbered. The VAX line > itself (which VMS is hopelessly tied to) seems to be nearing the > end of its useful life span. yes, there are indeed starting to be rumors around of what's next - I'm looking forward to it eagerly! Unix is now 15 years old, VMS is now 8 years old, it's most certainly time to get on to the next generation! > 10. UNIX has better games :-) true I've tried to limit myself to saying which of your reasons for changing I don't consider valid; I haven't gone into all the obvious reasons for *not* changing (the original poster didn't ask for them.) But I will mention one: well-thought-out and graceful user interfaces are very important to me, and I suspect as time goes on they will become more important to the rest of the world too. It *pains* me for students to be subjected to Unix first, with commands and error messages that read like modem noise, and for them thus never to be aware that the user interface is something anybody considers worth worrying about. Enough said. > -Barry Shein, Boston University garry wiegand (garry%cadif-oak@cu-arpa.cs.cornell.edu) PS - I'm looking forward to our new Cornell/Dec pricing agreement - they're bundling in bunches of interesting compilers and layered products for a cheap flat price.
doug%noah.arc.cdn%ubc.CSNET@RELAY.CS.NET (Doug Konkin) (01/13/87)
Oh Garry, why did you do it??? Let me start by saying that most of what you posted boils down to personal preference, and you and Barry (and me, once I'm done) will all be guilty of gross personal bias because of what we have said. Let's face it, however well reasoned an argument is presented on this sort of thing, someone on the other side can refute it, whether sensibly or not. That said, I side with Barry. I have a lot of experience with both UNIX and VMS, and my preference is for UNIX. I am a computer scientist, and I find that UNIX is a far richer 'playground' than VMS. Sorry, but I can't agree with you about the quality of utilities and user interface. For the experienced user, which I believe CS students had better become, UNIX is quicker and easier to use (see that bias creeping in :-) ). Documentation is also strictly personal bias. While the UNIX documentation is often cryptic, it fits on the corner of my desk. The VMS documentation doesn't quite fill my bookcase. Funny, but I can't report any more success finding the infomation that I need in the VMS documentation than in the UNIX books. I guess obsfucation is as big a sin as being overly terse. I think that the thing to keep in mind is the audience (CS students) and UNIX becomes an excellent choice, VMS only a fair one. Without source, for example, how are the operating system students going to hack the kernel? And do you want to do without pipes, redirection, and hack? Doug Konkin Alberta Research Council Calgary, Alberta. T2E 7H7 (403) 297-2662
cetron%utah-ced@UTAH-CS.ARPA (Ed Cetron) (01/13/87)
hoping to head this off at the pass, here is a response I used in trying to stop this the last time: Date: Tue, 8 Jul 86 10:39:43 MDT From: cetron@utah-cbd.UTAH-CS (Ed Cetron) Message-Id: <8607081639.AA00157@utah-cbd.ARPA> To: info-vax@sri-kl.arpa Subject: RE: fortran follies and un*x vs VMS I thought that I'd stay out of this fracas but some of the garbage the fanatics are printing has finally got me upset enough to write.... From where of I speak: I speak both FORTRAN and C, as well as being familiar with both VMS and Un*x to the point of some systems programming on both....(but my true love is RSX-11M/M+....) First off, any one who tries to use FORTRAN directly to do some of the functions mentioned earlier ( :fortran follies) obviously doesn't even remotely understand the interaction/synergy between the OS and the programming language. FORTRAN is NOT designed to do these kinds of things and it is hard to do them via FORTRAN but that is not the folly, the folly is even thinking/trying to... ( And yes, Un*x F77 is even worse to use....) On the other hand, C does this kind of stuff extremely well - but it is terrible for large brute force number crunching - and it is ludicrous to use it in that fashion.... One must also remember that certain combinations of OS's and languages have been 'designed' to work well together. The most obvious examples are VMS & VAX FORTRAN and Un*x and C. Using the other combinations are ok, but not the optimal usage (unless you only have one machine....) Indeed, both VMS and Un*x have their own strengths and weaknesses.. I have spent 17 years working on computers, the last 7 on DEC hardware, an have played with most everything out there at one time or another (except maybe Burroughs and Honeywell). And except for the IBM which are user unfriendly (try hacking in JCL sometime) and have NO redeeming values (other than the backing of BIG BLUE which in some cases IS important), each had a particular type of activity for which it was optimized. The CDC 6600 I programmed in COMPASS (60 bit words...with multiple assemble instuctions in each word....and different syntax depending on the position within the word) was terrible to program - but it was extraordinarily fast for large numerical calculations. To the extrapolate and therefore say it would be the 'best' for controlling the lights/heating/cooling/etc of your home is insane... A true computer engineer ( distinct from computer scientist - we engineers like to USE computers for practical applications - not postulate them) use the right tool for the job at hand. A carpenter doesn't use a hammer to screw in a nail, or a screwdriver to drive a nail...so how can he say the hammer is 'better' than a screwdriver. So how can a person who represents himself to be an 'expert' or a PROFESSIONAL !EVER! say Un*x is better than VMS or VMS is better than Un*x... I use both, fluently and PROFESSIONALLY. When I need to develop large interactive programs which need to be tweaked and massaged I wouldn't dream of using anything but Un*x. But when it is done, and the users need access to it, it wouldn't put it on anything but VMS. (sorry, Un*x fanatics, for the novice unsophisticated user how wants to use the program/enviornment to find answers not questions, VMS is easier to teach and master - try convincing an MD that ls makes more sense than dir...and yes I know that you can alias things about but try explaining NO version numbers compared to explaining versions and them safeguards that they provide.) And for large simulations I always use FORTRAN and the VMS machines. It only makes sense, that is what FORTRAN and VMS were designed for. Anything else is folly. Of course, there is always some overlap, and in that case things are going to get kludgy, REGARDLESS of which system/language that you use... The bottom line: VMS is a solid, reliable system whose innards are convoluted (but so are our guts) which is a general purpose machine perfectly adapted for getting things done. Un*x is a quirky, mostly reliable, system whose innards are fairly straight forward (and a mere mortal CAN aspire to understand it all) which is perfect to create an environment conducive to create the things which can be used to get things done. Neither is 'better' - but both are better when used together... AND any fool who says one is 'BETTER' than the other is no professional but simply a fool.... ( and heaven help the fool how tries to do really high-speed data acquisition and control with either.....) -ed cetron%utah-cbd@utah-cs.arpa _________________________________________________________________ I think the above says it all, so lets let the fanatics argue over in talk.bizarre and let the rest of us do our jobs......