TTG Home » Robert Home » Daynotes Journal Home » Journal for Week of 30 April 2001
|
Week of 30 April 2001 Latest
Update: Friday, 05 July 2002 09:16 |
|
Monday, 30 April 2001 [Last
Week] [Monday] [Tuesday] [Wednesday]
[Thursday] [Friday] [Saturday]
[Sunday] [Next Week] |
|
| I'm covered up this morning with the regular
Monday morning stuff--running web access stats, rotating weekly web pages,
and so on. Barbara and I had decided to start having regular weekly
planning/progress meetings, but in retrospect Monday morning seems the
wrong time to hold those. We've decided that Monday lunch is a better
time, because that'll let us get our regular first-of-the-week stuff out
of the way before we meet.
Yet another cut in CPU prices this morning, as AMD not only matches the recent Intel cuts, but goes beyond them. Here are the unofficial new Athlon/Duron prices (not yet confirmed by AMD):
These are quantity 1000 prices for OEM processors (no heatsink/fan and a short warranty), but the prices for the retail boxed processors are unlikely to be all that much higher once the price cuts propagate. I am flabbergasted by the low cost of fast processors, both from AMD and Intel. This price war can't be doing any good for either bottom line, but we consumers benefit, at least in the short term. The true marginal cost to AMD or Intel to produce one more processor, of course, is only a few dollars. But it's the overhead that's going to kill them. Both of them spend billions of dollars on designing processors and building fabs, and those costs need to be recouped, which they aren't doing by selling processors for $40. Fortunately for both of them, a lot of their unit sales are up toward the more expensive end of the line, which I would guess does generate sufficient revenue to recoup design and fab costs. One thing is certain, though. The days when Intel had 70% gross margins are gone forever. So which processor should you buy? It depends. In the real world, the $183 Athlon/1333 is about twice as fast as the $40 Duron/750. As always, you'll pay more proportionately for the higher speed processor. If the bang-for-the-buck ratio was the same across processors, you'd pay about $80 for the Athlon/1333 (or $91.50 for the Duron/750). In percentage terms, you pay a lot for a relatively minor bump in performance. The Athlon/1333 is twice as fast as the Duron/750, which translates into a computer that "feels" noticeably faster, although probably not twice as fast. But in real dollar terms, that large percentage difference translates to only $143, which many people will think worth spending for a processor that's twice as fast. At these price points (and those of the corresponding Intel processors) I have no argument with whichever processor you choose. If it were me, I'd buy the Duron/750 for a computer that was to be used primarily for web browsing, email, office productivity apps, etc., and buy the Athlon/1333 for a system that was likely to be CPU-bound, e.g. a software development box or one that would be used for heavy multimedia or graphics apps. But what Intel and AMD are actually accomplishing when they set price points this low is commoditizing processors. When an entry-level processor cost $150 and a performance processor $750, there was a choice to be made. That's no longer true. More and more people are realizing that it doesn't matter much whether there's an Intel or an AMD processor under the hood. More significantly, more and more people are realizing that even the slowest current processors are more than fast enough for 99% of tasks. For example, I cheerfully continue using a Pentium III/800 in my main system, even though I have literally a dozen or more faster processors lying around. This can't be good, either for Intel or AMD.
There's an interesting article on The Register this morning. Microsoft will do anything they can to kill Linux, of course, and this article details their latest heavy-handed tactics to do so. Microsoft is paying bounties to computer resellers who rat out customers who've ordered PCs without an operating system installed. If you order a bunch of Microsoft-free PCs, Microsoft is likely to show up at your door to do a software audit. Since any reasonable person must realize that all computers run Windows and that ordering a computer without Windows pre-installed means you're planning to pirate a copy of Windows, I'm completely behind Microsoft on this one. The nerve of some people, trying to buy a PC without paying the Microsoft Tax. Something must be done or before you know it Microsoft will actually have to produce software that people are willing to pay cash money for. Click here to read or post responses to this week's journal entries Click here to read or post responses to the Linux Chronicles Forum [Top] |
|
|
Tuesday, 1 May 2001 [Last
Week] [Monday] [Tuesday] [Wednesday]
[Thursday] [Friday] [Saturday]
[Sunday] [Next Week] |
|
| Reader Tom Brosz sends in a report that
describes his new home-brew PC project. He built a dual-processor Pentium
III/933 Ultra-160 SCSI box on an Intel OR840 motherboard with 512 MB of
Rambus PC800 RDRAM. That's about the fastest PC-class system I can
imagine, and certainly faster than anything around here. You can read
about his experiences here.
Tom's experiences with Outpost.com, incidentally, echo my recent warning
about them.
A tragedy was reported in the paper this morning. A young man drowned at a local lake, and the sheriff's department is being criticized for the actions (or lack thereof) of one of its deputies. Apparently, the drowning victim and two other men, all of whom were poor swimmers, had swum out to a buoy about 50 yards off shore. The men were tired and resting by hanging on to the buoy. The deputy cruised over in his patrol boat and attempted to warn them away from the buoy, apparently because many power boaters make a habit of running over the buoys and so being close to the buoys is a very dangerous place for a swimmer to be. Unfortunately, none of the men spoke English very well, and the deputy had little or no Spanish. Eventually, he got his point across and the men began to swim back toward shore. The 17 year old man who died was apparently too tired or too poor a swimmer to make it back to shore. He began floundering and went under. The sheriff's deputy threw the men a line, but it was too late for the victim. The criticism of the sheriff's department comes from those who believe that the deputy should have jumped in and attempted to save the man, which he did not. The sheriff's department, reasonably enough in my opinion, points out that the deputy might easily have drowned as well. He was fully clothed, including body armor, boots, and weapons, and would have sunk like the proverbial stone. Even had there been time for him to strip down before the victim disappeared from sight, the deputy was not a lifeguard, nor was that the purpose of the patrol boat. The beach is clearly marked with warning signs that there is no lifeguard and that swimmers do so at their own risk. In retrospect, it would clearly have been better for the deputy to take the men aboard his boat and deliver them safely to shore--if indeed his boat was capable of carrying them--or at least to throw them life preservers. But the language difficulties meant that the deputy apparently didn't realize that the men were in trouble. Surely had he realized that he would have taken some other action than encouraging them to swim for shore, which was, after all, only 50 yards away. It's the job of cops to protect civilians, and almost without exception the cops I have known take that responsibility very seriously indeed. Compounding the disaster, in my opinion, was the tone of the newspaper article. Although the article never said so directly, the implication was that the sheriff's deputy had done less than he might otherwise have done because the swimmers were Hispanic. I find such an accusation, even by implication, to be contemptible in the absence of any evidence. I'm sure that the deputy feels bad enough already, without being accused of being motivated by bigotry. Click here to read or post responses to this week's journal entries Click here to read or post responses to the Linux Chronicles Forum [Top] |
|
|
Wednesday, 2 May 2001 [Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week] |
|
| Dr.
Keyboard commented yesterday that he'd scored a 15 of 20 on the Serial
Killer Test, so I decided to visit that site and take the test myself.
In what will no doubt come as a surprise to some of my readers, I scored a
measly 10 on the test. According to the scoring key, that means "Not
to worry -- you're not a threat to society. Although a few quirks in your
personality may present some challenges in life, those same idiosyncrasies
may very well explain the reason your name continues to appear on
"A" party lists." So I guess the world is safe from me.
If you're running IIS5 on Windows 2000, you need to read this security bulletin from Microsoft. This one is as serious as it gets, folks. By exploiting the buffer overrun vulnerability, an attacker can take complete control of the server with root privileges. IIS5 is configured by default to enable this bug to be exploited, so if you're running an IIS5 server, chances are you are at risk. Although this security vulnerability is more serious than most, it's just another in the long line of IIS security vulnerabilities, which makes me wonder why any admin in his right mind would put an IIS server up even on a private Intranet, let alone on the public Internet. If you're running II5, I suggest the following plan: (a) right now, download and apply the patch, and (b) as soon as you finish that, start migrating your web sites to Linux/Apache immediately.
I spent all day yesterday working on Barbara's and my e-book. I started at about 9:00 a.m. and, with the exception of making breakfast for Mom and outing the dogs occasionally, worked straight through until 6:00 p.m. It's an interesting change, going from thinking and writing about PC hardware to thinking and writing about poisons. I have a lot of stuff roughed out now, and Barbara is going to start flowing text into the framework. We may still be a year away from releasing the book, but I want to get the first one in the series right. Click here to read or post responses to this week's journal entries Click here to read or post responses to the Linux Chronicles Forum [Top] |
|
|
Thursday, 3 May 2001 [Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week] |
|
| When Barbara and I walk our Border Collies
after dinner, she takes Malcolm on the leash (because otherwise he might
run in front of a car), I take Kerry on the leash (because otherwise he
might forget where he lives and dodder off), and Duncan runs free. It
occurred to me the other night that some people must really think Barbara
and I are strange, because we shout directions to Duncan. "Duncan,
stand still!", Barbara shouts when a car approaches, and Duncan does
indeed stand still. "Duncan, go home!", or "Duncan, go to
Bob!" Barbara shouts when a jogger approaches, and Duncan does indeed
head for his own yard or come to me so that I can hold his collar.
But it's one thing I shout that I suspect makes some people question my sanity. As Duncan trots down the street and approaches each cross street, I shout "Duncan, look both ways!", which he does. Actually, being a smart Border Collie dog, Duncan usually looks both ways before he trots across a cross street anyway, but he does sometimes forget, so all I'm doing is shouting a reminder. But I suspect that many people don't realize that it is indeed possible to train a Border Collie to look both ways before he crosses the street, and so think I'm very odd.
Here's a very disturbing spam I received yesterday. It's bad enough that NSI sells our domain name record information to spammers, but now one of those spammers has turned around and is reselling that information on CD to any moron who has $1,024.95 and a desire to spam millions of domain owners.
Speaking of domain registrations, I see that pair Networks, which hosts my web sites, is now offering domain registrations. They charge $19/year, which is somewhat more expensive than Joker.com, with whom I currently have three domains registered, but pairNIC has several advantages. For one, I can manage all of my domains that are registered with pairnic from one central control panel. For another, it makes it very easy to add a domain to my existing account without going through hoops to get DNS sorted and so on. Also, pairNIC makes it very easy to transfer domains from other registrars and transferring a domain loses you nothing. For example, I just renewed ttgnet.com and hardwareguys.com on NSI in February. If I want to transfer those domains to pairNIC, all I need do is initiate the transfer on the pairNIC web site. I'll get a confirming email that lists a URL and includes a password. I visit that URL, enter the password, and the transfer is complete. What's very nice is that pairNIC extends the old expiration date by a year. So, for example, those domains that I paid to renew in February currently expire in February 2002. When I transfer them to pairNIC, I'll pay one year's renewal, and the domains will expire in February 2003 rather than May 2002. I assume that pairNIC is taking the hit on that.
I really must do something about network access in the den. I'd been using the 802.11b/WiFi card in my notebook, but that stopped working for some reason. The alternatives are to (a) continue connecting my notebook to the physical LAN in my office to sync it, (b) troubleshoot and repair the 802.11b problem, or (c) install a hardwired LAN connection in the den. Ultimately, I'll need to figure out what the problem is with the 802.11b link, but I'm afraid that'll take more time than I have available at the moment. There's already a modular faceplate in den behind the sofa (one of those four- or six-position faceplates that accepts snap-in modules) so the only thing that needs doing is running a cable from the den to my office. The only thing that's a pain is that the den is paneled and there's a horizontal 2X4 ledger/nailing strip about halfway up the wall, so the usual plan of dropping a cable between studs won't work. I solved that problem when I ran the phone lines to the den by using my 6-foot (~ 2-metre) drill bit to drill a hole through the ledger strip. I just hope I left a pull string in that outlet box. Also, the current phone wiring runs from the den up to the attic and then down to the basement where the telephone switch is located. That shouldn't be a problem, though. I'll just run the LAN cable the same route to the basement and then up through the hole in the floor plate to the modular faceplate in my office. One remaining problem, though. I just realized that, although I have literally a mile or more of boxed Category 3 cable in the basement, I don't think I have any Cat 5 at all. I'd hate to buy a 1,000 foot box of Cat 5 when all I need is 50 feet or so, so perhaps I'll just cable it with Cat 3 and run the link at 10BaseT rather than 100BaseT. It really didn't make a very noticeable difference running WiFi at 11 Mb/s, so I suspect I won't notice much difference running 10 Mb/s versus 100 Mb/s either. About the only time I could tell the difference was when I copied a huge file to or from the network. Other than that, 10BaseT is plenty fast enough. Click here to read or post responses to this week's journal entries Click here to read or post responses to the Linux Chronicles Forum [Top] |
|
|
Friday, 4 May 2001 [Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday]
[Friday] [Saturday] [Sunday]
[Next Week] |
|
| Please give a moment's
thought today to Alison Krause, Jeffrey Miller, Sandra Scheuer, and
William Schroeder. Thirty-one years ago today at Kent State University,
the Ohio National Guard murdered those four students and wounded nine
others. Two of those dead were merely spectators at the anti-war protest
that occurred that day. The other two were minding their own business on
their way to classes. No students were armed. No Guardsmen had been
injured, nor were they in any danger. The Guardsmen fired randomly for 13
seconds--13 seconds--into a group of unarmed civilians, killing students
as much as 700 feet away. No one has ever been brought to justice for this
massacre, nor even suffered administrative discipline.
Reader Paul Robichaux tells me that, in a surprising display of honesty, Microsoft admits that having too many users in a Windows 2000 group is enough to make your hair turn gray. Anyone who's administered a large Microsoft network will appreciate the truth of that one, although some might argue that many fewer than 500 users are needed to make an administrator's hair turn gray.
The more I work on this e-book on poisons, the more I'm aware how wrong is the conventional wisdom that says everything is available on the Internet. Not even close. In fact, the Internet is more notable for what's not there than for what is. At this point, the Internet as a whole falls far short of being a complete repository of durable human knowledge. How short? Who knows? At a guess, perhaps five or six orders of magnitude. Perhaps more. What's there is simply a gloss overview of knowledge. Nor am I speaking of ephemera, but only solid information of lasting value or interest. If ephemera are taken into account, the shortfall is probably another dozen orders of magnitude. And even ephemera can be useful. If I'm writing a book on Jack the Ripper, for example, I might like to be able to look up the weather for London on September 30, 1888. Or a complete list of train schedules for that same day. Or a list of ship arrivals and departures, including rosters of passengers and crew. Or that day's arrest reports and suspect interviews for every precinct in London. Or be able to call up and read the entire Times for that day. Much of that data no doubt exists somewhere, but if it's on the Internet I can't find it. Multiply that by a trillion or whatever and you get some idea of what's not on the Internet. What's there is useful, certainly, but so far from complete that it beggars the imagination. For more about this from a professional researcher's perspective, see Barbara's page for today.
I've mentioned before that Adaptec/Roxio Easy CD Creator is on my not-recommended list. To reinforce that advice, this article from The Register begins "The latest version of the popular CD recording software Easy CD Creator, version 5, is killing Windows 2000 machines stone dead". Indeed. I'll say it again. If you have a CD burner on a Windows 2000 system, don't install Adaptec/Roxio software for it. Instead, get yourself a copy of Nero Burning ROM. It's fast, stable, and robust, and it just works. In fact, I recommend Nero for any computer running any Windows version. It's simply better than anything Adaptec/Roxio has ever shipped. This is just the latest in a series of Windows 2000 related problems, which makes me think the fundamental problem is with Windows 2000 itself. For example, Matrox, which was famous for writing rock-solid video drivers, has been trying for more than a year now to ship solid Windows 2000 drivers. Their latest are a lot better than their earlier efforts, but are still not up to par in my opinion. HP is seemingly unable to release any Windows 2000 drivers at all for several of their products, and many other companies have released Windows 2000 drivers that are deficient in stability or in functionality, or both. Many of my industry contacts have told me privately that writing drivers for Windows 2000 is a bear, and I've heard that often enough to believe that it must be true. Other than for notebooks, where the power management features of Windows 2000 are useful, and USB support, I see no reason whatsoever to install Windows 2000. All of my Windows NT 4 systems, including many that run older hardware, are rock solid, with the exception of one that has hardware problems. All of my Windows 2000 systems are built on recent hardware, and all have or have had various odd problems. Many of those problems may be related to ACPI, and one thing I've considered is installing Windows 2000 with ACPI disabled (choosing "standard PC" at install time) to see if that makes the problems go away. Perhaps I'll try that on the next system I build. I only wish that Microsoft would be foolish enough to release an NT4 service pack that added USB support. Then I'd never need to install Windows 2000 again. Of course, hardly anyone else would either, which no doubt is the reason that Microsoft has never released USB support for NT4, although rumor has it that they were prepared to do so as long ago as SP2. Click here to read or post responses to this week's journal entries Click here to read or post responses to the Linux Chronicles Forum [Top] |
|
|
Saturday, 5 May 2001 [Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week] |
|
| I think these web pages are okay now,
but I have a lurking fear that they're hideously corrupted. It all started
yesterday when I used pairNIC.com to
register two new domains and add them to my web hosting account. The first
domain, technomayhem.com, will be the base domain for our ebook reference
series aimed at fiction authors. The second domain, researchsolutions.net,
will be the home for Barbara's research business. Registering the domains
and adding them to my existing hosting account went fine, although I was
horrified later to read in the pair Support Forum that they'd been having
problems all day with the registration process. Oh, well. I dodged that
bullet.
The problems started when I decided to go ahead and relocate some stuff that had been in subfolders in the ttgnet.com web to two new subwebs for the new domains. I did a cut/paste from the ttgnet subdirectories to the technomayhem and researchsolutions sub-webs, as appropriate. The cut/copy (which FrontPage treats in UNIX style as a rename) seemed to proceed properly, although it took a long time because of all the links that were being corrected as the files were moved. So then it came time to publish the changes up to the server at pair. I marked the box to publish only changed pages and to publish all subwebs as well as the main web. So far, so good. During the first part of the publish process, there's a long wait while FrontPage scans first the local copy of the web and then the remote copy, deciding which files have been added, changed, or deleted. Okay, that took quite a while, as I expected. What I didn't expect was what happened next. Instead of publishing only changed pages, FrontPage insisted on republishing every single file in the web--not just HTML files, but graphics files and so on. Okay, that I could live with. But then Barbara went to work patching up her new arrangement of pages, removing links between her business page and her personal page, etc. After dinner, she told me she had done enough that I could go ahead and publish. I did, again telling FrontPage to publish only changed pages and to publish subwebs as well as the main web. Once again, it set off to parse the local and remote copies, and then began publishing--EVERY SINGLE FILE IN THE WHOLE WEB. Arrghhh. So I decided that perhaps something was out of sync. I did a local resync, telling FrontPage to verify links, etc. I then republished yet again, but this time told FrontPage to publish the entire web. I hope that fixed the sync problem, but I'll find that out when I publish this morning. If it's still not fixed, it's rapidly going to become a pain in the begonia to have to publish the whole web every time I change one file. Have I ever mentioned that I hate Microsoft?
One of Microsoft's NetMeeting pages offers the following marginally literate advice for those who want to use NetMeeting for video conferencing:
Microsoft is famous for being ruthless, but I'm surprised they come right out and suggest this. I presume they're recommending merely ceasing to associate with those unfortunates who do not have a cable modem or DSL connection, but that's not clear from context. They may instead be suggesting murdering them. "Not Guilty, your honor. It was justifiable homicide. He was using a 56K dialup!"
Spammers just get more disgusting. I've been getting junk penny stock spams for a long time, but the latest twist is truly foul. Instead of sending "To:" undisclosed recipients or some obviously bogus address, this new series of spams is sent "To:" an aol.com or earthlink.com address which appears to be a real address, like jimwhite27 (at) aol (dot) com or bobwilson11 (at) earthlink (dot) net. The message itself starts out "Dear Jim," or "Hi, Bob" or whatever. The body text tries to sound like a personal email sent from someone offering insider trading information to a friend of his that somehow (we're not supposed to wonder how) ended up in our mailbox instead of or in addition to the intended recipient. Presumably we'll all rush to buy that stock (purchase information is conveniently included) on the basis of this "insider information". Yeah, right. It's always said that you can't con an honest man, and I suppose that's true. The obvious target of these spams is people who are (a) stupid, and (b) dishonest. And, on reflection, I suppose there are enough people that are members of both categories to make it worth this slimeball's time. Oh, yeah, I forgot to mention. These spams are of an increasingly common type; those marked "high priority". What is wrong with Outlook 2000? Surely it should be able to filter a message sent using both an "X-envelope-to" address and "high priority" as almost certainly spam. Those two fields are for anyone I know mutually exclusive unless I'm missing something entirely. Perhaps something like CERT advisories might reasonably use both fields, but those situations are few and far between and could easily be handled with an exception based on sender's address or subject line.
The Register announced that Rambus has lost its legal battle against Infineon, which is good news for all of us. Of 57 claims of patent infringement originally made by Rambus against Infineon, US District Court Judge Robert Payne had dismissed 54 earlier. Yesterday he dismissed the remaining three claims, which leaves Rambus without a legal leg to stand on. Rambus will appeal, of course, but this is the worst possible news for them. It's a win not just for Infineon, but for all of the other RAM makers who've been sued by Rambus, including those who have already settled and will now no doubt be going to court to void those agreements. Rambus stock should take a major hit, and it's possible that Rambus will now find itself defending rather than attacking. Mention has already been made of prosecuting Rambus under RICO statutes, so it'll be interesting to see what happens. As far as I'm concerned, it couldn't have happened to a more deserving company. I hope that the legal system eventually punishes Rambus by voiding all of their patent claims and putting those patents in the public domain. Click here to read or post responses to this week's journal entries Click here to read or post responses to the Linux Chronicles Forum [Top] |
|
|
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week] |
|
| Our two new domains, researchsolutions.net and
technomayhem.com, are responding to DNS queries, and I've already set up
mail for them. There isn't much up yet on either of them--technomayhem.com
has just a placeholder index page--but at least they're registered and
working now.
I've been mucking about this morning trying to get the copyright notice back on all the pages. That notice is contained in what FrontPage calls a "shared border", which is text that is entered once and then appears on every page in the web for which shared borders are enabled. Until yesterday, that copyright notice appeared on nearly every page in this web, but as of today my local copy of the web has no copyright notices. FrontPage removed all the shared borders, but it was just doing what I told it to do. I'd created two subfolders in the main web, named researchsolutions and technomayhem. Shared borders operate only within a web and not across subwebs. So, for example, changing a shared border in my main ttgnet web has no effect on the shared borders in the researchsolutions or techomayhem subwebs and vice versa. I'd already converted the researchsolutions folder to a subweb and modified the shared bottom border to show Barbara's copyright notice rather than mine. Then I made my big mistake. I called up a page in the technomayhem folder and modified the shared border there. When I saved the document, I was surprised to see the hourglass appear for a long time. That usually happens only when FrontPage has a lot of pages to modify (it has to modify each page in which the shared border appears). After a minute or so I was really wondering what was taking so long. After all, there were only a few pages in technomayhem. Then I was horrified to see that I'd forgotten to convert technomayhem to a subweb before I changed the shared border. Technomayhem was still just a folder within my main web, which meant that FrontPage was merrily changing the shared border in every single page in the main ttgnet web. Arrghhh. So this morning I'm fixing that, which involves long waits as every page in the web is re-reconverted to show the copyright notice. And, of course, that also means I'll have to republish the entire web. Talk about shooting myself in the foot.
I noticed this in Cringely's column yesterday: "Microsoft, in its ultimate wisdom, has embedded code in [Windows Me] to make it impossible to make duplicates. After first burning a bunch of coasters, my spy reloaded Windows 98 and made the backup copy. There's a free tip for the anti-licensing lobby." If that's true, and I have no reason to believe it isn't, it's yet another good reason to avoid Microsoft upgrades.
I have laundry and other chores to do, so I'd best get started on them. Click here to read or post responses to this week's journal entries Click here to read or post responses to the Linux Chronicles Forum [Top] |
|
