|
Week of 3/15/99 Friday, July 05, 2002 08:11 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, March 15, 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've given up on using the IE5 off-line browsing feature. I never could get the scheduled updates feature to work properly. I could have lived with that, and simply started a manual update before I went to bed or when I sat down first thing in the morning. The real problem is that the process keeps blowing up. I've gotten repeated GPFs from mobsync.exe, which trash the current update and require me to restart the whole process. So, for the time being at least, I recommend that you not use this feature. I hope that the release version of IE5, due the end of this week, fixes these problems because this could be a really useful feature. I'll try again once I have a copy of the release version and have had a chance to test it exhaustively. And I've gotten several reports from readers telling me that they're having exactly the same problems with the IE5 off-line browsing feature (and IE4 "subscriptions") that I've been having. This would be a nice feature if it worked, but it's apparently just not ready for prime time. * * * * * A couple of readers have commented on the timestamps on my pages. When my site was hosted on BigBiz, those times reflected the local time at BigBiz, which is Pacific Standard Time (PST, -0800). I'm located in North Carolina, USA, which is Eastern Standard Time (EST, -0500), and my web site is now hosted on pair Networks, which is located in Pittsburgh, Pennsylvania (also EST). I'm still updating and publishing the pages at about the same time as before--usually around 0700 to 0900 my time, or about 1200Z to 1400Z (UCT or GMT). Some confusion may have resulted while I was having all the problems with BigBiz because I often couldn't get the updates to complete at the usual time. Before, I'd publish around 0800 my time, and the pages would show 0500. Now I update at 0800 my time, and the pages show 0800. Sorry for any confusion. * * * * * The Register reports that all 0.25 micron Intel CPUs have the serialization "feature." That includes Deschutes-based Pentium IIs, the current Mendecino Celerons, and Xeons. So it appears that most of us have what may well turn out to be a rattlesnake coiled within our processors. * * * * * This from Francisco Garcia Maceda maceda@pobox.com
What a truly excellent idea. I confess that I'd thought about just lying a motherboard flat on my work table and connecting everything to it, but I'd not thought about doing what you suggest--essentially creating a breadboard system. My first thought was that the lack of the ground plane provided by the screw connections to the motherboard tray would cause problems, but if you're doing it that's obviously not the case. Perhaps the boards radiate a bit more than they otherwise would, but that doesn't concern me. What does concern me a bit is cooling, but I can put a small traditional fan pointing at the CPU. Thanks again for the suggestion. I think you've solved my problem. * * * * * More from Francisco Garcia Maceda maceda@pobox.com
I've seen systems misbehave due to improper grounding of motherboards and expansion cards, so I think I'll take some measures to prevent that. One thing I could easily do is put a grounding strap accross the rear of the board that supports the motherboard. I could then clamp the lower tongue of each expansion card to that strap. As far as power supply switching, that shouldn't be much of a problem. AT power supplies are available that have the traditional big red switch on the power supply case itself. ATX power supplies are switched on and off under the control of the motherboard, so all I need is a simple two-wire connection with header-pin connectors at one end and a SPST switch at the other. For that matter, I could simply jumper the motherboard power-switch connector forced-on and use the power switch on the back of the ATX power supply to turn power on and off. You're correct that I won't have much problem with cooling most of the time. Most of my CPUs have attached fans. But I do, for example, have a Pentium III CPU here that has no attached fan. Instead, it has the mother of all heat sinks on it, and it needs a fan to blow air past it. If I trust to convection only with this CPU, it'd cook itself in very short order. * * * * * The following from Bruce Denman [bdenman@FTC-I.NET]:
Yep, that article pretty much says what I've been thinking. AMD appears to be in deep kimchee. They're selling many of their low-end K6-2 CPUs at or below their cost to produce them, and that's the *good* part of the picture. They can't get the K6-III produced in any reasonable quantity. Even when (or if) they solve their K6-III yield problems, they're going to have to sell it at a very high price to break even, and I don't think the price they need to get for it will be sustainable against the Celeron. And they might as well be eating their seed corn. I see that they are shipping only one example each to the Taiwanese motherboard manufacturers. One each. How on earth do they expect companies to produce motherboards for the K6-III if they can't even supply a reasonable number of engineering samples? It's no wonder they're facing stockholder suits. It wouldn't surprise me if AMD went under by the end of this year. * * * * * And more from Bruce Denman [bdenman@FTC-I.NET]:
That brings up something I should have mentioned, and it's pretty important. I mentioned in passing to Jerry yesterday that I'd subscribed to his site and downloaded about 20 MB of stuff. I also said that that wasn't a very "friendly" thing to do to a site. Here's why, using me as an example. My web hosting account allows up to 200 MB of data transfer a day. If I go over that, I have to pay something like $3 for each 10 MB. So, if my site is 20 MB and eleven people subscribe to it in one day, I end up having to pay $3 overage charges that month. If 25 people do it, I end up getting hit with a $90 overage charge that month. A lot of people have traffic figured by the month rather than by the day as pair Networks does, so that's less of a factor for them. But even so, having a lot of people subscribe to a site can cost the site owner some significant money. Obviously, updates on sites that use static pages (like mine and Jerry's) are much smaller. I have no objection at all if someone wants to subscribe to this site. But please don't do it all at once. If you're going to do it, wait a random number of days between one and ten and then subscribe. Better yet, flip a coin three times, treat heads as zero and tails as one, and use the results as a three digit binary number from 000 (do it today) to 111 (do it in seven days.) How's that for a solution appropriate to my readership? * * * * * And more still from Bruce Denman [bdenman@FTC-I.NET]:
I try to make at least one quick visit to all of my readers' sites, although I confess that I sometimes am unable to do so under the press of other business. As far as the IE5 focus problem, the method you describe works, but any method works if I *type* the URL into the box. The problem occurs only when I select the URL from my drop-down menu list of Favorites. Anand's site (and to a lesser extent, Tom's) are about unworkable for subscriptions. Anand uses essentially 100% dynamic content, which means the whole site is "new" each time you visit it. Tom uses mostly static html for his actual content, but there's so much dynamic ad content that it takes quite a while to refresh his site. I tried it this morning, and IE downloaded about 5 MB of "new" stuff, almost exclusively ad material. And it's slow responding, so the update took about half an hour. Given those problems, and the GPF problem I keep having with mobsync.exe messing with this feature just isn't worth it for me at the moment. Maybe Microsoft will fix the problem in the release version, but I somehow doubt it. * * * * * This from Bo Leuf [bo@leuf.net] Leuf Network [www.leuf.net]
I've never had much problem from a resource conflict perspective getting NT installed. I suspect that at least some of your problems may be due to your BIOS/chipset settings for the PCI bus. There are a few things to keep in mind:
Other than that, the only solution I've found is just to continue playing with it until you get it stable and have no conflicts. Someone else may have some additional advice, so I'll go ahead and post this. * * * * * This from joshua d boyd [jdboyd@cs.millersv.edu] http://catpro.dragonfire.net/joshua
As far as your your car, I'm pretty sure you can still get Freon R-12. It's not cheap, but the last time I checked it was still available. I believe it's now illegal to manufacture and sell R-12, but not to sell R-12 that was made before the ban. I think companies salvage it from junked cars, etc. and then re-sell it to people who have old-style auto A/C systems. At least, I certainly hope so. One of our Isuzu Troopers (the '93) uses R-134a, but the '92 Trooper uses R-12. * * * * * This from Warrick M. Locke [warlocke@mail.wf.net]
I suppose it could happen as you suggest, but I really don't think it will. If and when I see Microsoft formally announce that they're porting Office to Linux, I'll begin to reconsider. When I see an actual shipping beta, I'll believe it more. But until that happens, I'll stick to my belief that this is simply a rumor. * * * * * Noon: My sources tell me that NT4 Service Pack 5 is now in beta. If you hadn't heard about SP5, well neither had I until this morning. I was under the impression that SP4 was the last SP Microsoft would release for NT4 and that any other required patches would be released piecemeal as hotfixes. But apparently that's not the case. I don't have any other details about SP5 other than informed speculation from my source that it's primarily bug fixes. This may be an indication that Windows 2K is slipping even further. At this point, I half expect Microsoft to ship an interim NT 4.5 as a stopgap. That'd be a desperation move on their part, however, as doing NT 4.5 would extend the ship date for W2K even further. * * * * * This from Edmund C. Hack [echack@crl.com]. His first paragraph refers to the website I mentioned last week, whose owner savaged Pournelle's site.
It's interesting that her site has disappeared. Perhaps she started to exceed her traffic allowance based on all the hits from here and Pournelle's site. I'm not sure what AOL's traffic allowance is for a personal site, but I suspect it's very small. As far as Juan Rico, I didn't make the statement based on what I'd heard. I made it based on what I read. The first time I read that book was many years ago, probably the early 1970's. But I distinctly recall that, toward the end of it, Heinlein made clear in context that Juan Rico was black. I remember the light bulb flashing on in my mind, because, until that point in the book if I'd assumed anything at all about Mr. Rico, I'd assumed that he was a white of Hispanic heritage. I wouldn't have assumed that he was black based simply on a Filipino heritage, so I'm pretty sure that Mr. Heinlein inserted at least a strong hint that Juan is black. I remember thinking at the time that it was a very clever device. And I suppose it is possible that later versions of the book modified the text somewhat. The version I used to have was one of the very rare copies entitled "Starship Soldiers" rather than "Starship Troopers." I made the mistake of lending that copy to a friend who now claims he doesn't have it and doesn't remember borrowing it. At any rate, I'll forward this to Dr. Pournelle. Perhaps he can cast further light on the question. * * * * * I got an interesting press release from PowerQuest Corporation (the PartitionMagic folks). They have a new product called DataKeeper™ 3.0, a real-time backup utility for Windows 95/98. Although I use Windows NT almost exclusively, this looks like it might be a very useful product for Windows 9x users. I have an eval copy in hand, and will play with it once I get some free time. If you want to read about the details, you can do so at http://www.powerquest.com/dkpromo * * * * * This from Chuck Waggoner [waggoner@gis.net]
You're certainly correct that Microsoft programs have some, ahem, undocumented features. I guess it's really a philosophical question. Do we want X number of features, all of which are 100% stable and bug-free, or do we want 10X or 100X number of features, 90% or 95% of which work as they should? Overall, I think Microsoft does a pretty good job. IE5 beta overall is certainly a better, faster product than IE4, let alone Navigator, which is beginning to look pathetic in comparison. So, although I'd love to see these problems fixed, I think I'd just as soon have Microsoft adding these new features even if some of them aren't quite 100% right yet. But then again, tomorrow I may feel exactly the opposite.
|
|
Tuesday |
Tuesday, March 16, 1999 Yesterday was a bust. I accomplished very little, although I did get Barbara started on doing some stuff that needs done. Everything that could break broke. Everything that could fail failed. My main workstation was responding so slowly that I finally decided to restart NT. Task Manager showed normal numbers for memory usage and CPU utilization, but things just weren't running right. My speaker icon disappeared from the task bar for no apparent reason. And on and on. So I did a Restart. Windows NT shutdown normally to the point where it goes to the green screen and flushes cached data to disk. But it then just sat there beating the hard disk to death. The hard disk LED was lit solid red for literally 15 minutes or more, accompanied by the banging of the drive heads. I finally did what I usually to hasten shutdown. I pressed Ctrl-Alt-Del again. As usual, that caused NT to pop up an information box to tell me that it was flushing cache to disk and to please wait. But, also as usual, the shutdown completed normally within a half minute or so after I did that. I don't know if it's coincidence--perhaps I always reach my frustration limit a minute or so before NT would otherwise have finished the shutdown--or whether doing Ctrl-Alt-Del actually breaks some kind of endless loop or something. But at least it did finally shutdown and restart, and everything appears to be back to normal now. * * * * * pair Networks is driving me nuts. I signed up with them on March 1st. They had a special deal. Triple discounts for prepaying. The normal discount for paying a full year in advance is 8%, and during the month of March they're discounting by 24%. So far, so good. I signed up using their automated process, and captured the final page, which listed the amount I owed them. $194.10. On March 5th, two Fridays ago, they sent me an invoice copy that showed they'd charged my credit card $234.97, which was due to them calculating the amount using the 8% discount. I immediately fired off a protest. Since they'd already charged me card, I wanted them to issue a credit that day. I heard nothing from them other than their standard autoresponder message. Finally, last Tuesday, I mailed their urgent address, asking them to please correct their billing error immeidately. I got a human response from that address, saying that they had nothing to do with billing, only with fixing urgent system problems. Fine. I mailed the billing address again. No response except the autoresponder. I mailed them late last week again, and still no response but the autoresponder. I mailed them again yesterday afternoon. I know there's someone there, because they also screwed up Paul Robichaux's bill. They charged him the correct amount, using the 24% discount, but the invoice they sent him said his service expired in six months rather than a year. He got that message from their billing department last Wednesday. So why are they ignoring me? They don't have a phone number to call, and they apparently ignore email sent to them, including messages marked urgent. I'm starting to get seriously annoyed. They have $40 of my money that they shouldn't have. I'm going to have to pay my credit card company an extra $40 and wait a month for the credit, and this is pair Network's fault. I am livid. And they tell you not to dare dispute the amount they bill with your credit card company, or they'll cut off your service. This has been going on for ten days now. They are more than happy to take your money, including money they are not entitled to. They aren't so quick about rectifying their mistake. This sucks. * * * * * This from Jerry Pournelle [jerryp@jerrypournelle.com] http://www.jerrypournelle.com
Okay, thanks. I'm going to have to back and re-read Starship Troopers and find out what it was that made me think Juan Rico was black. Or perhaps I wasn't so much thinking that he was black as that he was not white. When I first read Starship Troopers 25 years ago and more, it was still pretty unusual in mainstream science fiction to have a lead human character who was other than white. But I definitely remember being surprised when Mr. Heinlein dropped that revelation in near the end of the book. At the time, I remember thinking that he'd carefully developed Juan Rico as a hero that any reader would admire, only then revealing that this hero happened to be of a different racial group than most readers would have assumed. At the time, I thought it was Mr. Heinlein's clever way of pointing out that racial intolerance existed only between people who did not know each other as individuals. Re-reading Starship Troopers is no bad thing anyway. * * * * * And Pournelle adds:
* * * * * This from joshua d boyd [jdboyd@cs.millersv.edu] http://catpro.dragonfire.net/joshua
The price I quoted was for the Promise FastTrak IDE RAID controller (about $100) and four Maxtor DiamondMax Plus 2500 10 GB hard drives (about $225 each). The controller is just a standard PCI expansion card, although Promise does sell (or soon will) a solution that includes hot-pluggable drive carriers. As far as your drives dying, that does seem an excessive number. It may be coincidence, but I'd be inclined to suspect overheating as the cause, given what you've said. A drive should at most be warm to the touch after running for a long time. If it's uncomfortably hot to the touch, that's too hot for its health, and you need to do something. What you can do depends on the specifics of your situation. First, if possible, get some separation between the drives. I've seen systems that had multiple 3.5" hard drives stacked three and four high in bays with almost no separation between the top of one drive and the bottom of another. This kind of arrangement makes it impossible for the drives to rid themselves of heat, and the drives in the middle of the stack can literally become hot enough to fry an egg. If your case allows doing so, separate the drives, putting some in the front bays and others in concealed internal bays, always leaving at least one open bay between two drives. Second, depending on the space you have available and the how the drives are mounted, you may be able to install drive coolers.If that's not an option, you may be able to mount the drives in external bays farther back than usual, pop the external bezel, and put a standard fan in front of the drives to pull air out of the case and past the drives. Finally, although it's an ugly solution, you may be able to run with the cover off and set the drives on top of the case itself, obviously using a sheet of cardboard or something similar to insulate the bottom of the drives from the metal case. If you're buying new drives, go for larger ones that will allow you to use fewer drives to replace multiple smaller drives. Although drives with a high rotation rate are desirable from a performance point of view, they also often run much hotter than standard drives. I've seen several reports of the Maxtors I'm using running hot, but they don't do so in my machines. They're just barely warm to the touch. * * * * * This from Francisco Garcia Maceda maceda@pobox.com
Good point. I never even thought about NT Diagnostics. The way I read Bo's message, he wanted a way to change stuff from within NT as opposed to just looking at it, but this may indeed have been what he meant.
|
|
Wednesday |
Wednesday, March 17, 1999 Anand is a maniac. Last weekend, the poor guy was running a temperature of 102°F (that's without a Kryotech cooler) and was rushed to the emergency care center where they found he had the flu, a strep infection, and something else that I don't recall. They told him to get bed rest, but he couldn't stay put. So he posted his review of the Promise FastTrak IDE RAID controller yesterday afternoon. I don't know why it surprises me, but Anand basically trashed the FastTrak. Now, in my opinion, the FastTrak provides more bang for the buck than just about any other disk-related upgrade you can make to a PC. So Anand has relatively nice things to say about the Kryotech cooler, which I think only an idiot would buy, and bad things to say about the Promise FastTrak, which I think is a top-notch product. I don't know if this is just an extreme difference in philosophies, or what. Anand dings the FastTrak because it provides only moderate disk performance benefits under Winstone, although he admits that it shows dramatic performance benefits under WinBench. But he sniffs at WinBench as a "synthetic" benchmark. Well, of course it is. Winstone is first and foremost an application benchmark that focuses on CPU and memory performance. Application benchmarks, including Winstone, tend not to stress the disk subsystem. WinBench, on the other hand, forces gigabytes of data through the subsystem, which minimizes the effects of caching and uncovers the true underlying performance of the disk subsystem. And I'd think that's what he'd want to test for. I sent Anand email a couple of days ago, telling him that I was savaging him over on my site, and suggesting he visit it. Actually, the truth is that I admire Anand. He's doing a big job, and overall he's doing it pretty well. Certainly, he's more ambitious than I was at 17 years old. Since I'm in Winston-Salem and he's in Raleigh, we're separated by only a couple hours' drive. I suggested that we should get together some time. I haven't gotten a response to my email to him, but I suspect that's because he's covered up with get-well messages. * * * * * From the interesting business ideas department: someone posted a message on the Computer Book Publishing mailing list yesterday about a university professor and his wife who are selling $4,000 worth of books per day from a web site that cost $2,500 to start up. According to the message, this guy has to buy only five copies of a title to equal Amazon's cost on that title. He undercuts Amazon's price on popular books, and donates 10% of your purchases to whatever charity you designate. I'm not sure that this business has any long-term future, but it's interesting that they're doing as well as they are. * * * * * Another good episode of Buffy the Vampire Slayer last night. This show gets my vote for the best on television. The writing is intelligent. There aren't many television shows that can drop in quotes from Shakespeare in context and make it work, and I'm pretty sure I've caught other quotes from sources as diverse as Suetonius and Swift. The cast is marvelous and the acting is first rate. Sarah Michelle Gellar is a fine young actress. She manages to portray Buffy as formidable and vulnerable at the same time, no easy task. Alyson Hannigan as Willow is always delightful. And the rest of the ensemble cast is equally good. Once you get past the high-camp premise, this is one extremely well done television show, and I say that as a person who generally dislikes television. I'm very glad that Pournelle mentioned Buffy, because I'd never have watched it if he hadn't recommended it. If you haven't seen it, give it a try. Better yet, give it a couple tries. Your first reaction will probably be the same as mine, "I can't believe I'm watching this." But give it a chance. It'll probably grow on you. * * * * * UPS just showed up with eval units of the Autotime Processor Protector. This is one of those products that you don't know you need until you find out it exists. It comes in Socket 7 and Slot 1 versions, which plug in in place of the CPU. When activated, the unit displays the voltage being provided to the socket or slot. This is useful in two respects: First, although you may know what voltage your CPU requires, it's not always immediately obvious how to configure the system board to provide that voltage, assuming that it is capable of doing so. Sometimes, that's simply because the system board is poorly labeled or because you've lost the manual. Other times, it's because the voltage you need is available but undocumented. As anyone who has done much work with Socket 7 system boards knows, you can usually configure a system board to provide undocumented voltages by connecting more than one pair of voltage jumpers. But figuring out which pairs to jumper to provide the voltage you want is another story entirely. If you know what you're doing, you can use your DVM to read voltages directly from the socket, but the Processor Proctector is much easier to use. You drop it in in place of the CPU, and read the voltage directly from its display. I wish I'd had this Socket 7 Processor Protector a couple of weeks ago when I was trying to build a Socket 7 system around an EPoX EP-58MVP3C-M system board and an IDT WinChip CPU. The WinChip was one of the 3.52 volt versions. The system board had jumper settings for 2.1, 2.2, 2.4, 2.8, 2.9, and 3.2 volts--a useful range of settings for most purposes, but nothing even close to the 3.52 volts I needed for the WinChip. I finally gave up, because I didn't want to track down which Socket 7 pins I needed to connect the DVM to. If I'd had this device then, I could have verified which combination of voltage jumpers would have given me 3.52 volts, or close to it. The second use for the Processor Protector is what its name implies. Over the years, I've run into a few system boards with voltage settings that were labeled incorrectly. This is surprisingly common, particularly on inexpensive and off-brand system boards. Sometimes it's simply a matter of the jumpers being mislabeled. Other times, the board itself has malfunctioned. In either event, before you drop that new $500 K6-III CPU into a system board, it's worthwhile checking to make sure that the system board is delivering the voltage that you expect. I haven't had a chance yet to actually put the Processor Protector through its paces. But if it functions as expected--and I have no reason to think it will not--it's going to appear on my Recommended List. The Socket 7 version lists for $39, and the Slot 1 version for $69. These aren't tools that the average home PC user needs, but if you do much swapping around of CPUs, these look like good things to have in your tool kit.
|
|
Thursday |
Thursday, March 18, 1999 Well, Internet Explorer 5.0 is due to be released today. I haven't tried to download it yet, but I'll do so as soon as I finish and post this. Thankfully, the release of IE5 was not delayed by the Blue Mountain Arts lawsuit. If you're not familiar with this one, it's just another example of how twisted our legal system has become. It seems to me that Microsoft is the Good Guy in this dispute. Blue Mountain Arts allows people to send "greeting cards" via Internet mail. The junk-mail filtering technology that Microsoft built into the Outlook Express module of IE5 filtered these messages to the Junk Mail folder. Blue Mountain Arts sued Microsoft to prevent them from including this filter. I've read what both Microsoft and Blue Mountain Arts have to say about this lawsuit, and I have to say that from my non-lawyer's point of view I agree completely with Microsoft. Microsoft says that "[...] the ruling does establish a dangerous precedent that essentially substitutes the view of Blue Mountain Arts for that of the customer in the marketplace[...]" and that seems to sum things up pretty well. In essence, Blue Mountain Arts has sued to prevent Microsoft from including a feature that allows the individual user to decide whether or not he wants messages moved to his Junk Mail folder. Microsoft is trying to give people a choice. Blue Mountain Arts is trying to prevent people from having that choice available. I hope Microsoft's legal department buries Blue Mountain Arts. They richly deserve it. * * * * * Well, I didn't really expect to be able to download IE5, and I couldn't. I had to be one of the first to attempt the download, because I kept the IE5 page up in my web browser and refreshed it frequently. Apparently, a lot of other people were doing the same thing. I got the initial 470 KB setup file after many retries, and then started to download the distribution files to my disk. IE5Setup has a "resume download" feature, but it appears to work only on a file-by-file basis rather than byte-by-byte. That means that when the download blows up in the middle of a huge CAB file, I have to restart downloading that CAB file from the first byte. Not good. After downloading a few of the early small files and getting throughput of only about 50 bytes/s, I decided to give up on this. I'll try again in a few days, once the goldrush has tapered off a bit. * * * * * This from Bruce Denman [bdenman@ftc-i.net] [http://web.infoave.net/~bdenman]
Thanks, but I think I'll just wait until the rush at Microsoft has petered out. I get so much new software that I'm never really in a hurry to load yet another new product. I'd like to get the IE5 release just to see if it fixes a few of the problems in the beta. But I can wait a few days. I've lived with it this long. * * * * * Afternoon: I'm off to Computer & Software Outlet to grab some stuff to build a test-bench breadboard system around. I'm covered up in system boards, processors, and memory, and need some quicker way to build a system than having to install the motherboard in a case. I think I'll pick up a CD-ROM drive, a floppy drive, an ATX power supply, a 10/100 Ethernet card, a video/keyboard switchbox and some cables. That ought to give me what I need to build the test stand. * * * * * A follow-up from Bruce Denman [bdenman@ftc-i.net] [http://web.infoave.net/~bdenman]
Yep. I'm not desperate enough to put up with the download timeouts. But I'll post this for those who just have to have it right away. * * * * * This from Chris Holden [chris.holden@gte.net]
Yep. I found three incidences, all referring to the Matrox Millennium video card. I actually do know how to spell the word, but my journal is completely unedited and I don't worry much about typos. I gave up using the FrontPage spell checker to find typos. It's pathetic. So what you see is what I type. Welcome to the site. I hope you enjoy it. * * * * * Late Afternoon: Back from CSO, where I picked up the stuff I need to build a test stand. When I got back, I decided to give the IE5 download one more try, this time with rather strange results. I restarted ie5setup.exe. As usual, it prompted me to resume the download, which I did. Also as usual, it showed that I'd download 326 KB out of more than 9 MB total. This time however, the status bar continued to sit at 326 KB for several minutes. I had an NT Explorer window open, and watched several 1 KB CAB files being written to the network drive where I told setup to store the distribution files. After several minutes with no activity, the download status bar suddenly ran all the way to 100% complete, and a box popped up telling me to run ie5setup.exe to install IE5. With no little trepidation, I did just that. What appeared to be a normal installation routine followed, although it took only a few seconds to run. At the end, it told me that IE5 had been installed successfully, and didn't even force a reboot. I started up IE5, and it appears to work normally. Help - About makes no reference to this version being the beta, which I'm pretty sure it used to do. It tells me that I'm running version 5.00.0910.1309, but I don't recall what number identified the beta version. The only thing I can conclude is that ie5setup.exe did an incremental download, going out and grabbing only a few small files necessary to update the beta to the shipping version. I wish I knew for sure what the heck happened.
|
|
Friday |
Friday, March 19, 1999 Well, apparently I was mistaken. Just for the heck of it, I tried running iesetup.exe again last night. A dialog popped up to tell me that I'd not downloaded all of the files needed for the configuration I selected, and the download started again. It blew up almost immediately, of course. But that begs the question: what am I currently running? It no longer says "beta." It appears to work, although all of the problems I've noted with the beta version are still present. Bruce Denman [bdenman@ftc-i.net] tells me:
So I guess I need to try the download again. I probably should have just ordered the CD, but by the time I thought about it it was too late. Enough of this. IE works well enough for now, and I need to get back to work. * * * * * It's two weeks today since pair Networks overbilled my credit card by more than $40, and I haven't heard a word from them despite sending them repeated messages protesting the overcharge. I am aggravated beyond belief with pair Networks. This situation long ago passed the inexcusable stage, and is beginning to border on criminal. They have basically stolen more than $40 from me by putting an unauthorized charge on my credit card and refusing to even acknowledge, let alone correct, the error. And their standard boilerplate threatens people in my position by saying that they'll cut off your service if you dispute the charge. I sent them another message this morning, which I copied to their urgent@pair.com address, telling them to credit me the overcharge immediately. This is actually the second message on this topic that I've sent to urgent@pair.com. They replied to the first one to tell me that that address was only for urgent technical problems. But if their billing department doesn't answer their email, what's a victim supposed to do? At this point, I have wasted much more than $40 worth of my time trying to get these bastards to fix their own mistake. No matter what happens now, I lose. Even if they credited me the entire amount, including the $194.10 they were supposed to charge me for a year's service, I'd still be out of pocket because of the lost time and aggravation. If you're looking for a web hosting service, don't even think about signing up with pair Networks. Definitely Not Recommended. * * * * * This from Edmund C. Hack [echack@crl.com]:
Good point, and one I should have made in my original post. When I first installed Outlook, I wanted to modify the filter that determines which messages go to the Junk E-mail folder. And I quickly found the filters.txt file. I was somewhat surprised that it was a pure ASCII text file, but I figured that perhaps Microsoft had done it that way to make it easy to add, remove, or modify filters. So I added a filter criterion, using the same syntax as was used elsewhere in the file. It didn't work. After some experimentation, I was able to learn that the filters.txt file simply documents the filters that are being used to determine which messages are sent to the Junk E-mail folder. Changing that file has no effect on filter action. It's easy enough to add a filter by creating a rule. For example, I created a rule that forwards messages from hotmail.com, juno.com, aol.com, and several other domains from which I was receiving a lot of spam to my Junk E-mail folder. That rule works fine. And, for that matter, you can simply right click on a message to add the sender to your junk mail filter. But it would be nice to be able to modify the junk mail filters rules directly. All of this really misses the issue, however. Blue Mountain Arts claims that their messages are being trashed, deleted, or blocked by Microsoft's junk email filter. None of that is true. As Microsoft states, they are simply moved to a different folder at the same level as the inbox.
|
|
Saturday |
Saturday, March 20, 1999 A strange thought occured to me the other night. I was sitting here working at my main computer, which is built around an Intel Seattle system board and a Pentium II/300 CPU. Barbara's main system is similar. It uses an EPoX system board and a Pentium II/300. But on the credenza behind me sits a bunch of Pentium III, Pentium II, and Celeron processors, all of which are faster (some significantly so) than the processors in our main systems. So why do I continue to use these slower processors when it would take only a few minutes to upgrade? Well, part of the reason is that these are eval processors. I need them to swap in and out of various system boards for benchmarking and testing purposes. But that's not the whole reason. As strange as it may sound, I don't feel the need for any more speed. The Pentium II/300 is more than fast enough for what I do--Outlook mail, web browsing, creating and updating web pages, and so on. If I did things that were very CPU intensive and time-consuming, the upgrade would be worthwhile. If the Pentium II/300 took half an hour to do a job, putting in a processor that would cut that to 15 minutes would be a very worthwhile upgrade. If I did repeated compiles that took 30 seconds on the Pentium II/300, swapping processors to drop that to 15 seconds would be worthwhile. But mostly I do things that take only a few seconds or less. So there's little point to swapping processors to drop the time required from one second to half a second. So then I began to wonder why I bother to benchmark these things. After all, the difference between "really, really fast" and "really, really, really fast" is sometimes not even perceptible to me. But in another sense, the difference is real and important. In order to make rational decisions about how much money to spend on a processor, one must have some concrete evidence of what tradeoffs are involved. So I think when I bring up my new hardware site, if I do, I'm going to include a frequently-updated "CPU Bang for the Buck" page. Basically, I'll continue to run widely recognized benchmarks like WinBench 99 against various processors, and then divide the results by the typical street price of those processors. I'll have to do separate charts for integer performance versus floating point, because different CPUs have different strengths and weaknesses, so using an arbitrarily weighted combined number would be unfair and misleading. * * * * * I finally received a response from pair Networks yesterday afternoon. Their message and my response to it follows:
Thank you for your courteous response. It is nice to finally hear from a human, however belatedly. However, issuing a credit "sometime in the coming weeks" is not an acceptable resolution to this problem, morally or legally. If you are going to bill someone's credit card, you need to bill it for the correct amount in the first place. If you do make a mistake, you should issue a credit the same day that you are notified of that mistake. In the mean time, because of your delay in correcting your mistake, I will have pay the credit card company the full amount you billed me. You billed my card quickly enough, and issuing a credit to my account takes no longer than issuing a charge against it. Please ask your billing department to issue me a credit in the full amount today and send me email to notify me that they have done so. Several of my readers have suggested that I simply dispute the charge or request a chargeback on my account. I will not do (a) because I've wasted enough of my time on this already, and do not want to waste still more dealing with the bank, and (b) because your threat to discontinue service to an account that requests a chargeback raises the possibility that you will discontinue service for my account if I simply refuse to pay the amount that you should never have charged me in the first place. This is unconscionable. Your error and your subsequent failure to rectify it immediately has created a situation in which I lose no matter what happens. I have now wasted much more than $40 worth of my time to get your company to correct a mistake that should never have been made in the first place, and having been made, should have been corrected immediately on the basis of my initial message. In fact, even if you issued a credit for the full $234.97, I'd still be out of pocket on the basis of the time I've wasted on this. In short, your mistakes have cost me both time and money, and simply issuing a credit for $40 at some indefinite future date is a completely inadeqate response. Your company has caused the problem, and I end up having to pay the price. I do not have time to change my web service provider yet again, or I would do so in a heartbeat. Your failure to respond in a timely manner has made your 30-day guarantee absolutely meaningless, as I no longer have time to change providers within that limit, even if I had the time to spare to do so. Your company's errors and non-existent customer service has turned what should have been a relatively pleasant experience into a nightmare, and you may be sure that I will do everything I can to make sure that none of the readers of my website or my books ever consider signing up for your services. * * * * * This from Chuck Waggoner [waggoner@gis.net]:
Yes, it is pretty disgusting. I won't allow pair Networks to delay any longer. When I get my credit card statement, I'll have my wife contact the bank and request a chargeback on the basis that the charge is unauthorized, which it is. I may check into the possibility of filing a criminal complaint against pair Networks on the basis of fraud. I won't resort to lawyers, though. I have some friends in the Pittsburgh area who spend a lot of time riding motorcycles. I may ask them if they'd mind dropping by pair Networks and having a heart-to-heart chat with their billing department. * * * * * And this followup from Chris Carter at pair Networks:
* * * * * This from Frank McPherson [frank@fmcpherson.com] www.fmcpherson.com
Exactly. The following quote from www.pair.com/billing describes their policy on credit card disputes: "We are happy to post refunds and adjustments to cards that we have processed. It works best for everyone involved if you contact us and allow us to make the adjustment, before disputing the charges with your bank. Disputing the charges typically takes two to six weeks and creates confusion, whereas a posted refund is immediate. The cost to us is the same in either case. Please note also that upon receiving a chargeback notice, we reserve the right to suspend services while we investigate the cause, as chargebacks often arise from fraudulent credit card usage, which is a very serious matter." That "immediate" posted refund they mention turns out not to be so immediate, as my experience shows. And the cost to them may be the same in either case, but as anyone with a merchant account will tell you, a pattern of excessive disputes and chargebacks annoys the bank. It may result in the bank charging higher transaction costs and, if it is not corrected, the bank may ultimately revoke your account. And the sanctimonious explanation they give for threatening to suspend services really annoys me. What would they lose by leaving the site active until the problem was resolved? By threatening to suspend service, they essentially force people to give up their rights to dispute incorrect charges against their credit cards. What really concerns me is Chuck Waggoner's message about what happens when the dispute period expires. I don't know what the period is on my card, but you may be sure that I will take action before it expires. And I will contact the bank that pair Networks ran my charge through to complain about their actions, policies, and threats.
|
|
Sunday |
Sunday, March 21, 1999 Tom Syroid and I have been corresponding privately about the configuration for his proposed new system. He plans to install a CD-R drive and use it, among other things, for backup. I commented that that could get pretty expensive because of media costs. He said he'd just bought 20 CD-R blanks at Costco for about $30CA, which given the difference in exchange rates equals something around $1US per blank or less. I told Tom that I remembered a reader telling me that Canada had passed a "music tax" that would charge (as I remembered) a $0.50 tax per 15 minutes of recording time on blank digital media, which translates to a $2.50 tax per blank CD. This was news to Tom. I know I'm not making this up, but I can't remember who told me or when. I think I've read something about this since in the on-line news that said that the tax wasn't going to take effect immediately, and that there was a shortage of blank disks because everyone was stocking up. But I did think that the tax took effect January 1, so I can't explain why Tom was able to buy disks at $1.50CA each. * * * * * This from Warrick M. Locke [warlocke@mail.wf.net]:
Hmm. Perhaps the message you forwarded applies to IE5 running under Windows 9x, but I could find no evidence of any of this with IE5 running under Windows NT. I couldn't find a folder named \windows\application data\identities. I did find the folder \WINNT\Profiles\thompson\Application Data, but it contained no subfolder named identities. It contained only the Microsoft subfolder, which contained the subfolders Internet Explorer, Outlook, and Shared. None of these contained anything except routine configuration and data files. Nothing identity related. As far as the registry, the key he named contains three value entries on my computer. The value entry named (Default) has the data (value not set); the value entry named Last User ID has the data {00000000-0000-0000-0000-000000000000); the value entry named Last Username has the data "". * * * * * This from Bo Leuf [bo@leuf.net]:
* * * * * I started to write a reply to this message, but it turned into a rambling essay on stuff only distantly related to what Bo was talking about. The essay follows: Well, I know that using the words "Sweden" and "free market" in one sentence is tough, but it seems to me that if the banks have any freedom to decide their own policies on this issue then SET must be doomed. All it will take is one bank that is willing to allow its credit card holders to transact business on the Internet, and all other banks will begin losing business to that bank. I think most people have not yet realized that doing business on the Internet is a quantum change from old methods, one that has dramatic implications for anyone who continues to do business the old way. Electronic commerce removes friction from the interaction between buyer and seller. For example, it used to be that when I wanted a book, I'd get in the car, drive down to the mall, and visit the bookstore. If they had the book I wanted, I paid the price they asked. If they didn't have the book, I came home with a different book or with no book at all. Or I might have stopped in a competing book store nearby, but probably not. Comparison shopping simply took too long. Someone else might have had exactly the book I wanted, or had it at a lower price, but discovering that information took more time than it was worth. Nowadays, when I want a book, I connect to Amazon.com and spend a few minutes checking reviews, reading about competing books, etc. If I don't like the price Amazon is charging, it's easy enough to check out the competition. In the long run, all this means that consumers will have more choices and pay much lower prices. But it also means a sea change in retailing. Brick and mortar stores will not be able to compete on selection or on price. Increasingly, they will not be able to compete on service, either. One of the biggest remaining elements of friction is per-transaction costs. Existing commerce mechanisms focus on completing a relatively small number of relatively large transactions efficiently. If you buy $100 worth of clothes at the mall, a $0.25 charge for that transaction is no big deal. But if I want to buy a $.01 view of a web page, any kind of per-transaction charge clearly makes that impossible. What we desperately need is a mechanism designed to efficiently process quadrillions of miniscule ($0.001 to $.10) transactions. When that mechanism arrives--and technologies like SET and smart cards may be the basis of it--everything changes overnight. What scares a lot of businesses is that this will make Darwinian mechanisms much more a factor because things that are now bundled, with the strong subsidizing the weak, will now be unbundled. Everything will have to survive on its own merits or perish. Take magazines as an example. When you pay, say, $3.00 for a magazine, you're buying a bundle of articles (and advertising). Some of those articles are good and useful, while others are poor and useless. Which is which, of course, depends on the reader. But the point is that you have to buy all the articles as a bundle, and you're paying equally for all of them. What if those articles were unbundled and you could buy just the ones you were interested in? If you wanted only one article and could buy it for $0.10, you would save $2.90. But you could then spend that $2.90 buying 29 other articles from 29 other magazines, or whatever. The total size of the pie would not be reduced, but the strong articles would prosper at the expense of the weak. Which is as it should be. The same thing applies in many other areas. Take television shows, for example. Right now, they're paid for in bulk. Advertisers write big checks and run their ads in the midst of the program. But that mechanism exists only because it's a lot easier for the content providers to make their money by accepting one big check from an advertiser rather than trying to collect a few cents from each of the millions of people who view the episode. But what if the friction of processing millions of small transactions disappeared? I'd happily pay $0.25 per week to watch Buffy if I didn't have to suffer through the commercials. We're seeing this happening with music right now. Traditionally, the music bundle has been an album. You paid the price of the album even if you wanted only one piece of music from it. But MP3 will be the death of that mechanism, and the death of the traditional music business as well. The friction of packaging and distribution ensures that CDs continue to sell for $10 or $15, very little of which is ever seen by the people who actually create the content. But MP3 and the Internet mean that content creators can eliminate the middleman entirely. Once a micro-money mechanism is in place, the Rolling Stones can bring up an electronic commerce site and sell their music song-by-song directly to their listeners, and make a lot more money doing it that way than they've made under the old distribution methods. But the Rolling Stones never did have a problem with distribution. Where this kind of distribution mechanism really pays off is for lesser-known content providers who have no record contract. The Web means that everyone can own a printing press, and that every band can compete on an equal footing with the big names. Almost everyone wins here. The content providers, because they'll actually make more money than they do now. The consumers, because they'll pay less for the same music. The banks and the ISPs, because they're paid for their services. The only losers are the record companies, because the value they've added has always been only packaging, promotion, and distribution--things peripheral to the music itself. And the same thing is likely to happen with book publishing. The advent of the electronic book is very bad news for traditional book publishing and distribution companies, many of whom are now taking pre-emptive steps to control electronic publishing. They see it on the horizon, they know it is a deadly threat to their core businesses, and they will do what it takes to prevent it from happening. They're bound to lose this fight, though. The day that Sony introduces a BookMan is the day that the death knell sounds for traditional publishing. Once again, the economics of publishing are interesting. Most people don't realize just how little the author actually earns for each copy of his book that sells. Take a Dummies title as an example. The book retails for $20, and probably sells to the distributor for half that. On that $10 net, the author makes a royalty that may vary between 2.5% and 15%, with the lower figure much more likely on high-volume titles like the Dummies series. Amazon discounts that $20 book by 20% to $16, leaving them with about $6 gross margin. The publisher makes $10 or so gross margin on the book, which covers editorial, production, and fulfillment costs. The author probably makes $0.25 to $0.50 on each copy sold. The economics of fiction publishing are somewhat different. A 15% royalty is the norm on hardback sales, and it's calculated on list price. That means that the author earns about $3.75 for each sale of a hardback copy of a $25 novel. But fiction is fungible, and few hardbacks sell more than a few thousand copies. Few sell for long, either. Most hardback fiction sells the vast bulk of its total sales within a month or two after it hits the bookstores. So consider the actual costs involved for the key elements required to produce a book. The author makes only pennies per book sold. Editorial costs typically add another dollar or two. The entire remainder goes to things like the cost of paper, printing, shipping, and distribution. It's an incredibly inefficient mechanism that exists only because it has been the only effective way to distribute content to a mass readership. But electronic books have the potential to change all that. Removing the inefficiencies of production and distribution would allow that $25 novel to be sold directly for, say, $5 or $8. That amount would pay for editorial services (still required), and still allow the author to earn more than he does under the traditional publishing mechanism. If a $100 Sony BookMan becomes ubiquitous, many people would choose to download the latest Clancy or Grisham directly from the author's web site for $5 rather than pay $20 or more for a paper copy of the book. There are many issues to be resolved, of course, notably copy protection, but all of them are well within the capabilities of existing technology. And electronic books would solve the out-of-print problem, which tax law changes under Reagan exacerbated. As things are now, the average novel goes out of print within a few months (or even weeks) after it is published. But electronic books would allow authors to maintain an electronic backlist without inventory tax obligations. Once the $5 or $8 sales of Clancy's latest became a trickle, he could move that book to his backlist and price downloads of it at $1 or $2 or whatever, allowing it to continue to earn revenue. This is analogous to the current hardback to paperback transition. Once a new novel has sold all the copies it's likely to sell in hardback, the publisher brings out a paperback edition. That paperback may sell in huge numbers, but the author makes only $0.10 or $0.15 royalty per copy sold. By maintaining their own backlists and distributing them electronically, authors could continue to profit from ongoing sales of their older work. And, for that matter, books that are distributed electronically don't necessarily need to be read in that form. As electronic content proliferates on the Internet, expect hardware manufacturers to begin introducing printers designed to cope with it. In five years, I fully expect to be able to go out and buy a $300 HP printer/binder that will take electronic content and produce a full-color bound copy from it for a dollar or two in materials. Once that happens, traditional publishers are in deep trouble.
|