mark@cbosgd.UUCP (Mark Horton) (05/09/85)
In article <166@timeinc.UUCP> greenber@timeinc.UUCP (Ross Greenberg) writes: >Escaping to shell (or forking to vi) from >$ ls -lR / | more (or ls -lR / | page) >barfs out. Funny error messages, some endless loops, etc. >Looks like a piping problem. Why would you want to vi the output of the ls command? This is an old bug in more that was fixed by 4.2BSD - it's execing vi with stdin set to the pipe, so the output of ls is being taken by vi as keyboard commands. More now beeps if you try to do something this silly. I'm not sure what pg does in SVR2. >AT&T claim 9 users. Could be. I felt it was *VERY* >slow with one user. More specifically, the disk is slow, compared to the speed of everything else. CPU processes seem quite fast. However, in a disk that small and that inexpensive, you can't expect a lot of speed. Mark
dwight@timeinc.UUCP (Dwight Ernest) (05/10/85)
>...They claim nine users. I felt it was VERY slow with ONE >user. Funny. I had about half and hour's hands-on with about 20 tasks running (I checked a ps) and I felt it was VERY FAST. Window management, however, was slow; I'm talking about running with 'wmgr' present and executing, but measuring response under the Bourne shell. Also, I've used the Korn shell at other sites, and I sure do wish that AT&T might have considered making it standard (instead of the unadorned Bourne 'sh') in the UnixPC. ----------------------------------------------------------------------------- --Dwight Ernest KA2CNN \ Usenet:...vax135!timeinc!dwight Time Inc. Edit./Prod. Tech. Grp., New York City Voice: (212) 554-5061 \ Compuserve: 70210,523 Telemail: DERNEST/TIMECOMDIV/TIMEINC \ MCI: DERNEST "The opinions expressed above are those of the writer and do not necessarily reflect the opinions of Time Incorporated." -----------------------------------------------------------------------------
greenber@timeinc.UUCP (Ross M. Greenberg) (05/11/85)
In article <1160@cbosgd.UUCP> mark@cbosgd.UUCP (Mark Horton) writes: (Quoting me...): >>Escaping to shell (or forking to vi) from >>$ ls -lR / | more (or ls -lR / | page) >>barfs out. Funny error messages, some endless loops, etc. >>Looks like a piping problem. > >Why would you want to vi the output of the ls command? This is an old >bug in more that was fixed by 4.2BSD - it's execing vi with stdin set >to the pipe, so the output of ls is being taken by vi as keyboard commands. >More now beeps if you try to do something this silly. I'm not sure what >pg does in SVR2. > Mark, if you note from above, I talked about *escaping* to the shell or vi. You know, like when more gives that funky --More-- jazz and awaits your input?? I typed a simple "!sh" or a simple "!vi" and was taken out for breakfast, lunch *and* dinner. Page did it too. You're right in that it appeared to be taking some of the stuff in the pipe as input to the new child process. >>AT&T claim 9 users. Could be. I felt it was *VERY* >>slow with one user. > >More specifically, the disk is slow, compared to the speed of everything >else. CPU processes seem quite fast. However, in a disk that small and >that inexpensive, you can't expect a lot of speed. > > Mark Yes, I can expect a faster performing *machine* from AT&T. If it isn't fast enough to do any practical work, then (whatever the reason) the machine just doesn't have the bang for the buck that I was hoping. I mean, lets be serious: $7000 for a machine that requires over a minute for simultaneous logins on two terminals??? Now this machine only had 512K so there was a lot of swapping going on, but why would AT&T send an army (well, about eight people) to a Uni-Group meeting with machines that were configured as such toys *UNLESS* they actually consider that to be what people will buy. I don't want to buy a Super-Duper Disk Unit from Smith's Computer company to make up for problems in the configuration and pricing scheme that AT&T choose. Whew! -- ------------------------------------------------------------------ Ross M. Greenberg @ Time Inc, New York --------->{ihnp4 | vax135}!timeinc!greenber<--------- ^^^^^^^^^--->New and improved, with added zing! "If ever the pleasure of one has to be bought by the pain of the other, there better be no trade. A trade by which one gains and the other loses is a fraud." --- Dagny Taggert
roy@phri.UUCP (Roy Smith) (05/12/85)
> Now this machine only had 512K [...] with machines that were > configured as such toys *UNLESS* they actually consider that to be > what people will buy. The AT&T guys were vary careful to point out several times during the presentation and in the Q&A that followed that for the type of use most of the audience would have for the machine (i.e. program development) that you should almost certainly only consider the full-blown 2Mb RAM, 20Mb winnie system, with the optional utilities package (C compiler, etc). There was no hint of trying to deceive the audience that the stripped down version would be satisfactory for the intensive environment you seem to have in mind. BTW, I see nothing wrong with making the C compiler, and other utilities optional as long as you make no secret of it. If you don't want them, you don't have to buy them. If you do, you know what you are getting into and how much the option will cost. My biggest complaint about the 7300 (I only played with it for a few minutes; no serious shakedown, no real effort to learn how to take advantage of its features) is that a large percentage of the screen is wasted on the wide borders around the windows. Screen area is at a premium; why waste so much of it on 3/4" wide borders. My personal opinion is that as a stand alone it is probably worth it, but I'd rather sink a few more $k into a diskless Sun. -- allegra!phri!roy (Roy Smith) System Administrator, Public Health Research Institute
mark@cbosgd.UUCP (Mark Horton) (05/13/85)
The UNIX PC is clearly intended as a single user machine. It runs UNIX, so in theory you can have more than one user. They even provide a way to plug in another user or two. But this is also true of the IBM XT running any of the UNIX ports, yet you would never claim the XT is intended as a multi-user machine. So the time it takes 2 users to log in on a stripped model is hardly an important benchmark. $6K gets you a fully loaded machine. (Plus software, and I'm not sure what that's priced at.) At that price, the UNIX PC has no competition. It's MUCH faster than an XT. Unlike the IBM AT, it has a 12.5MHz processor (about 6 times the benchmark speed of the XT), a full 32 bit machine (well, it's a 68010, it LOOKS like 32 bits), and there are no known reliability problems with the disk. The other products in this market, such as the 3B2, the Fortune, and the Tektronix, will run you $10K or more, and you have to supply your own terminal (which probably won't be bitmapped.) (The HP will run you $9K if you want a hard disk.) (These products do have additional capabilities, such as decent multi-user performance and faster disks, but are often used as single user machines anyway.) And yes, given the choice of a UNIX PC or a Sun, I'd choose a Sun any day. The Sun has a much larger screen, is faster, and speaks TCP/IP/Ethernet. However, the Sun is in a totally different market. You have to spend over $16K (list price) for their most stripped standalone system, plus $4K for a 2nd MB of RAM (4.2BSD is a pig with only 1MB) and another $4K if you want the bitmapped display. You need to have several users to make their fileserver oriented system pay off financially.
dwight@timeinc.UUCP (Dwight Ernest) (05/14/85)
BTW, there is an interesting article about the history of the development of the Unix PC (a.k.a. 7300) in the May issue of _Popular_Computing_. As many of us already knew, the article explains that the Unix PC was not developed in-house, but was rather specified in-house, in a small team effort, and then "farmed out" to an outside systems house (Coherent, I think). The article says that the biggest mistake made was the choice of the 68010 instead of the 32100 chip set, and that this mistake was made mostly due to proprietary marketing considerations which no longer existed when the machine was finally completed. Comments? -- ----------------------------------------------------------------------------- --Dwight Ernest KA2CNN \ Usenet:...vax135!timeinc!dwight Time Inc. Edit./Prod. Tech. Grp., New York City Voice: (212) 554-5061 \ Compuserve: 70210,523 Telemail: DERNEST/TIMECOMDIV/TIMEINC \ MCI: DERNEST "The opinions expressed above are those of the writer and do not necessarily reflect the opinions of Time Incorporated." -----------------------------------------------------------------------------
roy@phri.UUCP (Roy Smith) (05/14/85)
Oh boy! a real flame exchange, how fun! :-) > weren't we expecting a machine that would blow your socks off?? Yeah, I was a bit disappointed. But then again, most of the stuff AT&T has done (let's say since PWB) has been disappointing. > Corporate users [..] need decent response time, and a [..] bug-free system. True; the second point even more so than the first. Us "real programmers" know how to work around bugs, and are willing to do so if the payoff is high enough. Corporate users won't be able to handle unexpected situations as well and will just get discouraged. > It would have been nice to have a machine so powerful and inexpensive > that the biggest problem would have been which of the clones to choose > from. Do you see that happening? No, not with the 7300 I don't. This is not to say that I think there won't be 7300 clones. OEMs make clones not because they think the original product was technically good, but because they think it will become popular (look at all the VT-220 ripoffs). Especially with an item aimed towards a non-technical audience, there is little correlation between the two. -- allegra!phri!roy (Roy Smith) System Administrator, Public Health Research Institute
sdyer@bbnccv.UUCP (Steve Dyer) (05/15/85)
> The article says that the biggest mistake made was the choice of > the 68010 instead of the 32100 chip set, and that this mistake > was made mostly due to proprietary marketing considerations > which no longer existed when the machine was finally completed. Huh? Where's the "mistake" here? A UNIX box is a UNIX box as long as it performs to some minimal degree--to a large extent, you can forget just what chip is sitting underneath (unless it's an 80x86, I guess!) "Hey, this thing looks just like an Onyx," -- /Steve Dyer {decvax,linus,ima,ihnp4}!bbncca!sdyer sdyer@bbnccv.ARPA
greenber@timeinc.UUCP (Ross M. Greenberg) (05/16/85)
In article <395@rna.UUCP> dan@rna.UUCP ( Ts'o) writes: > ... For the price >(which I think you have wrong, my impression is that with standard discounts >it will be in the $4000 range), I think it is not bad, especially considering >an okay window interface and integral phone modem. > > I believe that you were misled at the demo by trying to do >simultaneous logins on a loaded machine. ... > >...is worth considering, especially if (as someone else thought) the machine >might be available to some for $3000. <Sigh...> There's a big difference to how *I* consider the machine for myself and how I consider it as something that AT&T is trying to market to corporate accounts. Now the corporate accouts will not be getting the hefty and spiffy developer discounts that were mentioned at UNI-Group. They *may* get some relatively modest volume discounts, if they buy enough machines in one shot. So I figure that the $6000-$7000 price would probably be what a Fortune 500 would pay for <5 machines. Remember: you have to buy the operating system as an add-on!!!! And, as a note, the machine that I had all my problems with was set up multi-user, with no windowing environment (although I did try that later. I could have had dinner while waiting for the login on the terminal to ask for my password!!) -- ------------------------------------------------------------------ Ross M. Greenberg @ Time Inc, New York --------->{ihnp4 | vax135}!timeinc!greenber<--------- Timeinc probably wouldn't acknowledge my existence, and has opinions of its own. I highly doubt that they would make me their spokesperson. ------ "If ever the pleasure of one has to be bought by the pain of the other, there better be no trade. A trade by which one gains and the other loses is a fraud." --- Dagny Taggart
adam@npois.UUCP (Adam V. Reed) (05/16/85)
The following views are my own, NOT my company's. Aah, Flames.... I have used the UNIX PC for three weeks now, and have found it an excellent machine if used right, with several caveats: 1. The consumer shell (called ua) is an excellent tool for easy performance of all those system adminstration tasks that I never quite became fluent in, like updating uucp control files. I never dreamed uucp administration, or lp queue setup, or port configuration, etc. etc. could be made so easy (even an utter non-programmer can do things on a UNIX PC that require a skilled SA on traditional UNIX systems, and never even notice having done them). But for things that one is fluent in, going through the menus of the ua just slows you down. 2. If you want the 7300 to show its stuff as a UNIX engine, USE IT AS A UNIX ENGINE. Note that you can avoid the ua altogether by logging in as root, or by removing the "exec ua" line from your $HOME/.profile. That is a good idea for anyone who knows how to use an "expert friendly" shell; the ua takes forever to start up (this is why a login takes so long if you leave "exec ua" in your profile), and slows everything down whenever it is running. If you need it, you can always "exec ua" later from your login shell. 3. If you avoid the ua, the UNIX PC can support up to 3 users with reasonable speed. It slows down with more than 3, but that is to be expected: after all, a 68010 is not a 32100. 4. You don't need the ua to have windows. I have implemented a blit-like borderless asynchronous windowing environment with a few simple kernel calls. This needs ksh aliases, though, for full functionality, so I don't plan to post it here unless there is a demand. Incidentally, ksh (as distributed by the AT&T UNIX Toolchest) ports very readily to the UNIX PC. Adam Reed ihnp4!npois!npoiw!adam (npoiw is a UNIX PC)
mash@mips.UUCP (John Mashey) (05/17/85)
Dwight Earnest ...vax135!timeinc!dwight writes: > BTW, there is an interesting article about the history of the > development of the Unix PC (a.k.a. 7300) in the May issue of > _Popular_Computing_. As many of us already knew, the article > explains that the Unix PC was not developed in-house, but was > rather specified in-house, in a small team effort, and then > "farmed out" to an outside systems house (Coherent, I think). > The article says that the biggest mistake made was the choice of > the 68010 instead of the 32100 chip set, and that this mistake > was made mostly due to proprietary marketing considerations > which no longer existed when the machine was finally completed. 1) Coherent ---> Convergent Technologies, of course. 2) The article indeed commented as described on the biggest mistake. However, regardless of whether the magazine heard this from someone inside ATT or simply speculated, the statements are at best an over-simplification. The article said: a) ATT WE manufactured the more powerful WE-32000. b) The WE-3200 was still proprietary, hence could not be used in any product manufactured externally. ATT couldn't release the specs to CT and thus couldn't use it. c) [BY IMPLICATION] ATT & CT would have preferred to use the WE-32000, but bureaucracy got in the way. a) ATT WE manufactured the WE-32000. It is not instantly clear that it is more powerful in this application, although this might be answered by running CPU benchmarks on a 3B2 vs a 7300 (or CT MiniFrame). [Note that the WE32000 of 3B2 & 3B5 and WE32100 are different chips]. b) The WE-32000 was still proprietary. As I recall, as part of the process of readying a bond offering last fall, that it was possible that some proprietary parts available only from ATT might be in short supply. This statement caused numnerous people to worry that CT wouldn't be getting enough 32000's. ALL of this was nonsense - there is of course a proprietary ATT part in the 7300, but that part NEVER was a 32000. c) It is not instantly clear that ATT and CT preferred the 32000. ATT first: it may be that somebody may have preferred to put the 32000 in this product. Remember that ATT IS (with whom CT worked) and ATT Technologies (who make 32000's) were fairly separate organizations, with (necessarily) different viewpoints, priorities, and goals. I'd speculate that there were some interesting discussions.... CT next: by the time this deal really got going, CT already had a demand-paged System V running in a system (MiniFrame) well-designed for low-cost manufacturing, using a 10Mhz 68010. Important considerations in this market are manufacturing cost and time-to-market. I don't know cost and availability figures on 32000's and supporting chips; I speculate that they might not necessarily be cheaper than the 68010s that CT had much experience with. As a fact, the 7300 started life as a MiniFrame variant, and if there ever was serious consideration inside CT of wanting to use a 32000, it died out very quickly. On a separate note related to the 7300, it seems that ATT can't win. First, people complained that ATT was too UNIX-conscious and was ignoring personal computers. Next, when the 6300 came out, people complained that it was boringly standard, wasn't exciting, and didn't even run UNIX. Finally, the 7300 comes out, combining many good attributes of UNIX + Mac, and people complain that it doesn't run MSDOS. Sigh... In my opinion, the 7300 is a great little box, with some good hardware and software engineering to get the cost/performance that it does. -- -john mashey [ex CT, ex BTL] UUCP: {decvax,ucbvax,ihnp4}!decwrl!mips!mash DDD: 415-960-1200 USPS: MIPS Computer Systems, 1330 Charleston Rd, Mtn View, CA 94043
mark@cbosgd.UUCP (Mark Horton) (05/20/85)
>> The article says that the biggest mistake made was the choice of >> the 68010 instead of the 32100 chip set, and that this mistake >> was made mostly due to proprietary marketing considerations >> which no longer existed when the machine was finally completed. In fact, the 68020-32100 choice (in general, not specifically for the 7300) is a somewhat controversial political decision. On technical grounds, the two aren't that different in practice, since we do all the coding in C anyway. The 32100 is a full 32 bit chip, as was the 32000, and the then-available 68000 wasn't quite 32 bits. But the 68K also costs less than the 32K, and was available sooner internally. So many prototypes were built with the 68K, like the Blit, and later turned into products like the 5620 containing a 32000. The decision to use the 68K in the 7300 seems to have been made by Convergent Technologies, not AT&T, and they were dealing with a part of AT&T IS that was actually willing to consider selling a product was not manufactured from the wire up by AT&T (this is almost unheard of, for those of you on the outside), so they were less constrained by politics. The decision was apparently made nearly 2 years ago, when AT&T had only the 32000, and it was in short supply. If they had it to do over again, the 32100 might make sense (but I am not familiar with the cost/performance/etc issues comparing the 32100 and the 68020.)
thompson@oberon.UUCP (mark thompson) (05/21/85)
> I tend to disagree with you. I feel that a system that is buggy and > as slow as this system was/is should not be put out by AT&T. Lets be > honest: weren't we expecting a machine that would blow your socks > off?? I mean, with AT&T coming out with a UNIX machine !!! > > Corporate users of the 7300 still need decent response time, and a > (as much as possible) bug-free system. > > ------------------------------------------------------------------ > Ross M. Greenberg @ Time Inc, New York > --------->{ihnp4 | vax135}!timeinc!greenber<--------- > This sounds like the starting salvo of a fine little war. I assume from the tone of this posting that you were all prepared to hate the machine when you met it, and sure enuf...you did. * Capsule review: Nice package...but not a bargain. * Wins: - CPU is plenty fast, and supports virtual memory. - All of the windowing and graphics is available in the C library. - 'ua', the desktop application, steps you painlessly through the incantations to create users, uucp links, etc. - the phone (of course) is built in, and it can toggle between voice (it dials, you talk) and data. * Losses: - Windowing, and mouse, PALE in comparison to a macintosh (not to mention EXPENSIVE machines). - The disk speed is, wellllll, not blazing fast. - It is UUUUGGGGLLLLYYYY, and has a big footprint. - It doesn't run BSD Unix (surprise). * Query: - Does AT&T plan to release a LAN for this thing (nudge nudge, wink wink)? -mark -- --- Everybody is welcome to my own opinion --- Written: mark thompson arpanet: THOMPSON@USC-ECLC usenet: { ihnp4 | hplabs | akgua | sdcsvax} !sdcrdcf!uscvax!oberon!thompson
george@mnetor.UUCP (George Hart) (05/21/85)
(eat this :-) In <185@timeinc.UUCP>, Ross Greenberg writes: > Now the corporate accouts will not be getting the hefty and spiffy > developer discounts that were mentioned at UNI-Group. They *may* get > some relatively modest volume discounts, if they buy enough machines > in one shot. So I figure that the $6000-$7000 price would probably > be what a Fortune 500 would pay for <5 machines. Remember: you have > to buy the operating system as an add-on!!!! While you'd be correct in saying that most companies don't get OEM type, discounts, those that buy alot, get alot. I've seen large, non-OEM companies get OEM discounts or better, because of quantity. Over and above that, my guess is that AT&T would *love* to get into the Fortune 500 companies and will be very aggressive in their marketing to large companies. If you really think it's too much for what you get, just think what over 4,000,000 purchasers paid (and got) for their IBM PC's & XT's (shudder). -- Regards, George Hart, Computer X Canada Ltd. {cbosgd, decvax, harpo, ihnp4}!utcs!mnetor!george
heiby@cuae2.UUCP (Heiby) (05/27/85)
In article <33@oberon.UUCP> thompson@oberon.UUCP (mark thompson) writes: >* Query: > - Does AT&T plan to release a LAN for this thing (nudge nudge, > wink wink)? AT&T has announced our StarLan (not sure of capitalization, but almost certainly a trademark of AT&T) local area network for the Unix PC and for the PC 6300 (MS-DOS based). See your AT&T account rep for pricing and availability info. -- Ron Heiby heiby@cuae2.UUCP (via wnuxa or wnuxb) AT&T-IS, /app/eng, Lisle, IL (312) 810-6109
dan@digi-g.UUCP (Dan Messinger) (05/30/85)
In article <33@oberon.UUCP> thompson@oberon.UUCP (mark thompson) writes: >* Query: > - Does AT&T plan to release a LAN for this thing (nudge nudge, > wink wink)? Yep. Its called STARLAN.* *The product announcement for STARLAN does NOT declare "STARLAN" to be a trademark of AT&T, although I suspect it is. Dan Messinger ihnp4!umn-cs!digi-g!dan