|
Week of 2/15/99 Friday, July 05, 2002 08:04 A (mostly) daily journal of the trials, tribulations, and random observations of Robert Bruce Thompson, a writer of computer books. |
|
Monday |
Monday, February 15, 1999 I wasn't feeling very well yesterday, and I'm feeling worse this morning. Some kind of virus, I suspect. That's one thing about working at home. You spend your sick days the same place you spend normal days. There's not much point to just lying around being sick, so you go ahead and work anyway. I spent most of this weekend working on the test-bed system, swapping hardware in and out and running various benchmark tests. I've been working a lot with the Promise FastTrak IDE RAID Controller, setting up, benchmarking, and then tearing down various arrays. I'll have a full report later, but for now I'll just say that this $99 card is letting me do things that I used to think required SCSI. The only problem I've had with it is getting it configured to allow the system to boot from the array, but that may be due to a mistake I've made rather than to a problem with the card itself. Once I've finished testing it, I plan to move it into my resource server and begin a long-term evaluation of it. * * * * * Late Afternoon: I've spent the day working on my current chapter, and gotten a surprising amount done, considering that I'm feeling rotten. I exchanged some mail with my contact at Integrated Device Technologies (IDT), the makers of the WinChip processor. I have some eval CPUs on the way, including one unreleased product. I can't talk about that one until IDT gives me the go-ahead, but it does sound like a very interesting product. In the mean time, if you have an old Pentium that you'd like to upgrade without spending a lot of money, check out the IDT WinChip line. For $50 or $75 you can upgrade an old 75 MHz or 90 MHz P54C Pentium system to twice or thrice its original speed. It still won't be a Pentium II, but it won't be all that far behind, either. WinChip floating point performance relative to the old Pentium remains inferior, but other than that the WinChip vies with additional memory as one of the most cost-effective upgrades available. Sadly, there's no WinChip available for the old 5.0 volt P54 60 and 66 MHz Pentiums. Incidentally, in the interests of full disclosure, I'm saying all this on the basis of information published by IDT and on feedback I've had from readers. I've not yet worked with a WinChip myself, but I'm looking forward to correcting that oversight soon. * * * * * Regarding the change to thisweek.html and lastweek.html, Tom Syroid made the following comment on his web site, to which I emailed him the reply that follows:
Well, one stumbling block that I ran into almost
immediately is that if you're going to maintain two identical pages (e.g.
thisweek.html and 0215RTDN.html) you can't just do a "Save-As"
to keep things in sync. For example, I called up 0215RTDN.html in FP
Editor, made changes, saved the changes under the current file name, and
then did a "Save-As" thisweek.html. That appeared to work fine.
But then I noticed a problem: the day-of-the-week links down the left
column in thisweek.html were still pointing to 0215RTDN.html. For example,
in thisweek.html, Tuesday should have been linked to
http://www.ttgnet.com/thisweek.html#Tuesday. Instead, it was still linked
to http://www.ttgnet.com/0215RTDN.html#Tuesday. Bummer.
|
|
Tuesday |
Tuesday, February 16, 1999 Important BigBiz.com (my web hosting service) is moving me to a different server, presumably today. That means I'll have a different IP address. Until that change flows through DNS, you'll still get to the old site when you use www.ttgnet.com. You can get to the new site by entering the new IP address, 165.90.95.251, directly into your browser in the form http://165.90.95.251 At this point, it's not clear when my site will be moved to the new IP address or when I'll next be able to update my site, so you may not even see this message until it's overtaken by events. If I am able to update the site, and if you do want to read the current stuff before the change takes effect, entering the new IP address will get you there. In the mean time, you don't need to do anything special. Eventually, the change will flow through DNS, and hitting www.ttgnet.com will take you to the new site. It may just take a couple of days for that to happen. I apologize for the inconvenience. I was having problems retrieving mail from my mail server again yesterday morning, so I sent a message to BigBiz.com to complain. By coincidence, the following mail exchange started with my friend Paul Robichaux [paul@robichaux.net] right around the same time.
Actually, I've kind of let it drop. I'm debating whether to stay with BigBiz or move to pair Networks. Until I decide, I figured it wasn't worth relocating to a different server at BigBiz. If I do decide to stay there, I'll certainly ask them to move me to another server.
Hmmm. Thanks for telling me that. When they mentioned moving me to server 05, they told me I'd have to move all my stuff manually. I'll contact them.
Maybe so. I've been a pain in the butt to them. That reminds me, I haven't looked at your site lately. I'll go there now...
Hmm. Getting the time to write them is less of a problem than it might seem. Just cut your sleep time by a couple hours a day and eat all meals in front of the computer. As far as not having interesting things to say, I never let that stop me... Then, at 10:21 p.m. I received the following response from BigBiz to the complaint I'd sent them that morning about the mail server:
Okay, moving to another server sounds good. I was under the impression that I'd have to reload all my stuff, but my friend Paul Robichaux tells me that you moved him overnight. My web site is all static pages, so presumably you can do the same for me. You'd mentioned web05 at one point. Is that still your newest server and the one with the lightest load? If so, I'd appreciate being moved there. Tell me what I need to do. Then, when I retrieved my mail this morning, I found the following waiting:
Okay, thanks. I note that I'm moving from 165.90.46.122 to 165.90.95.251. I presume the issue is that DNS will still be resolving www.ttgnet.com to the old address for a while, so any uploads I make would be going to the old location rather than the new. Even so, I'd like to update the original location to tell my readers that I'll be down for a short time, and perhaps to give them the new IP address so that they can use that in the interim. Is that okay? Also, once you get my stuff moved from the old machine to the new, I'd like to update my site using the IP address until DNS starts resolving it. Is that okay? I assume that my mail will continue uninterrupted, right? * * * * * This from Bill Grigg [bill_grigg@bc.sympatico.ca] from Vancouver, Canada:
I used to have a Spotmatic F myself. It was a nice camera, and I can't remember for the life of me what happened to it. I must have sold it. But my original 1971 Honeywell Pentax Spotmatic does have a self-timer and flash sync connections. As I recall, the main difference with the Spotmatic F was that it had full-aperture metering, while the Spotmatic had stop-down metering. Of course, Asahi had different distributors in Canada and the US, and used different names and model designations. You're right about the f/1.4 versus f/1.8 or f/2.0 issue. That extra half stop from f/1.8 to f/1.4 was a pretty expensive step, but the f/1.4 lens was a lot sharper at f/1.8 than the f/1.8 lens was wide open. As I recall, the f/1.4 Super-Takumar lens cost about $50 more than the f/1.8 version, but it was and is a good lens. That was an era of fast lenses. Although I believe that Pentax never went wider than f/1.4, f/1.2 lenses were available for Nikons and others, and someone (Canon, as I recall) actually shipped an f/1.0 monster.
|
|
Wednesday |
Wednesday, February 17, 1999 Well, the server move appears to have been done successfully without incident. I got mail from InterNIC yesterday morning asking if it was okay to change the DNS server addresses. Actually, Barbara got that message and forwarded it to me. That's because my contact address at InterNIC is "thompsrb@bellsouth.net", which is the email account associated with our dial-up ISP account. As things are set up now, mail to "barbara@ttgnet.com" is auto-forwarded to the thompsrb account, from which she POPs. When she forwarded me the message, I immediately replied "YES" that it was okay to change the DNS server information. I did realize that the way things are set up might mess up the InterNIC auto-parsing mechanism, so I changed the reply-to in that response to read "thompsrb@bellsouth.net" rather than "thompson@ttgnet.com". Still, I got to thinking that that might not be enough, so I went back and re-replied to the message from Barbara's computer, so that the "from" would read "thompsrb@bellsouth.net". Later that morning, my friend John Mikol--who is listed in the DNS record as the Technical Contact and Zone Contact--called to tell me he'd also gotten the message and to ask if he needed to do anything. I first told him no, but thinking about it I realized that having InterNIC get too many YES responses was better than having them not get enough, so I told him to go ahead and tell them yes. In the process of all that, I noticed that BigBiz had also put themselves down as Technical Contact in the Modify Request. I let that pass for now, although I'll change it back later. As a matter of good practice, it's best to have your own domain record 100% under your own control, and not to allow your service provider to have any control over it whatsoever. That, incidentally, is one of my main objections to pair Networks. They insist on being Technical Contact no matter what. At any rate, I've now been moved from server WEB01 to server WEB05. Apparently, a lot of the problems I'd been having--slow web server response, trouble POPping mail, etc.--were due to WEB01 being overloaded. There's an on-line auction site hosted there, and at certain times of day they just suck all the ticks out of that server. It seems to me that BigBiz should tell them they need to run on their own, co-located, dedicated server. So things appear to be back to normal, except the slow-downs should go away. My mail seems to be working properly, and you should be able to get to this site normally. The old web pages still live on WEB01, but they won't be updated now and they'll go away in a couple of days. * * * * * Every once in a while, I do learn something from the morning paper, and something they published this morning was a Dusie. There'd been some discussion by people asking about registering an automobile in the names of both husband and wife. When it's done that way, each automatically inherits if the other dies, and without inheritance or other taxes or the need to re-register the car. If it is registered, say, only in the husband's name, and the husband dies, the wife has to spend quite a bit of money to get the car back, including several hundred dollars to re-register the car in her own name. But someone wrote this morning to ask if doing it that way doesn't expose other jointly owned property, such as their home, to claims if the liability insurance on the automobile is inadequate to satisfy a judgement. The answer was that it did. If the car is registered in only one name but the house in both, the house is protected against claims. If the house and car are both registered jointly, and insurance is inadequate to cover a claim where the owner was at fault, the person who wins the judgement can force the sale of the house to satisfy the judgement against the equity in the house. Obviously, this situation varies state to state and country to country, but it's worth looking into. When my father died in 1990, the car was registered in his name, and my mother had to go through hoops to reclaim it. So when Barbara and I bought our latest SUVs, we registered both of them in both our names. As it turns out, doing that had a downside that no one mentioned to us. * * * * * A follow up from Bill Grigg [bill_grigg@bc.sympatico.ca] from Vancouver, Canada:
Yep. That makes sense. I'd forgotten about the SP1000 until you mentioned it, but I do remember it now. I haven't messed with digital cameras much, but I've enjoyed the playing around that I have done. Maybe I'm a bigot, but I still see them as toys. With traditional film and lenses, you can capture a serious amount of data. As I recall, Kodachrome II could resolve 200 line pairs/mm. That was beyond the resolution of all but the very best lenses available, but a decent lens could resolve 100 lp/mm. That's 3,600 X 2,400 line pairs on a standard 35 mm frame. I think each line pair requires an absolute minimum of three pels, so that translates to 10,800 X 7,200 in the way that digital camera resolutions are measured. That's an order of magnitude larger in lineal resolution, and two orders of magnitude higher in areal resolution than a typical digital camera today, and even that ignores color depth. With traditional photography, the resolution of the lens was always the limiting factor. With today's digital cameras, the resolution of the medium is the limiting factor, even given the poor lenses used on the things. If someone brought out a $500 digital camera that would resolve, say, 9,000 X 6,000 pels in 30-bit color depth, I'd buy one in a heartbeat. Until then, they're really just useful toys. * * * * * This from Chuck Waggoner [waggoner@gis.net]:
I did? If there's a way to take care of the problem in NT, I'd love to hear about it. The only way I know to prevent IE running under any operating system from writing "bad" cookies is to disable cookies entirely. The problem with IE is that it keeps cookies as individual files, rather than in one consolidated cookie file as Netscape does. With Netscape, one can make the cookie file read-only, which effectively prevents "bad" cookies from being written to disk. Unfortunately, it also prevents good ones from being written. But with IE, I've not found any way to stop cookie files from being written to disk. Even making the cookie folder read-only doesn't help. IE happily writes the cookie without regard to security settings on the folder. Apparently, in that respect at least, IE is functioning as a system-level service that can override permissions that otherwise restrict the ability of the currently logged-in user to access the disk. * * * * * And Chuck Waggoner [waggoner@gis.net] responds:
Indeed. I suppose there are more important things to worry about, but having these companies co-opt my hard disk to serve their own nefarious purposes really aggravates me. I still wish someone would charge them under the existing anti-cracking laws for accessing a computer and writing information to the hard disk without the owner's permission. When I explicitly visit a web site, I have given at least my implicit permission for that site to write cookies to my system for its own purposes. But I sure haven't given permission to those third-party companies that write cookies for their own purposes, unbeknownst to me. At least doubleclick.net appears to honor Netscape's blocking of third-party cookies. But Netscape blocking has no effect on those obnoxious cookies from imgis.com or NGAdcenter.com, which appear to get around the blocking by using invisible frames. It really makes my blood boil that they take active measures to write their damned cookies to the hard drives of people who have the nerve to try to keep them off their systems. * * * * * And speaking of obnoxious behavior, I got a spam email yesterday that hit a new low. This was the first-ever spam message I'd gotten that was flagged high priority. Give me a break. * * * * * And now I'm going to try publishing this to my web server. It's about 11:25 a.m. right now. When I was running on the old web server, any attempt to publish after about 9:00 a.m. my time (6:00 a.m. web server time) resulted in a FrontPage time-out. The material almost always got updated okay, but it's disconcerting to see an error message from FrontPage saying that the update failed. We'll see what happens now that I'm running on a snazzy new lightly-loaded server. * * * * * This from Richard Jones, who asks that his email address not be published:
No, actually that slang term originated with the Dusenberg automobile and was originally spelled "Dusie." But you're right that "doozie" is now the more common spelling.
|
|
Thursday |
Thursday, February 18, 1999 I got quite a bit done yesterday on my chapter, although I'm still not feeling 100% well. I have a lot to get done today and tomorrow, so updates here will be shorter than usual. The web updates went fine yesterday. I didn't get any FrontPage timeouts, which I attribute to being on the new web server, but they still took a long time to complete when I ran them during the work day. * * * * * This from Chuck Waggoner [waggoner@gis.net]:
Good points, all. But even for family snapshots, digital cameras have two significant problems to overcome in my opinion. First, obviously, is resolution, or the lack thereof. The truth is that the best digital cameras--and here I'm talking about the $30,000 Nikon-based units that newspaper photographers use--do not begin to approach the resolution provided by a $5 throwaway drugstore camera, literally. The difference is apparent to a trained eye even in 3X5 prints, and anyone can tell the difference in 8X10s. Even more important is archival stability. Silver-based processes are inherently stable. Black and white photographs I took 35 years ago are unchanged today, as are family photographs from nearly 150 years ago. Granted, dye-based color processes are much less stable, particularly those based on color-coupler technology. E4/E6 Ektachromes I processed in my home lab thirty years ago have mostly faded, but Kodachromes I processed are unchanged. (And, yes, I know you can't process Kodachromes at home. I did it anyway.) Technologies based on dye-destruction are much more stable. Cibachromes I did a quarter century ago are still stable, as are the dye-transfers. But that's not the case with digital photographs. Certainly, in theory at least, digital photographs are permanent. After all, they're just bits on a storage medium. But will they be readable in even ten years, let alone fifty? Can you read a floppy disk you made even ten years ago? I can't. I don't have a drive that will read the 5.25" floppies any more, and that's assuming that the media itself hasn't degenerated. That CD you burn with all the family photos you took over Christmas may be readable next year or even five years from now. But what happens in twenty years when you find that CD in a box of other junk and want to look at the pictures? The Mark 1 Eyeball hasn't changed much over the years, but data storage technology continues to evolve. And finally, an observation on your comment about test shots and film not showing things as they appear to the eye. Thirty years ago, they were filming an episode of Star Trek. One of the characters was a young woman who was playing an alien who was to have a greenish tinge to her skin. Makeup applied a light green tinge, and they filmed the scene, in which she appeared alone. When the dailies came back, she was pink and healthly looking. Makeup applied a darker green tinge, they reshot the scene, and waited for the dailies. Again, when the film came back, she had a healthy rosy glow. Convinced that the makeup was ineffective in moderate quantities, they turned her literally Kelly green and reshot the scene. This time, the film came back showing her with a slight greenish tinge and a note from the lab saying that they'd been doing all they could to correct the color balance on all the prints, but this time they couldn't fix it completely and could the crew please be a bit more careful with makeup and lighting. * * * * * This from Bo Leuf [bleuf@algonet.se]:
Kodachrome was a wonderful film (and presumably still is, although I haven't shot any of it for probably twenty years.) But there was nothing magic about its high resolution. It was an ASA 25 film, with accordingly small silver halide grains. Any similar speed film had similar resolving power, as for example the ASA 32 Panatomic X I shot by preference when doing 35mm black and white. And you're right about the difference in color balance between lenses. I remember one time doing some tests on that. I was then at Rochester Institute of Technology (RIT), which is actually located in South Henrietta, New York rather than Rochester. But that acronymn wouldn't have done at all. I took a 36 exposure roll of Kodachrome (II or 25, I can't remember which) and ran part of it through my Pentax with various Pentax lenses, shooting a standard color chart with electronic flash. After shooting half the roll, I rewound it, put it in a Nikon and shot the rest of the roll with various Nikon lenses. When the slides came back, I put them on a $20,000 transmission densitometer. It told me that there were indeed minor color rendering differences between the Pentax lenses and the Nikon lenses. What was interesting was that both Nikon and Pentax had balanced their lines of lenses. There were no significant differences between a Pentax 28, 50, 85, or 135mm lens, and there were no significant differences between the various Nikon lenses. But there were significant differences between the brands, at least according to the densitometer. I must confess that I couldn't tell any difference visually. I mean, it wasn't as though the Pentax lenses had a visible pink tinge and the Nikon a blue tinge or anything like that. As far as storage, you're correct. A 9K X 6K image at 30 bits would require about 200 MB to store, assuming no compression. So assuming we want a reasonable number of images on a removable disk, that means the camera needs something like a 5 or 10 GB disk for image storage. That's certainly within the capabilities of current technology (see Seagate's recent announcement). And I don't have a 60 GB desktop. Right now, I have only 53 GB of disk storage on all the systems on my desktop combined. * * * * * This from Francisco Garcia Maceda [maceda@pobox.com]:
Huh? That's like saying that if I leave my back door unlocked while visiting a neighbor and a burglar comes in and steals my TV that he's not guilty of any crime because I didn't lock the door. Give me a break. * * * * * This followup from Bo Leuf [bleuf@algonet.se]:
Hmm. Without getting into stuff like Modulation Transfer Functions and other things that I doubtlessly mis-remember, I suspect the reason that he perceived a difference was that dye blobs have soft edges whereas developed silver grains have sharp edges. Visually, most people perceive something that is lower resolution but has high contrast and sharp edge delineation as being "sharper" than something with higher resolution but lower contrast and ill-defined edges. As I recall it, the predecessor to Technical Pan (I can't remember its designation, but it was a number rather than a name. 50xx, perhaps?) had more than twice the inherent resolution of Kodachrome. As you noted, the presence of three separate emulsion layers in Kodachrome put it at a severe disadvantage relative to single thin-emulsion black and white technical films. It's not clear to me what he meant by developing it as "black and white." The normal process for Kodachrome required, as I remember it, twenty-seven separate steps, or perhaps it was thirty-four. The first was to develop the film as a standard black-and-white negative. That was followed by three separate reversal processes, each occupying multiple steps. Basically, the developed film was exposed to monochromatic blue light, which exposed the remaining silver halide in the blue-sensitive layer, and then processed in a developer that contained yellow color couplers. Same thing for green light/magenta couplers and red light/cyan couplers. The order was important, too, although I can't remember the order the colors had to be developed in. These three reversal steps put the dyes in the right places. The film was then bleached to reconvert the silver from the emulsion and filter layers to silver halide, and finally fixed to remove the silver halide, leaving only the dyes. So he may have been processing the film with monochromatic color couplers, which would indeed yield a "black and white" image, but one that was dye-based rather than silver-based. If that was the case, he indeed would have ended up with an image that had no grain, because the overlapping dye blobs would have nearly eliminated edge contrast between individual blobs. But if he was talking about simply processing the Kodachrome in D-76 or Microfine, then what he ended up with was a standard black-and-white image with silver grains. Small ones, granted, but much larger than the "technical" films of the time, let alone the high-contrast "process" films. I remember using some high-contrast copy film back then that was rated at ASA 0.5. In terms of grain, the images that resulted made standard Kodachromes look like they'd been shot on Tri-X Pan pushed to 1,600. * * * * * And still more from Bo Leuf [bleuf@algonet.se]:
Okay. It sounds from your description that he just did standard black-and-white reversal processing. I used to do that years ago with Panatomic-X to produce black-and-white slides. As I recall, reversal processing didn't work too well with anything much faster than Panatomic-X. I never tried it with Kodachrome, but basically he was just treating Kodachrome as a black-and-white film (which it was and is). * * * * * And to change the subject entirely, this from Andrew Sabel [asabel@wrtsun02.svidaho.net]:
As far as DirecPC, that sounds pretty reasonable. I'd heard some real horror stories about people receiving multi-hundred dollar bills for a month's use. I assume that they're essentially billing you by the kilo-packet or whatever for only traffic that actually comes down to you from the satellite, so time spent dialed in doesn't count. As far as the CD-R, I'm leaning toward the HP 4X IDE device, although I may yet go with SCSI. I just haven't had a spare moment to even think about it recently. Speaking purely on the basis of what I've learned from my readers, my guess is that your older system will work fine, particularly copying from one SCSI CD-ROM to another. Assuming your hard disk is IDE, even that should work fine, particularly if you don't try to do 4X copies. And even those may work fine if you're running NT and/or leave the system alone while it's doing the copy. I'll let my other readers mail you and copy me if it turns out that what I'm saying here is bad advice. * * * * * And now it's about 12:45 EST, and I think I'll try publishing this to see what happens....
|
|
Friday |
Friday, February 19, 1999 It published successfully, although it took about 15 minutes to do so. Which brings up one of the many gripes I have about FrontPage 98. I got in the habit early of not touching my computer while FrontPage was publishing, and I still do it that way. Why? Because the status bar at the bottom isn't updated properly. If you start publishing with FP Explorer full screen and then minimize it or even window it, the status bar blanks to grey and isn't updated until the next status change. Actually, even if you start the program windowed and then shift focus to another program that covers the status bar, it goes to grey. Actually, this program shows incompetent programming in many respects. Perhaps the most aggravating is its inept password module. Each time I publish, I have to enter my account name and password. The stupid program insists on defaulting to my Windows NT account name, which is not the same as my web server account name. So every time I publish, I have to not only enter my password, but delete and retype my account name. You'd think it would have been pretty simple to have it default to the last account name used. For that matter, how hard would it have been to add a "save password" box? This is the least configurable Microsoft program I've ever used, and that's probably because it's not really a Microsoft program at all. They bought it from Vermeer and the program still shows a lot of its roots. People complain about Microsoft software, but let me testify that there's a lot worse out there. FrontPage was a great idea that was ineptly executed. Perhaps FP2K will be an improvement. People keep recommending Dreamweaver to me, but I hear that program has problems of its own, including a shallow learning curve and quite a few bugs. If I had the time to think about changing programs, I'd probably take a look at Dreamweaver, but I don't. * * * * * This from Bo Leuf [bleuf@algonet.se]:
My dictionaries provide about the same information yours do, although my Collins British dictionary gives the pronunciation of Eleonora Duse's name as doozie. Most give the origin of doosie (in various spellings) as source unknown, but dating to circa 1930. But I'll stick by my derivation, because it is based on contemporaneous sources. I read mysteries by the ton, and years ago I read one form the 30's in which one of the characters in the presence of a Dusenberg says something like, "Now that *is* a real Dusie." I'd never thought about the derivation of the term until I read that. In the sycnronicity that seems to crop up in my reading, not more than a week or two later, I read another 30's mystery (by a different author) that also made reference to a Dusie in similar circumstances. And the dictionaries that suggest derivation from a daisy seem to be missing the point. There's nothing particularly extraordinary about a daisy, or about Miss Duse come to that. But a Dusenberg was the height of extraordinary. If you've ever seen a Dusenberg, you'll understand what I mean. People who drove Rolls-Royces back then did so because they couldn't afford a Dusenberg. The things were huge, powerful, and refined. They were and are the most extraordinary automobiles ever built. And, as it happens, they were (and still are) commonly referred to as "Dusies." * * * * * And this follow-up from Francisco Garcia Maceda [maceda@pobox.com]:
Blame the victim for the crime, huh? Your logic is twisted. I'm forced to conclude that you do not understand how cookies work.The point is this: When I visit a web site, I give that site implicit permission to write its own cookies to my hard drive and to subsequently access their own cookies for their own administrative purposes. But I do not give permission to Imgis or NGAdcenter--different companies entirely and ones whose sites I have not explicitly chosen to visit--to write their cookies to my hard drive. And, no, I don't know when they're doing it. That's the point. I find their damned cookies later in the cookie file, and have no idea where they came from. And to say that I can prevent their obnoxious behavior by turning off cookie use entirely or by individually approving each cookie misses the point entirely. Why should I have to cripple my web browser and inconvenience myself to prevent them from doing something that is already against the law? Jesus. * * * * * Internet Explorer 4.01 has started behaving very strangely. Sometimes I click on a link and the following text pops up in a warning box:
If I click the link a second time, the same warning repeats. If I click a third time, the page displays normally. The first few times this happened, I thought the site was just busy or something, but that can't be the case. Three clicks is too reproducible, and it was happening on a lot of different sites. I'd seldom if ever seen this message until the last couple of weeks. Now I get it a dozen times a day or more. The only significant change I made right around the time this started happening was to turn off caching on my WinGate proxy server and start caching locally. I talked to Barbara, and she said she'd seen the same thing happening for the last couple of weeks. So it seems pretty clear that the problem lies with WinGate having caching turned off, or perhaps with a timing setting in its configuration. I reconfigured WinGate to use caching again, restarted the server, and haven't seen that warning box again since. If the next couple of days go by without a bunch of errors, I'll know that the problem was due to WinGate. But I should emphasize a couple of things: first, I'm running v 2.0 of WinGate, even though I have 3.0 in hand. In fact, I have 3.0 installed on my future resource server, but haven't gotten around to connecting a modem to it or reconfiguring my clients to point to it as their proxy server. Second, I haven't made any attempt to determine what's causing the problem in WinGate when caching is off. It may well be that there's a timing setting or something that I should have changed and didn't. So this is not a criticism of WinGate, but simply a reflection of the field-expedient "good enough" practices we use around here. I was getting aggravated enough with IE, unfairly as it turns out, that I was considering changing to Opera. I've downloaded a 30-day eval copy and played around with it some. Perhaps I'm missing something, but overall I have to say I think Opera is noticeably inferior to IE4. It doesn't feel any faster, and it's feature-poor compared to IE4. Or at least it is for the features that are important to me. Cookies can only be set on or off, for example. I requested a live eval copy from Opera, because 30 days is not enough for me to evaluate a product, but they never responded to my mail. One nice thing about their 30-day eval is that it's for 30 days of use rather than 30 calendar days. I installed it almost a month ago, and when I started it this morning it told me I had 23 days left. IE only needs a few things to be fully acceptable to me:
There are probably some other minor things that don't immediately come to mind, but fixing these three things would make IE Good Enough for my needs. Noon: I just looked out the window and noticed that it was snowing. I guess snow in February isn't surprising, but our high yesterday afternoon was about 68F (20C). This isn't light snow, either. It's those big, fat, heavy flakes and they're coming down at what I'd guess is a 1 to 1.5 inch/hr rate, say 2.5 to 3.75 cm/hr. The weather around here is simply amazing. How many other places on earth can one be under a Winter Storm Warning and a Severe Thunderstorm Warning and a Tornado Watch simultaneously? That's not the case here today, but it was the case a week or so ago in Raleigh, which is a couple hours east of here. * * * * * This follow-up from Bo Leuf [bleuf@algonet.se]:
Okay, I give up. Actually, there's a word for this--similar or identical sounding words that describe the same thing but arose from completely different sources--but I can't recall right now what that word is. As far as the spelling of Dusenberg, I suspect that the name was originally spelled Düsenberg, with the umlaut. Umlauts have always given Americans a problem, originally I suspect because American typesetters had no umlauts in their type cases.Even now, I had to go to "Insert Symbol" to find one. So, some of us drop the umlaut and use a following "e" to indicate the diacritical, while others simply use the unaccented vowel by itself. A quick search of the Internet shows all three usages, including examples of each from seemingly authoritative sources, e.g. collectors' clubs. Although I've seen these cars at shows, I don't recall how (or if) the name appeared on the car itself. As far as Eleanora Duse, I must confess that her name was not familiar to me before this thread started. I suspect if you did a poll, you'd find close to 100% of those polled would at least recognize Sarah Bernhardt's name, perhaps 5% would recognize Helena Modjeska (as I did), and very few would recognize Eleanora Duse. Perhaps that's more a measure of their ability to promote themselves than of their talent. * * * * * And while I've written this, about an inch of snow has accumulated. North Carolina weather is certainly never boring.
|
|
Saturday |
Saturday, February 20, 1999 Hmm. We finally got our package from Blue Cross/Blue Shield yesterday afternoon. They're charging us $275/month instead of the $187/month they quoted. I sincerely hope that's not because I'm a pipe smoker, because pipe smokers are statistically healthier and live longer than even non-smokers. Of course, in this Politically Correct climate, smoking anything is a no-no, and the bastards always make you pay for not co-operating with their view of the way things should be. I remember when sociological elements played no part in insurance. Rates were based, as they should be, solely on loss experience and actuarial tables. Teenage boys had higher car insurance rates because they wrecked much more often. Women paid much higher health insurance premiums than men, because women get sick much more often and much more severely than men, are hospitalized much more often, and have babies. Conversely, men paid much higher life insurance premiums than women, because men died younger. Construction workers paid higher premiums than college professors and blacks paid higher premiums than whites. There was nothing discriminatory about any of this. It was based solely on loss experience. But nowadays, insurance companies are forced to be Politically Correct, setting rates according to sociological issues that have nothing whatsoever to do with their actual costs to insure a particular group. Although race is unquestionably a major factor in loss experience, there wasn't even a question about race on the application. I suspect it won't be much longer before the Politically Correct manage to force insurers to stop asking questions about occupation, sex and age as well. If this goes on, they'll eventually put actuaries out of business. I think it's disgusting. * * * * * I also got a package from IDT yesterday with some WinChip2 processors in it. I have a couple of EPoX Socket 7 boards I've been itching to try out, and the arrival of the WinChip CPUs provides a good excuse to take the time to get it done. One problem I may have is that the WinChips are the 3.52 volt versions (many WinChips come in two versions, one for 3.3 and the other for 3.52 volts). The highest voltage my motherboards offer is 3.3 volts. I emailed my contact at IDT yesterday, and he said to give the 3.52 volt WinChip a try at 3.3. It may work. My friend Steve Tucker is finally back in town after spending a couple weeks mostly on the road. He signed up with pair Networks for his new web site, and I told him I'd spend some time getting him up to speed with FrontPage 98. I'm headed over there this afternoon to work on the web site, and Barbara will join us later for dinner and the evening. As long as I'm over there, we might just as well be working on benchmarking these EPoX motherboards and WinChip CPUs during lulls in the work on the web site. * * * * * Barbara picked up a couple of books yesterday. Both of them are from Microsoft Press, one on FrontPage 98 and the other on Publisher 98. I flipped through the FP98 book yesterday and learned a couple of things. I think I'll sit down and read it cover-to-cover. The time has come to stop trying to do stuff by the seat of my pants and see if I can't learn the product more thoroughly. * * * * * And more from Bo Leuf [bleuf@algonet.se] about the Dusenberg/Duesenberg:
$18,000! And this at a time when the average worker was probably grossing less than $1,000/year, if he was lucky enough to have a job. And since an original Dusie now sells in the million dollar range, they've held their value pretty well. Actually, 75% existing and 55% operable doesn't sound very high to me, given the original cost and exclusivity of the product. I'd guess that Rolls-Royce motorcars of that vintage have considerably higher rates. Of course, they also have a company that still exists to support them. My great-uncle owned several Cords. I wish even one had stayed in the family for me to inherit. As usual, your persistence amazes me. Actually, Barbara (ever the librarian) wanted to research the whole doozie/Dusenberg issue when I mentioned it to her yesterday morning. I finally convinced her that we all probably had better things to be doing. But you're right. It is fascinating. * * * * * This from Andrew Sabel [asabel@wrtsun02.svidaho.net] regarding the problem with Internet Explorer:
Yep. I think I was too fast to blame WinGate. I re-enabled WinGate caching, and the problem is now happening less frequently, but it is still happening. On balance, I think I can blame this one on IE itself. Now the only thing I don't understand is why it just started happening. Perhaps the Internet is still very busy. Who knows? * * * * * And another follow-up from Francisco Garcia Maceda [maceda@pobox.com]:
No, I'm sorry if I sounded short. It's just that I've been through this whole thing so many times already that I get frustrated. Here, in a nutshell, is what happens with "bad" cookies. You point your web browser to http://www.altavista.com to search the Internet. AltaVista has signed a contract with DoubleClick.net to track people who visit their pages (and see the ads). AltaVista web site programmers put code within their pages that links to one or more servers in the doubleclick.net domain. (If you want to see what I mean, just hit AltaVista, view source, and search for "doubleclick"). When you retrieve the AltaVista home page, that doubleclick.net server sends a cookie to your browser, which it writes on your hard disk. The next time you visit AltaVista, they know you're the same person who was there during an earlier visit. But what's worse is that doubleclick has contracts with many, many web sites. And it can read that cookie when you hit any of them. So when you visit AltaVista and then visit another site that contracts with DoubleClick (most big sites do), DoubleClick now knows that you visited AltaVista and are now visiting this second site. And note that you have never explicitly chosen to visit the doubleclick.net site, and are probably not even aware of what has happened. The only evidence, unless you examine your cookie file, is the brief flash of the doubleclick URL at the bottom of the screen as the cookie is being sent to your browser. The danger also exists that if you provide personal information to any site, that information may be provided to one of the "bad" cookie companies, which can then track your web browing by name. Actually, AltaVista and DoubleClick are by no means the worst offenders. When you set Netscape to "Accept cookies only from the originating server", DoubleClick cookies do not get written to your drive. The obnoxious ones are Imgis and Net Gravity's NGAdcenter. They apparently use invisible frames to get around Netscape's filtering, and write those damn cookies to your hard disk even when you've chosen that option.
|
|
Sunday |
Sunday, February 21, 1999 Well, I finally bit the bullet and downloaded Internet Explorer 5.0 beta yesterday morning. I finally reached the "I'm mad as hell and I'm not going to take it anymore" stage with my current web browsers. Netscape Navigator 4.0 is slow, buggy, and crashes a lot. Opera is missing some features that are pretty important to me, and I find it clumsy to use. Internet Explorer 4.01 had started doing really weird things. The change to WinGate didn't help any, so it obviously wasn't WinGate's fault. The latest IE weirdity was that I could no longer right-click a link and choose Open in New Window. Whenever I tried that, the current instance of IE4 would freeze. The cursor would move around okay, but that was it. Calling up Task Manager, I found on the Applications tab that that instance was still listed as "Running." Perhaps so, but it didn't look that way to me. The CPU utilization wasn't high, and the memory looked okay. None of the expected problems existed, but IE was just frozen. This has been happening for at least a week now, and I needed to do something about it. I was on a web page when this happened, so I used Task Manager to kill that instance of IE. At least the other instances of IE were unaffected, which is better than Netscape Navigator. Killing a frozen instance of it was killing all instances at the same time. At any rate, I tried again to right click and open the link in a new window. It froze again. I killed it, went back to the page and clicked on the link directly to open it in the existing window. It froze again. I killed IE, went back to the page, right clicked the link, copied its location, exited IE, started another copy of IE, pasted the link into the URL line and hit enter. The site came up fine. So for some reason, IE had started not allowing me to jump to a link. That's a pretty fundamental function of a web browser. The problem wasn't occuring every time, or on every page, but one of these "poisoned links" behaved the same way every time. The first thing I tried was re-installing the IE4 update. That didn't help. I was considering uninstalling IE4 and then reinstalling it from scratch. But I figured if I was going to do that, I might just as well try IE5 beta instead. I had a couple of hours before I was due over at Steve Tucker's house to help him set up his web site and mess with the EPoX motherboards, so I called up the Microsoft download page and went to Beta Software, and then the IE5 beta page. The link to read more about IE5 before downloading it was poisoned. Thankfully, the link to download IE5 itself worked. I downloaded the small IESetup program and ran it. After agreeing to the license agreement, I clicked the Advanced button on the next screen, and told it to (a) not change file associations, (b) allow IE5 to co-exist with IE5, and (c) to download the IE5 setup files to my hard drive rather than installing directly over the Internet. The subset of the files I chose was about 10.5 MB, and they downloaded reasonably quickly. I was a little bit concerned, because the stuff I'd downloaded was apparently last updated in October, 1998. It seemed to me that there should have been a later beta version released in the last four months, but what the heck. I ran IESetup again, again told it to let IE5 co-exist with IE4, and started the install. Setup proceeded normally, and as usual forced a reboot at the end. When the system restarted, I fired up IE5 and clicked Help - About to find out what I'd just installed on my system. It displayed the following:
Okay, that sounds reasonable. I note that it apparently noticed that I had 128-bit encryption installed for IE4 and installed the same for IE5. I also noticed immediately that IE5 Setup had pulled in my links bar. Hmmm. The very first thing I did was click on one of the folders in the links bar. It displayed normally, with one difference from IE4. In IE4, the drop-down list of sites truncates long entries, whereas in IE5, the width of the drop-down list is set wide enough to display the full length of the entries. Then, the acid test. I clicked another folder to see if they'd fixed the Large Icon problem. Hurray. They did. Now I can click any of my folders and see a Small Icon/Text list of the entries it contains. That by itself is almost enough reason to move to IE5 from IE4. The next thing was to see if the Open in New Window problem was fixed. It was. It wasn't until I did this that I realized I'd forgotten to go in and set Proxy Server options in IE5. It picked up the correct Proxy Server configuration automatically from IE4. Very nice. Thinking about it, I couldn't believe that I had downloaded a beta product and was considering using it as my main browser because the so-called release versions of my other browsers were buggy. Geez. That's pretty bad. Using a beta in the hopes that it's more stable than official shipping versions. But after an exhaustive five minute evaluation, I've just about decided that IE5 is better than the other browsers I was using. It seems faster than IE4--which was much faster than Netscape Navigator 4 to start with--and I've been doing all my usual things without crashing IE5, at least so far. I'm almost afraid to go in and start modifying it's default configuration, though. Then I restarted IE5. One change the original installation did make was to change my home page. In IE4, I had it set to the home page of the local disk copy of my web site. The first time I started IE5, I noticed that the default home page was now a Microsoft IE5 page. Fair enough, and easy enough to change. But I didn't change it before closing and restarting IE5. This time, it took me to a different Microsoft page that announced that a later version of the product was now available. Okay, that seemed a bit strange. After all, I'd just now downloaded and installed IE5, so one would think that the download they'd just sent me was the latest IE5 beta. But, thinking about that October, 1998 date, it seemed reasonable that perhaps there indeed was a later beta release. So I clicked on the Tell Me More button, and it displayed a page that talked about IE5. I couldn't see anything different about that information than what I already had. I mean, it didn't mention IE5 beta 2 or anything. But, what the heck. I pressed the Update button, and it took me to a page that announced that I appeared to be running IE4.01 and that IE5 beta was available. That seemed strange. I went up and clicked on Help-About, and I was indeed running IE5 beta. It looked as though there wasn't a later version available, so I closed IE5. It gets weirder still. I had the FrontPage Editor open, and was writing this as I was playing with IE5. I decided to copy and paste the URL for that page into this document, so I restarted IE5. This time, it displayed my old default home page. Curiouser and curiouser, said Alice. At that point, I decided I wanted to make sure that IE4 was still available, as promised. I went to Start - Programs - Internet Explorer and clicked on the Previous Internet Explorer Ver. 4 item. It works fine. I decided to go check on how cookie handling was configured. IE5 appeared to have imported all my settings from IE4, but Microsoft seems to do everything they can to force people to accept cookies. If I believed in conspiracy theories, I'd assume a dark motive here... The first strange thing was that the option for Internet Options no longer appeared on the View menu. As it turns out, they moved it to the Tools menu. It never did make sense to have it under View anyway, so that was a reasonable change. I displayed Internet options, and found that cookie settings were no longer under the Advanced tab. My first thought was that Microsoft had removed entirely the ability to configure cookie usage. But I decided to look at the security tab (where cookie configuration should have been all along.) Sure enough, cookie configuration is now under Security, and can be set independently for each zone. They've also added separate settings for persistent versus non-persistent cookies. But I don't think it's coincidence that IE5 imported my IE4 cookie settings (reject all cookies) and converted them to Accept All Cookies. Way to go, Microsoft. First, set the IE4 default to Accept All, rather than the more reasonable Reject All. Then, bury the location for changing that option somewhere where hardly anyone will find it, even if they know what cookies are. Then, don't give them any reasonable choices for cookie handling, even if they do manage to find the location for setting those options. Then, change the new version of the software to put cookie handling options in an entirely different location. Then, when someone upgrades, change the cookie settings he explicitly entered to keep cookies off his hard disk back to your original wide-open settings. Finally, make a meaningless change (adding the persistent vs. non-persistent option), but don't do anything meaningful to help people manage cookies. But at least they fixed the Links bar menu, and IE5 "feels" much faster than IE4, which was itself the fastest browser I've used. * * * * * I spent yesterday afternoon and evening working with Steve Tucker. We started by working on his web page. In theory, I was going to show him some of the tricks and traps involved in using FP98. In practice, we spent the whole time trying to resolve a problem with publishing. Steve had already created a stubbed out web site. He used FrontPage themes, which generate large numbers of files--background gifs, button gifs, page banner gifs, etc. To make matters worse, he'd used different themes on different pages. His daughter, Katie, had one theme on her pages, Steve had a different theme on his, and so on. The result was that a web site with fewer than a dozen actual HTML pages and a few photos turned into a 2.5 MB web site with 808 files! The service level that Steve signed up for with pair Networks does not include FrontPage Extensions. This means that FP98 publishes using WebPost, which is essentially just a specialized FTP client. The problem was that we couldn't get it to publish only changed files. Every time we changed just one file and published, WebPost would attempt to upload all 808 files. I say "attempt" because it never completely succeeded. We kept getting a timeout error, giving the IP address of the client machine and a numeric error message, 2634 as I recall. I kind of fixed the problem with publishing all files by going to another machine on the network, firing up FrontPage98, and publishing Steve's web to a local disk copy. I could then edit that local copy and publish it to his real web site, and only changed files were being sent. My own web site has a different environment. When I first signed up with BigBiz. they set up a "root web" on the server. I then published that root web from their server down to my local machine and ended up with a copy (in a folder named "copy_of_www_ttgnet_com"). When I call up FP98, it shows two webs, <Root Web> and <Root Web> (Disk). When I edit the <Root Web> (Disk) local copy, the Personal Web Server doesn't start. I make changes to the local copy and then publish to the web server located at BigBiz. Every time Steve edits his local copy, the PWS starts. I think, but I'm not sure, that Steve's <Root Web> is the one on his local machine, and the one on the web server at pair Networks is a copy. Just for the heck of it, I tried publishing from the pair Networks copy down to a local machine. FrontPage generated an error message which made me think that the FP Extensions must be running on the remote server to allow that web to be published down to the local machine. At times, I find FP very confusing. To make matters even worse, Steve's ISDN service through GTE Network sucks dead bunnies through a straw. Steve has complained repeatedly to GTE without result, and has even complained to UUNet, who actually provide the POP. The problem is simply that they don't have enough inbound ISDN lines to handle the traffic from the number of ISDN subscribers they've signed up. The result, according to Steve, is constant busy signals almost every evening. We sat there for literally an hour trying to get connected and got only busy signals. I was convinced that GTE must be having hardware problems, but Steve said that was the norm for an evening with GTE. The moral here is to be very careful which ISP you select if you decide to sign up for ISDN-based Internet access. As it turned out, we finally did get on, and we finally did get everything published up to Steve's server. His web site is now officially open, although there's not much there yet. I suspect that Steve will eventually sign up for an account level at pair Networks that includes the FP Extensions. I suspect he'll also drop his GTE ISDN account and move over to BellSouth. The offer unlimited 64KB access and 50 hours/month of 128KB access for a fixed price. Better still, they control the switch, which should mean no busy signals. If everything is busy at one POP, they can simply re-route the call transparently to another POP that has free resources. That's the way their dial-up Internet service works, and only get one busy signal every three months or so. When that happens, I simply re-dial immediately and it always connects. The only time I've gotten multiple busy signals is when they've had hardware problems or were doing an equipment upgrade. This is Internet service as it should be, at least as long as people continue to use switched access. I can't wait for the xDSL roll-out, which should be sometime in the next year or so. Now, if only they'd provide a static IP address with xDSL service. They've already told me that the address will be dynamic. Damn. * * * * * We also installed the EPoX EP-58MVP3C-M motherboard, although our other problems meant we didn't have time to do as much as we'd hoped with it. When this board shipped last summer, it had the distinction of being the first Super7 motherboard that actually worked. Others had introduced Super7 boards before EPoX, but they didn't run reliably at the 100 MHz bus speed. The EPoX did (and does), which is an indicator of the superior engineering that went into this board. This is still the best Super7 motherboard we've seen in the AT form factor. We'll have much more to say about it once we have some processors that are worthy of it. We'd intended to install some WinChip2 CPUs in it, but all the WinChips required 3.45 volt core and I/O, which this EPoX board doesn't support. We hope to be getting some AMD K6-III "Sharptooth" CPUs soon, which will allow us to wring out both them and the board. But we wanted to play around with the EPoX board, so we ended up installing the only Pentium-class CPU we had handy, which was an elderly 166 MHz P55C Pentium/MMX. Kind of like dropping a lawnmower engine into a Ferrari, but at least it got us up and running. We have four EPoX motherboards in the evaluation queue right now, in addition to the EPoX EP-68LXR board that runs Barbara's main system. In addition to the EP-58MVP3C-M, we have the EP-BXT (Slot 1 with integrated Intel740 graphics and Yamaha sound); the KP6-BS (dual Slot 1 100 MHz); and the EP-MVP3G-M (Super7 ATX). The EP-BXT is a superb board. We've been working with it for about a month now, originally with the intention of reviewing it as a low-end Slot 1 solution for those who wanted integrated video and sound. As it turns out, this is one of our favorite boards ever. It's very well laid-out, rock solid, uses any memory we throw at it, and is flexible enough to use just about any existing or near-future Slot 1 CPU. We decided to leave it in our test-bed Slot 1 ATX system. As you might guess, we're big fans of EPoX motherboards. There are only two minor negative aspects to the EPoX boards:
EPoX motherboards are not distributed as widely as many competing brands, but they're worth taking some extra trouble to track down. * * * * * This from Bruce Denman [bdenman@FTC-I.NET] [http://web.infoave.net/~bdenman]:
Yep, it does snow here once in a great while. We get a trace of snow most years, one that covers the ground and stays for a while every couple or three years, and a "big" snow every five or ten years. The last big one was something like 19" (0.5 metre) and occurred several years ago. As far as Cookie Crusher, I've tried it and similar products, but never found one I thought was worth keeping. For a good book on networking basics, I'd suggest the Microsoft Networking Essentials book that's intended for MCSE test preparation. Windows NT TCP/IP Network Administration is really aimed at people who want to design, install, and administer TCP/IP networks in an NT environment. It would probably be of limited use to you, although it does provide a very good grounding in TCP/IP in the Microsoft environment, if that's what you're looking for. And I agree that Amazon screwed up. I'm not happy about what they did, and I suspect that they moved to correct the problem only because of public outcry, but at least they did fix it. * * * * * But before I'd read the preceding one, this one came in:
Hmm. I've had reports from many people who use and suggest Cookie Crusher, but this is the first time I've heard about any problems like this. Perhaps it was something other than Cookie Crusher that caused the problems. I've looked at several cookie-filter programs, including this one. One I'd just about decided to try was Internet JunkBuster, a free program that seemed to have all the functionality I needed. But it didn't want to play nice with my WinGate Proxy Server (not that I spent all that much time trying to make it do so), so I bagged it. In general, I dislike add-in programs designed to do something that the application should do natively.
|