ram@elmgate.UUCP (Randy Martens) (01/06/88)
Well folks, it's time for me to insert my 2 pfennigs worth into this discussion. For your reference, I have been programming for ten years, much of it in 'C', and much of it on UNIX or UNIX-like systems. I think GNU is a wonderful idea. It will be an excellent system for educators and for educational institutions, it will serve a a marvelous tool for students to learn about programming and operating systems, and it will be a real boon to the home users who want a decent UNIX-like system to work with. Personally, I really enjoy working with UNIX (on alternate days I despise UNIX, but that's only when i am in "debugging" mode). I would love a copy of GNU that would run on say, an Amiga or a Mac. It would be a great environment to play in. From a professional standpoint, I wouldn't touch GNU with a 10 metre LAN cable. Why ? Let me explain by way of example. I recently bought a new car. I paid $12K of my own hard earned money for this car, a Pymouth Reliant. Within two months of my taking possession of the car, it began to develop some severe but very intermittant performance problems. I returned to the dealer, and the car was checked over. They could not find the source of the difficulty, but decided to repair and tune to see if they could isolate the problem. No luck. Two weeks later I was back again. This cycle repeated itself several times. During this period, I wish to say, that the dealership made every effort to accomodate me (ie, loaner cars, etc). They acknowledged that there was a problem, and that they could not find it. At one point, the manager of the dealership proposed that if they couldn't fix it this time, that they would take the car back, and give me a new one. As luck would have it, they found the problem this time. It was a very subtle sort of bug, involving the interaction of many components of the powertrain. The reason that they were able to find the problem - documentation. You see, there are literally thousands of Reliants on the road. A small percentage of them developed this particular problem. All these cars turned out to have certain features in common. Therfore, the engineers at the corporate level were able to isolate the cause of the problem, and issue what amounted to a "patch" to the dealerships and they were finally to fix the problem once and for all. A year later, the bug seems to have gone away for good. In case you haven't guessed, the problem turned out to be a bug in the software for the Engine Control Computer. The point of the example is this - I purchased what amounted to a standard system - one supported by a large company. Because of this level of support, standardization, and documentation, as well as the means to disseminate data to the suppliers and users, my investment was protected. I still have a good, usable car. Now, lets look at GNU : - It will not be backed by any company. Admittedly, some companies do not do a good job of supporting their users, especially in the software arena. If a company is to survive and prosper, it must give good service, no matter if it is selling software, shoe polish, or cars. And giving good service costs money. Which means the company must make money. Which means it CANNOT possibly function on the basis of giving it's product away. Companies that provide good service will get repeat business, and develop customer loyalty. My experience with my car was not happy, and I certainly wish that it had not developed in the first place, but you can be sure that the next time i am in need of a car, the first place I will go is to that Plymouth dealership ! On the other hand, becuase of some very negative experiences several years ago, I will never buy or use a Digitial Reasearch Co. product ever again. - There will be no standard version. The method of distribution basically guarantees this. And some versions will likely be wildly incompatible with others. The automotive analogy applies again. When buying a car, one has factory supplied choices: model, color, engine, stereo, power widgets, etc. These are all tested and guaranteed by the factory. If my air-conditioner breaks, the dealership will fix it. There are also factory approved modifications supplied by third party vendors, such as performance products. These have also been tested by the factory and the vendor, and are guaranteed to work with my car's configuration. A good example of this would be an after-market turbocharger. Third, there are options that do not seriously affect the vehicle, such as aftermarket stereos - which are still guaranteed by the manufacturer of the stereo. Therfore, there are a lot of different configurations available, to suit my needs. All of these standard and modified versions are akin to a software house supplying optional features with it's systems, allowing approved third party vendors to write tools for their systems, and lastly my going out and buying a copy of "Zork" for my machine. The worst that can happen in the last case, is that I trash my copy of Zork, and get a new one from Infocom. But the important point is this : ALL OF THESE ARE SUPPORTED SYSTEMS !! Gnu will have so many customized versions floating about, that it will be nigh impossible to tell if a given feature or tool will work with my system correctly. And, no, I don't wan't to have to debug my operating system every time I add a new feature. I'd never get any real work done. - The real killer is that if my version does something unpleasent to say an important database, I am on my own. And the loss of data, or the loss of use of a computer system can cost you MUCH MORE that the price of a new system. The support has got to be there, and has got to be ironclad. And yes, I will gladly pay for such support, either by buying a software licence, or by paying support fees. Mr. Stallman says that companies will step in to provide support for GNU. I don't see this happening. And even if some company markets "Fred's Gnu - with support services", the I am in the same position as if I had bought any other operating system, and Gnu's advantages have been erased. - Lastly, I will generally not care if I get the system sources for a piece of code or not. If it works as advertised, and is supported, the damn thing can be written in Bulgarian for all I care. If I am developing tools for an opsys, then I had better have sources, and they better be standardized, or my efforts will be wasted. And if I need to, I will gladly pay reasonable source fees. Yes, I know alot of companies have unreasonable licensing/source fee/site fee policies. But this is changing rapidly. I find that now I can deal only with the good companies, like Borland, who have reasonable policies. And the companies that don't, lose out. Well, thats my ramble on the subject of GNU and operating systems in general. My apologies if this is a bit long, but I wanted to make my point clear : - From a professional standpoint, as someone who makes their living working with computers, I cannot afford to use anything except tested, supported, standardized products. Period. Anything else is foolish. - From a personal standpoint, I look forward to hacking around with GNU, if for no other reason that to learn from it. And that, I think, will be it's greatest value. I will gladly respond to any comment by e-mail or on the net as appropriate. send comments or flames to the addresses below. 'nuff said. -- }} identity } Randy Martens ( rochester!kodak!elmgate!ram ) }} quote } "Reality - What a Concept!" - R.Williams }} disclaimer } The preceeding represents only my random babbling, }} } and certainly reflects no one else's opinions. Fnord.
earle@jplopto.uucp (Greg Earle) (01/06/88)
Hmmn, gee, let's take Randy Merten's article, and everywhere you see `GNU' substitute `4.3BSD', and everywhere you see `Richard Stallman' substitute `UC Berkeley CSRG'. Then re-read the article, and whisper the words `Gosh, no support and all the source to muck with sure didn't stop lots of people from using 4.3BSD' ... - Greg Greg Earle earle@jplopto.JPL.NASA.GOV Indep. Sun consultant earle%jplopto@jpl-elroy.ARPA [aka:] Rockwell Science Center earle%jplopto@elroy.JPL.NASA.GOV Thousand Oaks, CA ...!cit-vax!elroy!smeagol!jplopto!earle
ritzenth@bgsuvax.UUCP (Phil Ritzenthaler) (01/06/88)
In article <5166@elroy.Jpl.Nasa.Gov>, earle@jplopto.uucp (Greg Earle) writes: > Hmmn, gee, let's take Randy Merten's article, and everywhere you see > `GNU' substitute `4.3BSD', and everywhere you see `Richard Stallman' > substitute `UC Berkeley CSRG'. Then re-read the article, and whisper > the words `Gosh, no support and all the source to muck with sure didn't > stop lots of people from using 4.3BSD' ... Encore, encore!! Give that man a cee-gar! Phil Ritzenthaler |USnail: University Computer Services Computer Graphics Research Consultant | 241 Math-Science Bldg. UUCP :.!cbosgd!osu-cis!bgsuvax!ritzenth | Bowling Green State University CSNET: ritzenth@bgsu.edu | Bowling Green, OH 43403-0125 ARPA : ritzenth%bgsu.edu@relay.cs.net | Phone: (419) 372-2102
keith@nih-csl.UUCP (keith gorlen) (01/07/88)
In article <850@elmgate.UUCP>, ram@elmgate.UUCP (Randy Martens) writes: > > Well folks, it's time for me to insert my 2 pfennigs worth into this > discussion. For your reference, I have been programming for ten years, > much of it in 'C', and much of it on UNIX or UNIX-like systems. > From a professional standpoint, I wouldn't touch GNU with a 10 metre LAN > cable. > > Why ? Let me explain by way of example. > I recently bought a new car. [...questionable analogy between cars and software follows...] > Now, lets look at GNU : Yes, let's take a look at GNU EMACS, the part of GNU that's been around the longest. > - It will not be backed by any company. Admittedly, some companies > do not do a good job of supporting their users, especially in the software > arena. If a company is to survive and prosper, it must give good service, > no matter if it is selling software, shoe polish, or cars. And giving > good service costs money. Which means the company must make money. Which > means it CANNOT possibly function on the basis of giving it's product away. GNU EMACS has been ported to different machines and marketed with support by one or two companies that I've heard of. I think that, at least outside the PC mass market, adequate software support cannot be financed by software sales -- customers must pay specifically for software service, and that is how a company makes some of it's profit. Free software, like GNU EMACS, also effectively gets backed by "the public". It is good and it is free, so lots of people use it. Since lots of people use it, it is well-tested. If someone finds a bug, he can use the net to see if someone has a fix. If someone fixes a bug, he has an incentive to send it to the FSF so it gets incorporated into future releases and he won't have to install it again. Everyone benefits. It works because there is a network and a central organization willing to act as a coordinator and clearinghouse. You can see the same phenomenon happening with X Windows. Prior to switching to GNU EMACS, I had purchased two commercial versions of EMACS, and I also paid for support service for one of them. There was absolutely NO CONTEST as to the quality of the software. On one commercial version such features as mail, directory edits, and subshells never worked. On the other, regular expressions and the extension language didn't work. The company wouldn't/couldn't fix the problem, even though I had a "support" agreement. GNU EMACS, on the other hand, has worked almost flawlessly for us from the beginning. > - There will be no standard version. The version distributed by FSF is the standard. Users are discouraged from modifying it because (1) it does practically everything already, and (2) local modifications create a maintenance headache. > Mr. Stallman says that companies will step in > to provide support for GNU. I don't see this happening. I have seen this begin to happen with GNU EMACS. > And even if some > company markets "Fred's Gnu - with support services", the I am in the same > position as if I had bought any other operating system, and Gnu's advantages > have been erased. Not so. In a pinch, you have source code. You can fix the problem yourself, or hire a consultant to do it. One project of mine almost died because the outfit that sold me a compiler couldn't make it work (I had paid for a support contract), and I had no recourse. > - Lastly, I will generally not care if I get the system sources for a piece > of code or not. If it works as advertised, and is supported, the damn thing > can be written in Bulgarian for all I care. Those are big "ifs" in my experience. -- Keith Gorlen phone: (301) 496-5363 Building 12A, Room 2017 uucp: uunet!ncifcrf.gov!nih-csl!keith National Institutes of Health Internet: keith%nih-csl@ncifcrf.gov Bethesda, MD 20892
mwm@eris.BERKELEY.EDU (Mike (My watch has windows) Meyer) (01/07/88)
[Ok, I give up. I'm going to reply....]
In article <850@elmgate.UUCP> ram@elmgate.UUCP (Randy Martens) writes:
<- From a professional standpoint, as someone who makes their living
<working with computers, I cannot afford to use anything except tested,
<supported, standardized products. Period. Anything else is foolish.
And you'll be able to get that from GNU. Just like you can buy a
tested, supported and standardized Unix for the IBM PC, even though
neither AT&T nor IBM sell such a creature. Just like you could buy
4BSD for a VAX before DEC sold Ultrix, and you can buy SysV for a VAX
after AT&T has dropped support for the VAX.
Companies have already formed that sell either SysV or 4BSD for
hardware that it wasn't otherwise available for. They make money
selling *support*, not the OS (a years costs about what the OS does,
and buying the OS comes with a non-trivial fraction of a years
support). They provide real support (I've used it), ranging from
reading manuals over the phone to fixing kernel problems. They've even
supplied source patches I've handed them, though they couldn't see why
it would make a difference.
Being able to start a company without having to spend the cost of 10
to 20 workstations to get a license from AT&T can only cause more of
them to form. New hardware companies get the same benefit.
You don't see those companies forming. I do. I've been aproached about
joining startups that plan on supporting or porting GNU as soon as it
becomes available.
The key question is the standardization of the product. If GNU is
either SVID or Posix ("The nice thing about standards is that there
are so many to choose from."), it won't be any worse than "real" Unix.
I expect it be one of the two. That GNU Emacs & X have avoided turning
into a multiplicity of variants causes me to expect the GNU OS to do
the same. All it takes is a central authority willing to work with
people who would otherwise produce variants.
<mike
--
Tell me how d'you get to be Mike Meyer
As beautiful as that? mwm@berkeley.edu
How did you get your mind ucbvax!mwm
To tilt like your hat? mwm@ucbjade.BITNET
rob@nbires.nbi.com (Robert F. Prentice) (01/09/88)
In article <296@nih-csl.UUCP>, keith@nih-csl.UUCP (keith gorlen) writes: > GNU EMACS has been ported to different machines and marketed with > support by one or two companies that I've heard of. I think that, at A good example is A/UX for the Apple Macintosh: it comes with complete EMACS source. -- Rob Prentice {pyramid, ucbvax, uunet, rutgers!hao, colo!boulder}!nbires!rob These views are of course my own. If they have assisted you, then I am glad. May your day be filled with harmony, and your heart open to the Light.
peter@thirdi.UUCP (Peter Rowell) (01/10/88)
I believe Randy Martens' point of support was not that *any product* which is supported is desirable from a commercial viewpoint, but more specifically that a supported product based on a *recognized standard* is desirable. His analogy about the car made the point that the manufacturer created a standard by which to measure all products meant for that particular model. Although it is true that any random company can aspire to create an industry standard, the reality is that very few of them have the credibility to actually have other companies (including their competitors) acknowledge it as a standard. (E.g. Sun's NFS, e.g. AT&T System V). It is one thing for Stallman to say, "Hey! Here's this free software!", and quite another thing for AT&T or Sun or IBM to say, "Here is what we are supplying to 10,000 (or 1,000,000) customers. If you would like to sell into this customer base, you should consider using our standard." Customers (and developers) have a lot(!) to do with whether a standard actually takes hold. If people will not build or buy products which hove to a given "standard", then the "standard" is either a joke or a hobby - it is not something you base your company's future products on. IF GNU ends up being POSIX conforming AND IF a company selects it AND IF that company supports their customer base THEN it may get recognition - not because it was free, not because you can get the source, but because it met a recognized standard and was supported. Flame away, gentlepeople. Peter Rowell Third Eye Software, Inc. "You will need a larger rock." ...!pyramid!thirdi!peter (415) 321-0967
zrm@eddie.MIT.EDU (Zigurd R. Mednieks) (01/11/88)
In article <302@thirdi.UUCP> peter@thirdi.UUCP (Peter Rowell) writes: > >It is one thing for Stallman to say, "Hey! Here's this free software!", >and quite another thing for AT&T or Sun or IBM to say, "Here is what we >are supplying to 10,000 (or 1,000,000) customers. If you would like >to sell into this customer base, you should consider using our standard." > Peter Rowell has put his finger right on the issue. The Free Software Foundation is nothing more radical than a patent pool or other collaborative R&D effort. If GNU's standards are adopted then GNU will be a success. Thusfar, Richard Stallman's organization has shown they can produce product, which is rather more than many celebrated and lavishly funded efforts in this industry can claim. What is missing is a demonstration that Free Software can work with a manufacturer to bring a GNU-based product to market. Now I may be unaware of agreements that have yet to bear fruit, but the impression I get is that Richard Stallman has difficulty asking for money from the very people he castigates for creating problems in the software world, and that the people that fund collaborative research efforts are put off by some of Free Software's dogma. Yes, I do mean dogma. Would Richard compromise his policies, particularly if some marketing department functionary who stands far below him intellectually, was asking for compromise? I would bet against it. Would he let a board of industry stuffed shirts oversee his operation? It would stick in his craw. So I am not too optimistic about GNU. On the other hand I am all for it. The world could really use a "free" (apart from what it costs to support Free Software) Unix-compatible but rather more advanced OS. Would a manufacturer, say NBI or some other workstation company looking for a way to say, loud and clear, "Look, this is better than Sun!" be willing to fund FSF, stand clear, and hope to benefit from the result? I hope so, the Unix world is getting boring. -Zigurd -- ------------------------------------------------------------------------ Zigurd Mednieks MURSU Corporation (617)424-0146 25 Exeter Street Boston, MA 02116
djs@actnyc.UUCP (Dave Seward) (01/13/88)
In article <5166@elroy.Jpl.Nasa.Gov> earle@mahendo.JPL.NASA.GOV (Greg Earle) writes: >Hmmn, gee, let's take Randy Merten's article, and everywhere you see >`GNU' substitute `4.3BSD', and everywhere you see `Richard Stallman' >substitute `UC Berkeley CSRG'. Then re-read the article, and whisper >the words `Gosh, no support and all the source to muck with sure didn't >stop lots of people from using 4.3BSD' ... > > - Greg > But it did stop *lots* of others from getting involved...
andys@genesis.ATT.COM (a.b.sherman) (01/14/88)
In article <643@actnyc.UUCP> djs@actnyc.UUCP (Dave Seward) writes: >In article <5166@elroy.Jpl.Nasa.Gov> earle@mahendo.JPL.NASA.GOV (Greg Earle) writes: >>Hmmn, gee, let's take Randy Merten's article, and everywhere you see >>`GNU' substitute `4.3BSD', and everywhere you see `Richard Stallman' >>substitute `UC Berkeley CSRG'. Then re-read the article, and whisper >>the words `Gosh, no support and all the source to muck with sure didn't >>stop lots of people from using 4.3BSD' ... >> >> - Greg >> >But it did stop *lots* of others from getting involved... You might also consider why SUN went to so much trouble to merge the 4.3BSD features into System V and to pursuade (coerce) AT&T into taking back the merged product. It was precisely the "user written - not supported" nature of BSD that they wanted to get away from. That, and having to support more than one operating system. -- andy sherman / at&t bell laboratories (medical diagnostic systems) room 2h-097 / 480 red hill road / middletown, nj 07748 (201) 615-5708 / andys@shlepper.ATT.COM ...The views and opinions are my own. Who else would want them?
chip@ateng.UUCP (Chip Salzenberg) (01/15/88)
In article <425@genesis.ATT.COM> andys@shlepper.ATT.COM (a.b.sherman) writes: >You might also consider why SUN went to so much trouble to merge the >4.3BSD features into System V and to pursuade (coerce) AT&T into >taking back the merged product. It was precisely the "user written >- not supported" nature of BSD that they wanted to get away from. >That, and having to support more than one operating system. You've got to be kidding. Sun couldn't "coerce" AT&T with a nuke. Sun doesn't care about the "unsupported" nature of BSD Unix -- with all of Sun's proprietary work, SunOS is hardly BSD anymore. What Sun wants is the marketing appeal of "System V - From Now On, Consider It [Sub-]Standard". Pyramid has done well with OSx, their dual port of BSD and System V. Sun is doing one better; they're getting AT&T to take the result back, thus making the combined OS a new standard. "The nice thing about standards is that there are so many to choose from." -- Chip Salzenberg UUCP: "{codas,uunet}!ateng!chip" A T Engineering My employer's opinions are a trade secret. "Anything that works is better than anything that doesn't." -- me
zrm@eddie.MIT.EDU (Zigurd R. Mednieks) (01/17/88)
In article <151@ateng.UUCP> chip@ateng.UUCP (Chip Salzenberg) writes: > >You've got to be kidding. Sun couldn't "coerce" AT&T with a nuke. > Sun is the Big Gorrilla is Unix, folks. Whatever they do is a de-facto standard, and AT&T is, among other things, accepting that standard. Sun may not have "coerced" AT&T into unsing their software, using their hardware architecture, and blessing it all as standard, but market reality has. It is really amazing that at practically every twist and turn in the Unix market Sun has made the right decision and AT&T has made the wrong decision. Now if they can get all the software vendors to dance to the same user interface tune, watch out! -Zigurd -- ------------------------------------------------------------------------ Zigurd Mednieks MURSU Corporation (617)424-0146 25 Exeter Street Boston, MA 02116