dmr@dutoit.UUCP (03/01/86)
Herb Chong's delayed article propagates wrong history. Twice he quotes someone from Purdue who claims "Are you aware that Dennis Ritchie once said that if he had known about Tenex, he would never have invented Unix?" For the record, I'm not aware of saying that, or anything like it. First off, I've always been at pains to point out that Ken invented Unix, though I'm certainly pleased with my contributions. The only observation about Tenex I can remember making publicly was along these lines: we were very lucky not to have gotten a PDP10 to write a system for, because then very early Unix would have had to compete not only with DEC's operating system, but also with Tenex, for the small PDP10 market, and both TOPS-10 and Tenex were pretty decent systems. As it actually happened, there were lots and lots of PDP11s, the DEC software was ghastly, and so many groups were willing to risk trying Unix. As to the history in general, I think my BSTJ account is a little more authoritative than Bourne's, though his is not seriously misleading despite some mistakes (for example, the first PDP7 Unix was a time-sharing system; it supported two users). Herb also muses "... he [DMR] would rather forget that he invented Unix, despite its success. I have heard that he once said that he feels like someone who started a religion that he now sees all the flaws in, but no one else seems to want to listen. He feels caught up in something he no longer believes in." Perhaps someone caught me in a wry mood in which I muttered something about religious fanatics. I do try to be honest with myself and others about flaws, limitations, and failures of Unix to reach utter perfection and universality. However, to put matters as modestly as possible: I do not hold the feelings ascribed to me in the quoted paragraph. As to the more general comments in Chong's article: they can be attacked and defended in various ways; how it comes out depends as much on what one wants from an operating system as anything else. There are things on which I would comment: he says, "Unix was hacked together to do something until they had something else to do it right. That right thing never came along and so more and more got added to Unix... Unix is uniformly mediocre. It uses the lowest common denominator between a lot of different types of machines. In doing so, it doesn't try to do too much and it succeeds well at not doing too much." I assure you that Unix was not designed to be thrown away when something better came along. Rather, it exhibits a strong, coherent and manifestly successful set of beliefs about how to construct and furnish a certain kind of computing environment. It is not uniformly mediocre: it is absolutely excellent at providing interactive computing for program development, scientific computing, text processing and the like, and perfectly horrible for DP by banks and insurance companies, or transaction processing by airlines. Finally, I would agree completely with the last quoted sentence if it said, "it succeeds well by not doing too much." Dennis Ritchie
herbie@polaris.UUCP (Herb Chong) (03/03/86)
In article <2102@dutoit.UUCP> dmr@dutoit.UUCP writes: >Herb Chong's delayed article propagates wrong history. Twice he quotes >someone from Purdue who claims "Are you aware that Dennis Ritchie once >said that if he had known about Tenex, he would never have invented Unix?" >For the record, I'm not aware of saying that, or anything like it. in the absence of tapes of everything anyone has ever said about unix, i have to assume that someone has some idea what they're talking about without having to bring one of the originators to testify. in otherwords, since unix was invented when i was still in highschool and didn't know what FORTRAN meant, i had to believe somebody. 8-) >First off, I've always been at pains to point out that Ken invented and didn't know what FORTRAN meant i have to believe someone else. 8-) yes, i am aware that you are credited with inventing C, a totally different thing. >The only >observation about Tenex I can remember making publicly was along these >lines: we were very lucky not to have gotten a PDP10 to write a system >for, because then very early Unix would have had to compete not only >with DEC's operating system, but also with Tenex, for the small PDP10 >market, and both TOPS-10 and Tenex were pretty decent systems. As it >actually happened, there were lots and lots of PDP11s, the DEC >software was ghastly, and so many groups were willing to risk trying >Unix. i guess this is going to end up in one form or another as an anecdote from the dark ages of computing garbled to a greater or lesser degree by comments like mine a 30 years from now. of course, i'd rather be remembered for the wrong thing than not be remembered at all! 8-) >Herb also muses "... he [DMR] would rather forget that he invented >Unix, despite its success. I have heard that he once said that he >feels like someone who started a religion that he now sees all the flaws >in, but no one else seems to want to listen. He feels caught up in >something he no longer believes in." >Perhaps someone caught me in a wry mood in which I muttered something >about religious fanatics. I do try to be honest with myself and others >about flaws, limitations, and failures of Unix to reach utter >perfection and universality. However, to put matters as modestly as >possible: I do not hold the feelings ascribed to me in the quoted >paragraph. this was heard second hand from someone who attended a talk given at IBM T. J. Watson Research Center (where i work) that i did not go to. i forget who actually claimed to have heard you say this or whether in fact the talker had said he had heard it from you directly or not. i think Brian Kernighan was the speaker. that is why i worded it the way i did since i never heard it myself. as it is, i think it may have been appropriate for the audience, most who have only heard of unix as an operating system for lunatics and other fringe types running DEC (heaven forbid!) equipment. 8-) >I assure you that Unix was not designed to be thrown away when >something better came along. Rather, it exhibits a strong, coherent >and manifestly successful set of beliefs about how to construct and >furnish a certain kind of computing environment. It is not uniformly >mediocre: it is absolutely excellent at providing interactive computing >for program development, scientific computing, text processing and the >like, and perfectly horrible for DP by banks and insurance companies, >or transaction processing by airlines. which is all totally true, but you'd never know it from the marketing hype in such publications as Unix World. a lot of articles come by in this news group by people who think that unix is the be-all and end-all operating system. it isn't. it has a niche which it fills very well thank you. as someone who designs and implements operating systems for a living, i am well aware of the limitations of unix as well as it's advantages. the uniformity in the treatment of files as virtual disk devices is a great advantage from the programmers' point of view. the fragility of the filesystem implementation is not so great (having recently patched some filesystems where the free-list was being scribbled on by the hardware). the documentation is just horrible, especially for a new user. an operating system that is going to be used is much more than just an elegant design and careful implementation. it's also the supporting tools (which need not be programs and/or libraries of functions). perhaps when unix was still a small enough to be printed on a few hundred pages (for all of it), the documentation was adequate. today, it is mostly inadequate. word-of-mouth is more trusted. supplying source is not an answer either. the machine i am posting this from has an operating system that source cannot be obtained for. so much of the documentation assumes, no make that requires, source to translate what the english says in to something meaningful. >Finally, I would agree completely with the last quoted sentence if >it said, "it succeeds well by not doing too much." the flip side of this is for the people who don't want to build the medium level tools for applications. yes, it is essential that an operating system provide enough flexibility to do pretty much what the user wants without contortions of the worst kind in coding, but there is also providing enough medium level support to do things without re-inventing the wheel all the time. the main example i'm thinking of is file-locking, but the sys5 people will say they've had it for a long time. i realize that locking is not a trivial thing and i don't pretend to have the answers. in summary, since i'm too young to have actually participated in that bit of history, i can only quote from people who claim to know what they're talking about until i find out otherwise. the other thing is that i hate people who won't open up their minds enough to try something different, to study it and compare. the unix users community has a lot of tunnel vision but it is not the only one with it. Herb Chong... I'm still user-friendly -- I don't byte, I nybble.... VNET,BITNET,NETNORTH,EARN: HERBIE AT YKTVMH UUCP: {allegra|cbosgd|cmcl2|decvax|ihnp4|seismo}!philabs!polaris!herbie CSNET: herbie.yktvmh@ibm-sj.csnet ARPA: herbie.yktvmh.ibm-sj.csnet@csnet-relay.arpa, herbie%yktvmh.bitnet@wiscvm ======================================================================== DISCLAIMER: what you just read was produced by pouring lukewarm tea for 42 seconds onto 9 people chained to 6 Ouiji boards.
seifert@hammer.UUCP (Snoopy) (03/06/86)
In article <464@polaris.UUCP> herbie@polaris.UUCP (Herb Chong) writes: > the fragility of the filesystem implementation is not so great (having > recently patched some filesystems where the free-list was being > scribbled on by the hardware). The hardware screws up and scribbles on the disk and you're complaining about the software? Perhaps you should be looking at replicated hardware. >the documentation is just horrible, especially for a new user. Lots of people complain about this. I must be strange. I didn't have any problem with the documentation when I learned Unix. Snoopy tektronix!tekecs!doghouse.TEK!snoopy
geoff@desint.UUCP (Geoff Kuenning) (03/08/86)
In article <1843@hammer.UUCP> tekecs!doghouse.TEK!snoopy (Snoopy) writes: > The hardware screws up and scribbles on the disk and you're complaining > about the software? Hate to introduce you to reliable systems, but yes. A truly robust filesystem isn't bothered by little details like flaky hardware, even if you can't afford to replicate it. -- Geoff Kuenning {hplabs,ihnp4}!trwrb!desint!geoff
ka@hropus.UUCP (Kenneth Almquist) (03/11/86)
> > The hardware screws up and scribbles on the disk and you're complaining > > about the software? [Snoopy] > > Hate to introduce you to reliable systems, but yes. A truly robust > filesystem isn't bothered by little details like flaky hardware, even if > you can't afford to replicate it. [Geoff Kuenning] This is a rather brief introduction; perhaps you could expand on it? If the disk drive doesn't write correctly, you've lost the information written unless you have replicated it; am I right? Kenneth Almquist ihnp4!houxm!hropus!ka (official name) ihnp4!opus!ka (shorter path)
aglew@ccvaxa.UUCP (03/13/86)
>/* Written 3:43 pm Mar 5, 1986 by seifert@hammer.UUCP */ >In article <464@polaris.UUCP> herbie@polaris.UUCP (Herb Chong) writes: >> the fragility of the filesystem implementation is not so great (having >> recently patched some filesystems where the free-list was being >> scribbled on by the hardware). >The hardware screws up and scribbles on the disk and you're complaining >about the software? Perhaps you should be looking at replicated >hardware. Don't be so quick to lay all the onus for reliable filesystems on hardware. Xerox had a system where both cooperated to make a more robust system: disk blocks, in addition to being marked with track number and sector number, were marked with file (inode) number, and logical sector number in that file. Since these marks were part of the block header and not the data, they were less likely to be scribbled in error. Unfortunately, nobody except Xerox makes home-brew disk controllers. Sigh.
seifert@hammer.UUCP (Snoopy) (03/14/86)
In article <173@desint.UUCP> geoff@desint.UUCP (Geoff Kuenning) writes: >> The hardware screws up and scribbles on the disk and you're complaining >> about the software? > >Hate to introduce you to reliable systems, but yes. A truly robust >filesystem isn't bothered by little details like flaky hardware, even if >you can't afford to replicate it. Sorry, I disagree. What do you do when the drive decides to write stuff all over the disk? What do you do when you get a head crash? There are all sorts of things the hardware can do that a simplex system just can't protect from. Sure there are things that you could do that would help a little, but don't kid yourself about immunity to hardware failures. Snoopy tektronix!tekecs!doghouse.TEK!snoopy