|
Week of 5/10/99 Sunday, May 16, 1999 09:08 A (mostly) daily journal of the trials, tribulations, and random observations of Robert Bruce Thompson, a writer of computer books. |
|
Monday |
[Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] Monday, May 10, 1999 If you didn't read the updates last weekend, check back to last week. I posted quite a lot of interesting new stuff Saturday and Sunday. I'm building a to-do list for the ebook project that Barbara and I are seriously considering doing. One of the high priority items is registering a domain name for the supporting web site. Actually, it's not that we need the web site immediately so much as that domain names tend to be grabbed by other people if you don't grab them yourself. I'll probably hold off a while just to see what happens with all the new domain name registries. It'll be interesting to see what happens to the price of domain names. InterNIC has had a monopoly for way too long now. Competing registries can charge whatever they want, although they do have to pay InterNIC $9 per domain name registered. The competing registries claim that $9 fee is excessive, and I think they're right. Something in the $1 range might be a reasonable amount for the service that InterNIC will provide them. At any rate, I have two domains now, so this new one will make three. This one, a new domain that is currently "parked" and will host the web site for my hardware books, and now this one for the project that Barbara and I are considering. I have a hard enough time keeping up one web site. I can't imagine what it'll be like to have to maintain three of them. * * * * * This week is not a good time to buy an Intel CPU, but that'll change next week when Intel ships the 550 MHz Pentium III. This newest member of Intel's stable should be available next Monday, and is to cost $744 in quantity. As usual, existing processors will have their prices cut, some dramatically. You'll now be able to choose between the Pentium II/450 and the Pentium III/450 for the same price--$268. I can't imagine that many will choose the Pentium II. In effect, this announcement kills the Pentium II entirely. Some may still opt for the Pentium II/400 (at $193) or the Pentium II/350 (at $163), but these processor's days are obviously numbered. For the next several months at least, Intel will be selling the Celeron to the cost-sensitive market and the Pentium III—which includes SSE—to the performance market. Celeron prices won't change, so these new prices at least put the Pentium III in the race. Given the choice between a $268 Pentium III/450 and a Celeron/466 for about a hundred bucks less, many will opt to pay the extra cost simply to have the latest technology. If Intel can demonstrate compelling advantages to SSE--not in benchmarks but in real-world performance advantages in actual shipping applications--they could make the Pentium III/450 the smart bet. The Pentium III at $600 to $1,000 was simply not a good choice for most users. At $268, it may well be. Several sources have mentioned that the Pentium III/500 runs very hot. If I had a pre-production Pentium III/550 and was allowed to talk about it, I might have found that it does not run hot at all. In fact, if it was installed in an ATX case with proper cooling, I might have found that it stabilized at about 105F (40.6C). * * * * * This from rainer becker [info@rainers-radsport.com]:
Entschuldigen Sie bitte mein schlechtes Deutsches. Die sind in der Tat eindrucksvolle Benchmarkzahlen. Ich benutze nicht AMD-Mikroprozessoren viel, aber jene Benchmarkzahlen scheinen, die des Pentium II zu übersteigen. Glückwünsche auf Ihrem neuen Computer. (Please excuse my poor German. Those are indeed impressive benchmark numbers. I don't use AMD microprocessors much, but those benchmark numbers appear to exceed those of the Pentium II. Congratulations on your new computer.) * * * * * This from Ralph Mosqueda [ralferic@volcano.net]:
The SP2 was last made about 25 years ago, so coming up with a manual for it won't be easy. Your best bet may be to check E-Bay and similar auction sites. I've seen manuals for older cameras on offer there frequently. I've also seen such things on the web sites of large camera stores that sell used equipment, although I don't have any URLs handy. Herbert Keppler published a book called _The Honeywell Pentax Way_ back in the 1970's. One or another edition of that book may cover the SP2. You may be able to find a copy of that book by using one of the search services for used and rare books, such as Advanced Book Exchange, although I just did a quick search and didn't turn anything up there. * * * * * This from Bo Leuf [bo@leuf.com]:
That makes sense, although I'd have assumed a traditional percolator ran at 100C. I mean, I assume that what pushes the water up through the central tube and causes the burbling noise is live steam, so the water itself would have to be at 100C before steam formed, right? Well, at sea-level anyway. I have a thermo-couple probe somewhere. Perhaps I'll suspend it in the basket and see just how hot the water coming out of this Krups is. I can well believe that liability concerns might cause manufacturers to use cooler water. Nowadays, power mowers come with big labels telling you not to put your hands or feet into the path of the blade while the mower is running. Duh. I sometimes wonder what makes the people who mandate such warnings think that anyone stupid enough to need the warning is smart enough to read it.
|
|
Tuesday |
Tuesday, May 11, 1999 Change in plans. Yesterday, I finished cutting down a chapter for the Nutshell book. My editor, Robert Denn, asked me to take two chapters in the "fat" form I'd written them and compress them for him. Each of them started at about 14,000 words, and I cut each down to about 11,000 without seriously compromising content. That's hard work. I finished the second one yesterday, and I planned to get back to drafting new material today. In the mean time, Pournelle just sent me two chapters from the big book with some stylistic changes that Robert had requested. I wasn't expecting to get those back from Jerry for a week or two. I'm the one who does final formatting, so it's time to drop everything and get those chapters ready to re-submit to Robert. * * * * * This from Pinegar, Joe, HMR/US [Joe.Pinegar@hmrag.com]:
Good points. I think you are probably right. * * * * * This from Bo Leuf [bo@leuf.com]:
Ain't that the truth? Someone dumb enough to need a written warning not to put his hand into the lawnmower while it's running gets to reproduce. In the wild, a lawnmower or some other dangerous predator would have helped clean up the gene pool... * * * * * Here's another one of those messages I sometimes get from people who appear to be morons or worse. Sometimes I'm not sure whether it's simply someone having me on, because I find it hard to believe that anyone could be this dumb. I've cleaned up the more egregious grammar and spelling errors and removed the name to protect the guilty:
Duh. Please do the world a favor and don't have any children. I'm surprised you were able to access my web site and, having accessed it, were able to read it. Tell the truth. Someone at the mental care facility where you reside helped you, right?
|
|
Wednesday |
Wednesday, May 12, 1999 I got an interesting email this morning from Executive Software, the makers of the Diskeeper disk defragmentation utility, which I use and recommend. They've made their Emergency Undelete utility for Windows NT freely downloadable for a limited time. They ask for your name and contact information, where you found out about them, and have an opt-out mailing list. I haven't played with Emergency Undelete yet, but it appears from the description to be something like the Norton Unerase utility, but for Windows NT. Probably something worth having in your bag of tricks if you use NT. * * * * * Speaking of downloads, I see on Tom Syroid's web site that Microsoft has released Windows NT Service Pack 5. SP5 includes scores of fixes, which you can read about here. You can download SP5 here. According to Microsoft, the SP5 software itself is complete, but the documentation isn't. That will continue to trickle in over the next several weeks. * * * * * Speaking of Windows NT, I forgot to mention that Airborne showed up Monday with a padded envelope from Microsoft. I knew what was in it, but I didn't want to think about it. I was feeling guilty enough this morning that I decided I'd better at least open the envelope. It had a tab on each corner, labeled Pull Here to Open. So I pulled each tab, and each ripped off the envelope without actually opening it. What they should have written was "Pull Here and then Shred Envelope to Open." It was worse than I'd feared. Copies of Beta 3 of Windows 2000 Professional, Windows 2000 Server, and Windows 2000 Advanced Server. For both Intel and Alpha. And a CD labeled High-Security. There's a stack of CDs the better part of half an inch thick, and three separate sets of boot floppies. The enclosed letter told me that these betas expire after 444 days, so I suppose their time-out is long enough to make it worth loading them on a computer. Now the questions are "which computer" and "which operating system". I suppose it's about time to retire sherlock to duty as a test machine. Sherlock is a Dell Dimension XPSM200s with a 200 MHz Pentium/MMX and 64 MB of RAM. That should be enough to run Windows 2000 Professional (formerly NT Workstation). If not, Microsoft has some serious problems. The problem with using sherlock is that it has a bunch of data on it that I'd rather not risk. All of it could be replaced--ftp downloads, copies of CD's, etc.--but it'd be a pain in the butt to do so. I think I'll create a partition on a hard disk elsewhere on the network and migrate all that stuff to a different system. That's going to be one heck of an xcopy operation. I think there's something like 4 GB of data to be moved. Technically, I suppose I shouldn't be talking about W2K. I think I'm still under non-disclosure about Windows 2000, although I don't see any mention of the NDA in the printed material that came in this envelope. In the past, the cover letter always reminded me about the NDA. In practical terms, I can't imagine that Microsoft would care. They are, after all, shipping out a million copies of this beta, or close enough not to matter. So I suppose I'll talk about what I learn, assuming I can find the time to actually work with W2K right now. * * * * * This from Dave Farquhar [farquhar@lcms.org]:
In the immortal words of Bartles and James, "thank you for your support." I actually got mail back from that moron, accusing me of being "mean." Yep, that's me, greedy *and* mean. Geez. I've added his email address to my kill file. I started to write "I've added him to my kill file", but I was afraid he'd think I meant I was taking out a contract on him. One supposed benefit of civilization is that people so stupid that they would otherwise have been eaten by a predator or stumbled off a cliff are now protected against their own stupidity. I suppose this means that evolution effectively ceases when civilization coalesces. That such people not only survive but breed is certainly no good thing for the gene pool. * * * * * This from Scott Taylor [staylor@airmail.net]:
Thanks. That's useful information for anyone who still uses his Pentax Spotmatic. I haven't used mine seriously in probably 15 years or more. I do remember buying batteries for it at the drugstore, and I had been assuming that they were still available there. * * * * * This followup from Scott Taylor [staylor@airmail.net]:
Okay. That makes sense, I guess. I knew the government had banned Mercury in batteries, but I was thinking that I used to buy Silver Oxide batteries that were a drop-in replacement for the original battery. Perhaps not. * * * * * This from Tom Syroid [tsyroid@home.com]:
NT has a boot partition and a system partition. The names
are counter-intuitive in the sense that the system partition contains the
files that NT needs to boot (e.g. Ntdetect.com and ntldr), and the boot
partition contains the remainder of the NT system files. Usually, of
course, the boot and system partitions are the same partition. * * * * * Afternoon: I decided to go ahead and install Windows 2000 Professional on sherlock. Everything seems to have worked fine, and I'll have my detailed installation notes up soon. * * * * * This from Paul Robichaux [paul@robichaux.net], concerning the Service Pack 5 update:
Thanks. That's good to know. I believe I'll just wait until 5/19 and get the whole distribution with strong encryption. I prefer to install from a local copy. I'm installing Beta 3 of NT5 Workstation (er, W2KP) right now. It looks pretty enough. It'll be interesting to see what's still missing... * * * * * This followup from Paul Robichaux [paul@robichaux.net]:
I'd go for the second Pentium II/350 next week. They'll be $163 (quantity 1,000), and that's about the cheapest they're ever going to get. Intel historically cuts prices on a CPU constantly until that CPU is destined to disappear. The best evidence of that is that the price doesn't get cut when they cut prices on other ones. I'd grab the PII/350 now, while you can still get it. It's probably not going to be available much longer. The other issue is matching S-Specs with your existing processor, which you should try very hard to do. That is, if your existing PII/350 is a retail-boxed SL38M, you want another SL38M (or an SL36U, which is the OEM equivalent) for your second CPU. You can find the Master S-Spec table here.
|
|
Thursday |
Thursday, May 13, 1999 I spent some time yesterday installing Beta 3 of Windows 2000 Professional. If you're interested, my preliminary installation notes are here. Other than that, there's not much to write about, and all the mail has been personal. Back to work on the book...
|
| ' |
Friday, May 14, 1999 Horrible thunderstorms last night. We got a bit over 3.5" (9 cm) of rain, and the lightning and thunder came in waves from about 2:00 a.m. on. The power failed briefly about 2:00 a.m. to the accompaniment of multiple UPSs screaming, and then came back on immediately. I'd just rolled over to go back to sleep when the power failed again, came back on again, and then started ping-ponging on and off. I decided I'd better go shut everything down. One of these days, I'll install the APC PowerChute automatic shutdown software on my NT boxes. I say that every time this happens. At any rate, I ran around in my underwear powering down boxes. Everything went fine until I got to kerby, my main workstation, which runs Windows NT Server 4.0 SP3. I issued the Shutdown command and waited. And waited. And waited. For five minutes or so, the disk churned, flushing cached stuff out to disk. Fine. After ten minutes, I was still looking at the blank green shutdown screen, and disk activity had dwindled to a sporadic blink or two. This didn't look like it was going anywhere fast. I finally powered down the monitor to wait for a while longer. The system was being powered by the big APC Smart-UPS 1100, so I wasn't really concerned about running out of battery power. It'd probably drive this box easily for an hour or more. But I didn't see any point to running the battery down to nothing. With the monitor off, the load indicator LEDs on the APC showed zero load. Still, after about fifteen minutes, I powered the monitor back up. Still a blank green screen. By this time, my wife had taken the dogs downstairs in case a tornado showed up. And there I sat, in the dark, smoking my pipe and waiting for NT to shutdown. By that time, the first wave of storms had tapered off, so I continued to wait a few more minutes. Finally, after what must have been twenty to twenty-five minutes, I turned on the monitor again. Blank green screen. I finally just powered the damn thing off. The culprit was undoubtedly one NT service or another. The problem is that NT doesn't really have a shutdown command. UNIX, now UNIX has a shutdown command. You can tell UNIX to shutdown NOW. UNIX sends a kill signal to each process--die, Die, DIE--and that process dies on command. Not so NT. NT sends a command to its processes something like, "Hey, guys, we're thinking about shutting down the server whenever it's convenient for you. I don't want to rush you, but if you could, please get all your ducks lined up and then let me know. Of course, I won't actually do anything about shutting down until all of you say it's okay with you." Screw democracy. I want a shutdown command that shuts down the damn server. I don't know which service was guilty this time, or even that it was one service in particular. At times, I've suspected that it was the interaction between services that causes the problem. I run an older version of BackupExec (6.2?) on my resource server. BE by default loads several services, but only one of them is really required for BackupExec to run interactively, the way I use it. But that one service caused greatly delayed or failed shutdowns. I've actually timed it. With that service not loaded, the server shuts down in a minute or so. With that service loaded, the server takes more than 20 minutes to shutdown. So, each time I want to do a backup, I start that service manually. Each time I finish doing a backup, I stop that service manually. Geez. In this case, my first thought was that Diskeeper 4.0 was the culprit. I have it configured to run automatically every night at 2:00 or 3:00 a.m., so the timing was coincidental. That's really why I waited as long as I did. I feared that Diskeeper was doing something that shouldn't be interrupted. But, thinking about it, there's no way that Diskeeper was at fault. As it happens, I have it loaded on thoth, Barbara's computer, where it is configured similarly to mine. Barbara's box shutdown immediately, so Diskeeper can't very well be the culprit. I suppose I could use trial-and-error to determine which service or services are causing the problem, but I just don't have the time. Thank goodness for the APC UPSs, though. They do exactly what they're supposed to do, and they do it every time. Every time. That's a critical aspect of a UPS, and one that most people don't think about. Or, rather, they just assume that it will be the case. That's a bad assumption. Back before I finally got it through my head that a top-quality product was worth paying a little more for, I bought quite a few lesser UPSs. All of them were brand-name products, and I just assumed they'd work as advertised. They usually did, but usually isn't good enough for a UPS. Several times, one of these less expensive UPSs simply failed to switch. The battery was fully charged, the power failed, and the computer died instantly. Every time that happened, the cost of replacing the lost work for that one incident greatly exceeded the small price difference between the cheaper UPS and an APC. Never again will I use anything but an APC. * * * * * Lunchtime: At about 10:30, our across-the-street neighbor called Barbara, who was out doing errands, the gym, and grocery shopping. I listened as Paula left a message for Barbara on Barbara's answering machine. Paula sounded like she was in tears. When I picked up, she told me that her home alarm had been going off constantly since 2:00 this morning, and was about to drive her insane. Apparently, the company that installed it was too busy to come over and fix the problem. She asked if she could borrow a small screwdriver, and I told her I'd be right over. The first thing she pointed me to was a power brick that was secured to the receptacle with a small screw. I removed the power brick. There was a piece of old-style quad wire that disappeared back into the wall and had its red/yellow wires connected to one terminal of the power brick, and its green/black to the other. Paula said the guy had told her to remove the wires to it. That seemed odd, but I went ahead and did it. The siren kept sounding, as I expected. Then she showed me back to the hall closet, where the guts of the alarm were hidden. I pulled one wire off the backup battery and killed the noise. While I was putting the wires back on the power brick--which was difficult lying on my stomach in a built-in knee space with no light--Paula was on the phone with the guy from the alarm company. He gave me a hard time about taking the wires off in the first place, and started lecturing me about basic electricity. I told this moron that I'd gotten my First Class Radiotelephone License twenty-five years ago, which allowed me to tune a television transmitter for God's sake, and that I'd just been doing what the lady thought he'd told her to do. If he'd come out and fixed the problem as he should have, Paula wouldn't have had to listen to that alarm sounding for 8 hours, and I wouldn't have even been there. I hate being lectured to by someone who didn't understand what a "power brick" was. Once I'd explained, he gratuitously informed me that the proper term for that item was "transformer." Duh. * * * * * Paul Robichaux informs me that the 128-bit version of Windows NT 4.0 Service Pack 5 is now available for download here.
|
|
Saturday |
Saturday, May 15, 1999 Communications problems this morning. I don't know if they're at my end or at BellSouth. The connection stays up, but IP connectivity dies. Then the connection itself dies, but WinGate doesn't realize it's died. The only way short of rebooting the WinGate server--tough when you're using it to do a backup at the moment--is to stop the WinGate Server engine and then restart it manually. At that point, I can reconnect and the connection works for a while, but never for long enough to get this page published. * * * * * Congratulations to Leah and Tom Syroid:
* * * * * New Scientist magazine has a fascinating article about a new technology that may ultimately mean the end of books on paper. The article describes "electronic paper" which resembles ordinary paper in thickness, flexibility, and readability, but can display images in much the same way as a traditional LCD screen. Someone will, before long, use this technology to build an electronic book that resembles a traditional hardback book but will store the text of thousands of books and display it on demand. This isn't just a theoretical technology, either. The inventors have actually produced products based on it. Right now, as you might expect, it's too expensive to be a mass-market item, but that's just a question of production. Expect to see consumer products based on this technology in the next few years. Now the only problem, as with all electronic books, will be the stranglehold that publishers have on authors and their content. Still, they can't hold back the tide. * * * * * This from Robin Gould [rgould@ihcc.org]:
I've had this happen a couple of times, and each time it was because I'd not entered the correct subnet mask. NT defaults for some reason to a Class B netmask (255.255.0.0) rather than the more common Class C (255.255.255.0). The first thing I'd check is that you're using the same netmask on the new BDC as you are on the other machines on the network. I'd have been completely comfortable with this suggestion, except that you mentioned you were using both TCP/IP and NetBEUI on your network. I usually install only TCP/IP, but I'd have thought that NetBEUI by itself would have been adequate for the new server to find the existing PDC. I wouldn't think that using the BackOffice Server CD would have anything to do with this. I'll go ahead and post your message in case any of my readers have experienced similar problems. * * * * * This from M.F. or M.L. McDonell [mcdonell35@earthlink.net]:
Sorry to hear of your troubles. Although a few people I know swear by the 56K modems, most of them say that they don't notice a whole lot of difference subjectively. The early 56K modems often connected at little more than 33.6. The newest crop of 56K modems can make connections at greater than 40 or 45 Kbps, but need a pretty good phone line to do so. With cable modems and/or ADSL imminent or actually available for many people, I just don't see that 56K (or ISDN, for that matter) makes a whole lot of sense. Glad to hear you're working again, if only at 33.6.
|
|
Sunday |
Sunday, May 16, 1999 No mail overnight. None. No reader mail. No personal mail. No mailing list mail. Not even a return-receipt or a Spam message. My first thought, of course, was that my mail server was hosed. Not so. I just didn't get any mail. First time in years. Usually, I get ideas to write about here from my mail. Lacking that, I looked for inspiration at the web sites I visit every day. None of them had been updated. Last resort, the morning newspaper. Nothing there worth me writing about. Nothing there worth them writing about, actually. Well, one thing, but I've written about it before. Intel introduces the Pentium III/550 tomorrow, which is great news for many of us. Not that many of us would actually buy a PIII/550 at $744; but the downward price pressure on other CPUs is good news indeed. The Pentium III/450 at $268 is likely to be the sweet spot. About 82% the performance of the PIII/550 for about 36% of the cost. Of course, you can get even better bang-for-the-buck with a Celeron, even though their prices are not being cut, at least unless you're running software that is SSE-enabled. The Celeron gives you about four fifths of the performance for about one fifth the price. It's no wonder that corporations are beginning to buy Celeron-based systems in large numbers. AMD, as usual, is clueless. They've cut the price of the K6-III/450 to $226, thereby positioning it as a less expensive alternative to the Pentium III/450. The problem, of course, is that the K6-III doesn't compete against the PIII. However much AMD wishes things were different, the K6-III competes against the Celeron, and is grossly overpriced in that competition. Actually, there's some justification for AMD's pricing. They can't manufacture enough K6-IIIs to meet demand, so they might as well lose sales against the higher-priced PIII/450 as against the Celeron. As it is, AMD couldn't be doing a better job of converting themselves into a niche CPU manufacturer if they were trying. I wish AMD well. Without them, the Celeron would never have seen the light of day and we'd all still be paying very high prices for Intel CPUs. But I'll let others buy AMD CPUs to benefit the rest of us. I'll continue using Intel. The other alternative CPU manufacturers are also in trouble. National Semi is apparently selling Cyrix, and IDT is having money problems. The only reason these alternative CPU manufacturers are still in business is the economics of designing and producing CPUs. Most of the cost comes in designing the CPU and building the fabrication facility. Once that's done, the incremental cost to produce one more CPU is very small. Most of the selling price of that CPU goes to pay back the costs incurred up-front. Right now, I suspect that alternative CPU manufacturers are selling their products at prices too low to amortize the development costs. But better that than no return at all. It at least provides some cash flow. But AMD and the rest are eating their seed corn. That can't go on forever. When the AMD K7 ships, expect Intel to cut the price of the matching Pentium III dramatically. Once again, AMD will be in the position of shipping a decent processor, but at a price inadequate to repay development costs. Intel can keep that up a lot longer than AMD can.
|