|
Week of 17 April 2000 Friday, 05 July 2002 08:10 A (mostly) daily journal of the trials, tribulations, and random observations of Robert Bruce Thompson, a writer of computer books. |
|
wpoison
Search [tips]
|
Monday, 17 April 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] I finally got the hard disks chapter re-written and off for tech review. That was a true re-write. I cut it down to half its original size, and then started adding material on SCSI, which I hadn't covered in the first pass. Now to finish up the final three chapters--processors, motherboards, and memory. How smart are Border Collies? Here's how smart. While Barbara was cutting the grass yesterday, Malcolm and Duncan came into my office and informed me that they needed to go out, and that this time they weren't lying, really. Okay. I put on my slippers and took them out the front door, Malcolm on the roller leash and Duncan loose. Malcolm peed in the front yard. "Good dog," I told him, so he turned and made a bee-line for the front door, knowing that good dogs usually get treats. As we went up the walk, I shouted over to Duncan, who was sniffing around in the neighbor's yard. "Treat!" I shouted. No result. Whatever he was sniffing was more important than the prospect of a treat. "Treat!" I shouted again, again with no result. Time to bring out the big guns. "Big Treat!" I shouted. You see, we keep a jar of puppy treats near the front door. The dogs like those well enough, but what they really like are the big-dog treats that Barbara keeps in a jar in the kitchen. Same stuff, but much larger, and Border Collies aren't stupid. At any rate, as soon as I shouted "Big Treat!", Duncan stopped sniffing, popped his ears up, and headed for the front door on the dead run. When I opened the front door, both of them ran into the house, ignoring the treat jar in the foyer, which is usually where they stop, because that's where they get their treats on the way in the door. As I walked through the den, there they sat in the kitchen, looking up at the big treat container. The same thing happens when we're passing out human food. If I take a piece of meat, say, and break it into two pieces, whichever one gets the smaller piece looks at me as though I'd betrayed him. I'm thinking of having one of them divide the ort into two pieces and allowing the other one to pick first. It's the only way I know to keep them happy. After reading about Napster on Chris Ward-Johnson's site yesterday, I decided to download it and give it a try. He's right. It is frightening to anyone who makes his living creating content. Like Chris, I experimented by downloading a few tracks. I don't think I'll bother seriously downloading stuff with Napster, though. Most of the stuff I found was at 128 Kb/s or less, with some at 160, a bit at 192, and a very few tracks at 256. That's just not good enough quality for my taste. I'll just rip my own CDs at 256 or perhaps 320. One exception. I did download Louie, Louie by the Kingsmen, which is so garbled to start with that I figured even slow bitrates wouldn't make much difference. I was right. At 128, I couldn't tell any difference. Even at 64, it sounded pretty much like it sounds from a CD, which is to say terrible. But I can see why Napster scares performing artists and record companies. * * * * *
But they are chattel, and passing laws otherwise simply dilutes the value of ownership, which is a very slippery slope. Abused animals are a Bad Thing, certainly, but this cure is worse than the disease. As I said, I wish I could horsewhip people who abuse dogs, but I'll have to be content with shunning them and verbally abusing them. Having the state take an ownership interest in their property--which is exactly what you suggest--diminishes the concept of ownership in general. If I own something, I have the absolute right to treat it as I please. The concepts of ownership and property rights are under attack from all sides now, and the last thing we need is this assault from another direction. Before you advocate diluting property rights in a good cause, remember that ultimately all human rights devolve from property rights. Yes, I despise anyone who maltreats his dog, but I will defend his right to do so. But I will hope that he someday finds himself desperately needing something from me. I'd happily watch him drown before I troubled myself to toss him a rope. * * * * *
Okay, this is from memory, but I think it's correct in all essentials. One one of our networks, we had a Windows NT 4 Server PDC, a Windows NT 4 Server BDC, a Windows NT 4 Server member server, and clients running Windows NT 4 Workstation, Windows 95, Windows 98, and various other operating systems. In the first instance, we had the BDC running DHCP Server, correctly configured. I installed Windows 2000 Professional (although it was called Windows NT 5 Workstation back then), telling it to obtain its IP configuration information from the DHCP server. Setup completed normally, and the system restarted. Immediately, I heard a beep from behind me. Upon examination, I found that the new NT5 Workstation box was using the IP address 192.168.111.97, which was already in use by the system that beeped. The DHCP Scope was 192.168.111.1 through 192.168.111.254, with 1 through 10 excluded. There were more than 200 IP addresses available on that network, but NT5 grabbed one that was already in use. Thinking that perhaps I'd done something wrong, I stripped the new box down to bare metal and re-installed. This time, it correctly grabbed an IP address that was available. I wrote the problem off to something I'd done. But a week or so later, I was installing NT5 Workstation on another test-bed system. When Setup completed and the system restarted, my main system beeped. The new system had taken 192.168.111.203, which was in use by my main system. Subsequently, thinking that perhaps the problem was caused somehow by the fact that the DHCP Server was running on a domain controller, I removed DHCP from the domain controller, cleared all DHCP clients, re-installed DHCP Server on the member server, and restarted everything. Everything operated normally until I installed NT5 Workstation on another test bed system, at which point it again grabbed an IP address that was already in use. Obviously, it's the DHCP server that passes out the address, but the problem appears to be that the NT5 client was using a different IP address than the DHCP Server was giving it. I've never experienced this problem on an NT4 DHCP Server until I started installing NT5 clients. I wish I'd had time to put a packet grabber on the network to watch exactly what was happening, but I'm sure in my own mind that the NT5 client was using an IP address different from what the server provided. I reported the problem using the standard bug reporting mechanism, but never heard a word back from Microsoft other than the auto-reply. At that point (I think it was Beta 1), the DHCP client stub in NT5 Workstation was clearly broken, but I assumed they'd fix such a serious problem immediately. I haven't done much more with NT5W/W2KP since then, but I wasn't surprised at all to hear that IBM was having serious DHCP problems with W2KP.
|
|
wpoison
Search [tips]
|
Tuesday, 18 April 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Here's a warning if you have a dog and a home office. Barbara just stuck her head in my office door and told me about a Border Collie which had come into its master's home office and stuck its tongue in the paper shredder. The dog is not expected to live. We don't have any details about make or model, but if you have a paper shredder and a dog, it seems prudent to keep them separated, particularly if the shredder is one of those inexpensive models that sits atop a wastebasket. Barbara finished cleaning up the old Dell Pentium/200 box yesterday. After some quick research, I ended up pulling the OnStream DI30 tape drive. OnStream publishes Linux drivers, but implementing them involves recompiling the kernel, which I thought was a bit too challenging for my first Linux baby steps. I also didn't bother to install a larger disk drive for now. It has a 6 GB IDE drive. Installing a larger drive later will be a better learning experience than having it in there from the start. The box now sits on the floor under my credenza. I'll get to it eventually. It looks like Network Solutions is at it again. Not satisfied with a government-granted monopoly to charge outrageous prices for domain name registrations, they now appear to be selling their database--which is to say our names, addresses, and telephone numbers--to direct marketers. Steve Tucker notes on his page that he received a direct-mail solicitation to register wakeolda.cc. I got the same direct-mail solicitation yesterday, with first-class postage yet, but with a twist. Not only did they offer to sell me ttgnet.cc for a mere $100, they also offered to sell me jerrypournelle.cc. That in itself proves that they're operating from the InterNIC database, because I happen to be a contact on Jerry's domain record. Then this arrived, again obviously originating from the InterNIC database:
So, it now appears that if one wants a domain name, one has no choice but to provide contact information that will subsequently be resold to spammers. I suppose it's remotely possible that these direct marketers have simply gleaned the InterNIC database and that NSI is also an injured party, but I strongly suspect that NSI was an active participant in this latest outrage. As I have said before, something needs to be done about Network Solutions/InterNIC, preferably total destruction. And speaking of spam, I'm being repeatedly spammed by one of the most persistent spammers I have ever encountered. Here's an example:
I don't have any friends at BMG Music Service. Not only did I not resign from BMG, I was never a member. And I'm getting one or more of these messages each day, and have been for the last week or more. They are identical with one exception. The from address changes slightly. This one is "removeb602". I've gotten others with "removeb501", "removeb502", and so on--at least twenty so far. It's obvious that these people are cycling through various spam address lists, trying to validate the addresses on them. I assumed that whoever was sending these messages had nothing to do with BMG, so I went over and visited the cyberdev.net web site. It's just a banner page for Microsoft Internet Information Server, with a 1996 Microsoft copyright notice on it. So I sent a complaint message to BMG, addressed to abuse@bmg.com and postmaster@bmg.com. That message bounced, with a message from the mailer saying that neither of those addresses existed. How can a corporation the size of BMG not have valid abuse and/or postmaster addresses? * * * * *
Oh, I understand that, which is why I was careful to mention that the problems I had occurred with an early beta. But there has to be good reason why IBM cancelled a 300,000 machine roll-out, and explicitly mentioned DHCP problems as at least part of the reason for that cancellation. Obviously, DHCP ain't fixed. * * * * *
But there is a difference between "disturb" and "offend". There are laws in various jurisdictions here as well that require bystanders to stop and render aid. Those laws are wrong. One person's "need" does not place a valid claim on another's resources. Of course, any decent person would have used his cell phone to call for assistance. I've done that myself. But I was under no moral or ethical obligation to spend my money to help someone else. If I was under a legal obligation, then the law is wrong. In this country, we have what are called Good Samaritan laws. In essence, if I as a bystander without medical training stop and render aid, I am protected from being sued successfully if my well-intentioned actions cause injury. The standard is what a reasonable and prudent man would have done in the circumstances. For example, if I drag someone from an automobile because I fear that the automobile is about to catch fire, but by my action exacerbate a spinal injury and paralyze the person for life, that person cannot sue me successfully. Medical personnel are not protected by Good Samaritan laws, which is why you'll never see a doctor stop at an accident scene to render aid. He can't afford the risk, and his insurance probably won't cover him against a lawsuit. Nowadays, you never hear the classic cry, "Is there a doctor in the house?" because no doctor in his right mind would respond to it. And the sad truth is that our litigious legal climate means that even a Good Samaritan is at risk nowadays. I don't know if it's ever been tested, but I would hesitate to stop and render aid in the situation I described, simply because I don't want to be sued by the person I rescue. Even if the person has no legal grounds for filing a suit, defending any suit is expensive.
|
|
wpoison
Search [tips]
|
Wednesday, 19 April 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Barbara and I installed Linux Mandrake 7.0 yesterday afternoon. We booted Mandrake from the CD that Brian Bilbrey sent me. We clicked the "We're morons, so please make all decisions for us" button. It's not really labeled that, of course, but that's the general idea. The first problem was that the system contained a 6 GB hard disk partitioned as one FAT32 volume. Mandrake very nicely started to resize the partition to make room for Linux, but that wasn't what we wanted. We stopped the installation gracefully by turning off the power, and then booted up a Win98SE startup disk, intending to use fdisk to blow away the drive contents. That didn't work, because fdisk didn't recognize the partition that Mandrake had already created. So we fired up a PartitionMagic 3.0 disk (hey, it's old, but it was lying right next to the computer). It wouldn't do anything, either, so we pulled out the big gun, the IBM-specific copy of Disk Manager. That blew away the partition table in no time, and we re-started the install, again choosing the Moron option. While Barbara went off to take a shower, I sat there and watched the installer install about 500 packages, everything from Netscape Navigator (I hope it's better on Linux than on Windows) to some MP3 apps to a C compiler to Cyrillic fonts. I'm pretty sure I saw the kitchen sink being installed at one point. After half an hour or so, the installer had finished installing packages, and we finished up the installation. Everything went fine, including configuring X. We were even able to select the specific monitor by make and model. Now the holdup is that I can't use Navigator because this machine must access the Internet through my proxy server, and I have no idea how to configure that. Obviously, not while I'm still logged on as "thompson" because that is not a privileged account. I'll probably figure it out soon enough. I'd better. When Barbara saw Mah Jongg among the games, she decided she wants a Linux box of her own. I see that Ars Technica has joined Tom's Hardware in the obnoxious practice of using the no-cache pragma. If there's a reason for doing this on main pages and reviews other than artificially inflating page-read counts and ad displays, I don't know what it is. I played with using no-cache it here a while back just to experiment and see if it would help readers. What I found was that there were no advantages and a lot of disadvantages for readers. So now when I visit Ars Technica, I have to wait for the entire page to reload whether or not it has changed. Previously, IE5 displayed the old page and I clicked Refresh. If the page had been updated, it reloaded. If it hadn't, the refresh took only a fraction of a second. No more. That just means I won't be visiting Ars Technica nearly as much as I used to. I suspect other people will also cut down on their visits. I wonder how the trade-off between forced page reloads and fewer visitors will work out. It's for this reason (and the similarly-motivated actions of breaking of one page into many separate pages and making the site difficult or impossible to use with images turned off) that I seldom visit Tom's Hardware, Firing Squad, Sharky Extreme, and other similar sites. I mean, Tom's Hardware forces refreshes on every page and commonly posts report pages that contain only one or two paragraphs of text. The average twenty-page Tom's Hardware report contains enough content to fit comfortably on one HTML page, with links to graphics and so on. Give me a break. Life's too short to spend time on Tom's Hardware. Although I've criticized AnandTech in the past, Anand does things right: no forced reloads, a reasonable amount of text on each report page, the availability of "Printer Friendly" versions that include the entire report, and so on. Until this week, AnandTech and Ars Technica were the only two "enthusiast" sites I visited regularly. Now that's down to just AnandTech. If these other sites can't earn enough advertising revenue without these ridiculous contortions, they need to re-think their business plans. As it is, visiting Tom's Hardware or one of those other sites is like watching a television program that constantly alternates 15 seconds of content with 15 seconds of commercials. I don't know why anyone bothers visiting them. * * * * *
* * * * *
Sorry to hear it's that bad over there too. Our problem with lawsuits really started about 15 years or so ago, when changes were made that made it easier to file lawsuits. Those changes, of course, were sold on the basis of making it fairer for the little guy, but in fact all they did was help lawyers. We have too many lawyers in this country, by an order of magnitude, and they all have to find some way to make money. The solution to the problem is simple in theory, but impossible in practice, because we have lawyers making our laws and running our court systems. Talk about a conflict of interest. * * * * *
I'll probably just add the sender to my kill file. That way, I never see the messages. The only hint that I've gotten one is that sometimes Outlook leaves its new message flag raised with no new messages visible. In that case, if I look at my temp directory, I'll find a temp file with the killed message in it. As far as spamcop, some people seem to have good experience using it, but the truth is that most system administrators treat it as a joke. In fact, most sysadmins I know have messages from spamcop filtered into their junk mail folders because what they send you is pretty useless in determining the source of the spam. * * * * *
Sorry, I don't. I must confess that remote scanning is something I haven't tried. Perhaps one of my readers will be able to help. * * * * *
I don't think so, unless you are unwilling to concede that SCSI burners are better than ATAPI burners, which nearly anyone who has tried both will tell you. Also, I have said that the Plextor 8/4/32 ATAPI drive is the most reliable ATAPI drive I have tested, approaching SCSI in reliability. But even Plextor will tell you that one of their SCSI drives will burn more reliably than even the best ATAPI drive. I don't pay much attention to PC Magazine or their tests. These are, after all, the same folks who concluded that W2KP outperformed NT4W on the basis of tests using DMA mode for W2KP and PIO mode for NT4. Nor is DMA a magic bullet. I have ATA burners here that, although they support DMA mode, burn reliably with DMA turned off, but are unreliable with DMA enabled. You are certainly correct that faster processors, more memory, and faster disk drives make for more reliable burning. Another thing you didn't mention is OS. Burning under NT4 is much more reliable than burning under Win9X. * * * * *
Thanks for the kind words. We have three BCs of our own plus frequent visitors from Carolina Border Collie Rescue, for which Barbara volunteers. Kerry is 12, Duncan 5, and Malcolm 6 months. Barbara tried Duncan at flyball, but he wasn't really interested either. We found out last week that he has hip dysplasia, so flyball would be out for him anyway. Barbara does have high hopes for Malcolm, though. At six months, he's already a natural. He plays with balls constantly, and is very acrobatic. When he's on a tear, he leaps from floor to Ottoman to love seat to Ottoman to sofa to Ottoman to floor to Ottoman, without a pause. It's rather disconcerting to be sitting reading on the couch and have a 45 pound puppy come flying past. More so when he lands on my chest, which he frequently does. * * * * *
It's an interesting conundrum. I am unusual as a content creator in that I am not comfortable with the whole idea of copyright. The original basis of laws against theft was that the thief was depriving the owner of the use of his property. If someone steals my car, I can no longer use it. But if someone copies this page, copyrighted though it may be, he has not deprived me of the use of it. The flip side, of course, is that content creators must be paid or they have no motivation to continue to create content. I think the real problem is an economic one. Most people are happy to pay a reasonable price for things they want. The problem is that the middlemen--publishers, music companies, and so on--make the price unreasonable relative to the zero price of downloading the content from the Internet. When someone pays $6 or whatever for a Pournelle paperback, Jerry actually earns perhaps a ten cent royalty. All the rest goes to printing, distribution, overhead, and profit for his publisher and all the others in the distribution chain. When Barbara pays $15 for a CD, once again the content creator makes at most a buck or two. All the rest goes to those very costly intermediate services. What we desperately need is an efficient, universal micro-money scheme. If we had such a thing now, when Barbara wanted the latest Sarah McLachlan, she'd hit Ms. McLachlan's web site and download the music she wanted, for which she'd pay perhaps 25 cents per track or a couple of bucks for the entire CD. I'd probably set my site up to charge a few cents for each page load. I suspect my traffic wouldn't drop much, either. Most of my readers would happily pay a few cents per page to read what I have to say, just as I'd happily pay a few cents a page to read what's written on my regular web sites. What's aligned against such a scheme is formidable, though. Record companies and publishers will kill to prevent something like this from happening, because it puts them out of business. Or, more accurately, it changes their business. I still need editing, layout, and similar services to do my books, but I can shop for the best deal on those. Sarah McLachlan still needs studio and production services, but she can shop for the best deal or bring those services in house. What neither of us needs any more is distribution services, and that's the keystone of the publishing and music company business. Then there are the advertisers and all of the companies that support them. They don't want to see micro-money arrive, either, because it damages their businesses severely. Their power comes from their position as a centralized distributor of monies that are invisibly extracted from buyers. The last thing they want is for those buyers to be able to distribute those moneys themselves.
|
|
wpoison
Search [tips]
|
Thursday, 20 April 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] One of the drawbacks of using a 32 MB SmartMedia card in the Olympus D400Z digital camera is that I tend to just let images accumulate instead of downloading them as soon as I take them. I finally got around to downloading them this morning, and found a couple of interesting ones. I think I mentioned that the Dell Pentium/200 that I used to build the new Linux box hadn't been cleaned for several months. Unfortunately, it has one of the original ATX power supplies that blow the "wrong" direction, pushing air into the case instead of pulling it out. Here's what the front panel of that Dell looked like after perhaps six months without cleaning.
And here are a couple of shots I took yesterday afternoon of the exterior and interior of the Rural Hall branch library that Barbara used to run.
Thanks to everyone who has sent me messages to help me get Linux up and running. In particular, Brian Bilbrey has been very helpful. Well, I know I'll get flamed for this, but Linux is slow. I mean S-L-O-W. In fairness, it's not really Linux that's slow, but Linux running the KDE graphical environment. This Dell Pentium/200 with 64 MB was running Windows 98 immediately before I installed Linux, so I have a good basis for comparison. On average, the box "feels" about one-half to one-third the speed running Linux/KDE that it did running Windows 98. Not that it really matters, because I don't plan to use this box as a personal system. It'll eventually be configured as a Linux server, and I'm sure that a Pentium/200 with 64 MB is more than enough to make Linux fly in server mode, once I get rid of the GUI. When I build a personal Linux system, it'll have a 400+ MHz processor (or two) and 128 MB or 256 MB of RAM. I just got mail from BellSouth.net telling me that they now provide five POP mailboxes with each account, rather than the one mailbox that they've provided until now. That's very nice, although I'm not entirely sure what I'll do with the extra four mailboxes. Right now, I'm POPping directly from mail.ttgnet.com, and Barbara is POPping from our dialup account at BellSouth. I'm sure I'll come up with some use for the extra boxes. POP mailboxes are handy things to have around. * * * * *
Yes, MilliCent is exactly the kind of thing we need, except that it's proprietary. And it doesn't seem likely to fly. Pournelle has been talking about MilliCent for years, but it seems no closer to fruition now than it was when he first started talking about it. The problem is that it is proprietary and supported by very few web sites. The situation you describe is similar to the very early days of telephones, when there might be literally half a dozen competing telephone companies in one town. If you were on Bell, you could talk to other Bell customers, but not to people who subscribed to Home or Consolidated. What we really need is the equivalent of MilliCent, but as an IETF standard, support for which would be built into every browser and other Internet client. Microsoft easily could (and should) do this, simply by developing the standard and then giving it away with no strings attached. Such functionality built into IE (and IIS) would establish a de facto standard overnight, which could then be formalized as an RFC. Properly implemented, such a mechanism would avoid all of the hassles you mention. I could configure IE 7.0, for example, during casual browsing to pay for any page that cost $0.01 or less and to reject any page that cost $1 or more without prompting me, but to ask me about pages that cost between $0.01 and $1.00. For pages denominated in non-US currency, I could configure IE 7.0 to check exchange rates once per session, for each page, or never. I could specify that I wanted to be billed each month to a credit card, to my ISP account, or to my telephone bill. I could set a daily, weekly, or monthly limit on total charges, so that I'd be warned that I was approaching or exceeding the limit. I could set up password-protected accounts for children, allowing them a specific amount per day, week, or month. Unfortunately, Microsoft Wallet has nothing to do with micromoney. It's simply an organizer for your credit card information and personal data. * * * * *
Thanks. That was actually the first thing I tried, having set up the Windows version of Navigator many times to use a proxy server. Unfortunately, there's no such choice in the Linux version of Navigator. The Advanced dialog has two sections. The top one provides check boxes for loading images, using Java/JavaScript, enabling style sheets, and sending your email address as the anonymous ftp password. The bottom one provides cookie options. There are no options for setting Proxy parameters. * * * * *
Thanks. As I told Dave Farquhar (and several others whose messages I haven't posted), the first thing I did was look in Advanced preferences for proxy settings. They aren't there. I did find Start -- Settings -- Applications -- Web Browser, which brought up a proxy server configuration screen. I configured it appropriately, and fired up Nav with no apparent change. When I went back into look at the other tabs on the configuration screen, I saw that this proxy server configuration screen applied to Konqueror, which I understand is another browser. I'll keep plugging at Linux until I get it, although that's likely to take a while. Perhaps I should read a book or three. * * * * *
Thanks. Actually, I have no interest in setting up a dual-boot system at the moment. I'm trying to keep the number of variables as small as possible. Perhaps my original comments were confusing. The system now running Linux originally had Windows 98 installed on it, and Linux Mandrake 7.0 Setup attempted to convert it to a dual-boot system, when what I really wanted was to blow away Win98 entirely. But I'll keep this message handy, because I may some day want to build a dual-boot system, perhaps as my main personal system. * * * * *
Well, a browser which competently supported micromoney would obviously need to make provision for switching accounts on the fly. I should, for example, be able to sit down at a computer at a friend's house, the public library, or wherever, clear the existing active account (if any), and type in my own account number and PIN. At that point, I'd be surfing on my own dime. When I exited the browser (or clicked a Clear Account icon), my own account information would be cleared from that browser. In the example you give, the additional costs to the company would typically be trivial. Being charged a few cents per page would be nothing compared to the cost of the person's time who was doing the browsing. Unless, of course, they were hitting $1/page porn sites or something, which would be easy enough to restrict. If micromoney usage became widespread, it would indeed provide a disincentive to people who browse on company time, especially if they were using a company micromoney account to do so. That's no bad thing.
Noon: I've gotten several messages gently pointing out that the proxy settings are in a sub-dialog of Advanced, and telling me how to open that dialog. Well, gentle except for the one message that started, "You moron." That one I deleted without reading further. At any rate, the problem was that I was double-clicking Advanced rather than clicking the arrow to the left of it. Which brings up a wider question. I'm not a computer novice, nor am I stupid. If Linux is to succeed as a desktop operating system, it must be acknowledged that nearly all converts will come from a Windows environment. Why, then, are there so many seemingly arbitrary differences between the standard Windows interface and the KDE interface? I mean, things like check boxes and option buttons being gray-on-gray whether they are enabled or disabled, which makes it nearly impossible (for me at least) to tell what their current status is with a quick glance. What's the matter with a black X in a white box or a black bullet in a white circle? Surely Linux folks aren't afraid of being sued by Microsoft for using such low-level commonalities? I mean, I used to use an X in a screen form to indicate field-enabled back when I was writing business applications in Cobol and DIBOL before Microsoft, let alone Windows, even existed. * * * * *
Duh. That does it. Thanks. Here's how bad FrontPage 2000 running on Windows NT4 is about preserving case. Several days ago, I got a report from my search engine provider. Among other things, it lists broken links that it finds while parsing the site. There were many of those, nearly all of which were caused by case problems in file names. For example, the file 0621rtdn.html existed on the server (which runs Unix, which is case-sensitive), but the links were pointing to 0621RTDN.html. Within FP2K, the file on my local copy appeared as 0621RTDN.html, but when I used Windows Explorer to look at the directory, it showed as 0621rtdn.html, which was exactly the file on the server. FP2K will not allow you to rename a file by changing only case, so I renamed the file from within FP2K to 0621TDN.html and saved it. FP2K prompted me whether I wanted to update the 5 files (or whatever) that were linked to it. I told it yes. I then re-renamed the file within FP2K from 0621TDN.html to 0621RTDN.html, being careful to delete the "TDN" part and re-type it as all upper case. FP2K saved the file, again prompting me about updating the linked files, which I told it to go ahead and do. After fixing all of the similarly butchered file names (they were obvious in NT Explorer because they had lower-case characters in them), I published my web, telling FP2K to delete anything from the remote server that didn't exist in my local master copy of the web. That caused a lot of individual HTML pages to be published, not just the ones I'd changed, but all of those that linked to the ones I'd changed. Fine. Then when I published this morning, FP2K popped up a message informing me that the file 0621RTDN.html existed on the remote web but not on the local copy, and asked if I wanted to delete it. I clicked the "Hell, no" icon and let it publish. While it was doing that, I went out with NT Explorer to check. Everything was as it should be on the local copy. The file 0621RTDN.html was present, and the file 0621rtdn.html was not. Just from curiosity, as soon as the publish process completed, I told FP2K to publish again. After doing its usual time-consuming scan of the remote web server, it popped up a message similar to the one it displayed the first time. But this time, the message told me that 0621rtdn.html (rather than 0621RTDN.html, per the first message) existed on the remote site but not on my local copy, and asked if I wanted to delete 0621rtdn.html from the remote server. This time, I told it yes. It was lying both times. When I started publishing the first time, the file 0621RTDN.html existed on both the local copy and the remote web server, and the file 0621rtdn.html existed on neither. So, the first time, FP2K lied by telling me a file did not exist that did exist. The second time, it lied by telling me that a file existed that did not exist, and also by claiming to have deleted a non-existent file. How hard is it to keep case straight, anyway? It seems that FP2K should be able to get something this easy right. But it can't.
|
|
wpoison
Search [tips]
|
Friday, 21 April 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] J. H. Ricketson posted a query to his readers on his site yesterday, asking them to recommend a brand of CD-R blanks for his new burner. I sent him the following: [...] getting recommendations for CD-R blanks is meaningless unless you restrict it to those using exactly the same burner (and ROM rev) that you are using. I have, for example, a stack of Smart & Friendly 4X-certified blanks here. The S&F drive uses them reliably. The Plextor 8/4/32 ATAPI hasn't yet succeeded in burning even one good CD on these blanks, this despite the fact that Easy CD testing says the blanks are good for 1X, 2X, 4X, 6X, or 8X recording. Conversely, I have some generic blanks, supposedly 4X-certified, that the Plextor happily burns at 8X without a hiccough but that the S&F won't even admit are usable during testing phase. Nor can you assume that a blank usable at one speed is usable at lower speeds, although that is often the case. I have used blanks, for example, that had no problems burning at 4X, but generated nothing but coasters when burning at 2X or 1X in the same burner. Best advice is to check the manufacturer's web page for your burner, find the list of recommended media, and buy a small sample of it. If they work as expected, buy more. Note, however, that even different lots of media often have very different characteristics. Also, make sure to flash the BIOS in your burner to the latest rev. Good CD-R manufacturers constantly update their BIOSs to account for changes in available media. Pournelle called last night to talk about choosing components for the system he's building for his son, who is stationed on a warship, where the new system will literally be bolted to his desk, making ruggedness more important than sheer performance or any other normal factor. Jerry asked about cases. I told him I thought highly of one brand, which is built even more solidly than PC Power & Cooling cases, but the name escaped me momentarily. I told Jerry that my mind had gone blank, and he started laughing. "I love it." he said, "Just wait until you get to be my age. Your mind will go blank once an hour. It's not so bad when you're writing, because what you're trying to think of comes to you eventually, but it's hell when you're giving a keynote speech." Oh, yeah. It was Antec that I was trying to think of. Even their Value Line cases are built like tanks. I have a KS288 here, and it's built from sheet metal that's thicker than the PPC model. No sharp edges, and everything lines up perfectly. A very nice case, especially at about $65 street price with a 250W power supply included. * * * * *
Well, yes, but that's like saying that a pogo stick resembles your BMW except that it doesn't have four wheels, an engine, or a passenger cabin. Heck, I'd be willing to pay 5%. Heretofore, the problem hasn't been the percentage, but the fixed per-transaction fee. Until a couple of years ago, that fixed fee made it impractical to accept credit cards for any purchase much less than $5 or $10. Now, if you're on one of the "volume" plans, it's reasonable to accept credit cards for transactions of as little as $1, but that's still a far cry from the fractional-cent granularity we need. * * * * *
You may be right, but I think (and hope) you're wrong. The reasons all the pay-per-use schemes have failed are that, first, they are clumsy to use, as others have pointed out, and, second, that they are priced out of sight. Formalizing a mechanism as an Internet standard and building it into browsers addresses the first issue, by making it easy to sign up and pay for content and by making that single sign-up valid for accessing all premium content. Micromoney addresses the second issue. During casual browsing, for example, I might find a document on Northern Light that looks interesting. With the current mechanism, I'm not likely to pay for it, because it might cost $3 or $10. With micromoney, I'll probably pay for it, because it'll cost a penny or a dime. There's a concept taught in business school called price elasticity of demand, which basically says that when using value-based pricing (for something for which unit production cost is a tiny fraction of selling price) you can maximize overall revenue and profit by pricing your item such that the product of price and unit sales is maximum, with the optimax usually falling near the high-volume/low-price end of the curve for products that are generally useful or of interest, and near the low-volume/high-price end of the curve for specialized products. Right now, on-line content is priced an order or two of magnitude higher than it would be in the presence of a micromoney mechanism. They have to price it that way because of the relatively small number of prospective buyers in the absence of a micromoney mechanism (in effect, even things of general interest have an optimax near the low-volume/high-price end of the curve) and because the existing mechanisms are very costly to use. Whether or not a micromoney scheme is initially accepted by the masses is immaterial. The leaders here will be "boutique" sites--ones with very loyal readerships and with content for which readers are willing to pay. If the mechanism for that payment is built into their browser and all they need do is activate it, you may be sure that many will do so, if the alternative is not being able to access content they want to read. Imagine, if you will, that two years from now you are using Microsoft IE 7.0 (or whatever), and you visit Jerry Pournelle's web site one morning. Up pops a Microsoft e-Cash dialog, telling you that it'll cost you five cents to view this page, and asking how you'd like to pay for it, giving option buttons for billing your credit card, your ISP, or your telephone bill, defaulting, say, to your ISP account. All you need do is click OK to sign up and view the page. Do you click OK? Probably yes. Once you do, another Microsoft e-Cash dialog pops up asking you to configure options for using e-Cash, with the default set to "Don't Ask" for pages that cost $0.01 or less, "Ask" for pages that cost between $0.01 and $1.00, and "Warn" for pages that cost more than $1. Similar option settings allow you to specify warning levels for daily, weekly, and monthly usage. And so on. I think you'll find that people are willing to pay for what they want, if the choice is between paying and not getting it, particularly if paying is easy and relatively painless. I don't doubt you're right about PBS, but the comparison is not valid. People can watch PBS whether they pay or not. And, in fact, many consider that they have already paid via their taxes. In effect, when PBS solicits pledges, they are asking people to pay more than they need to to access content. And, yes, I understand that ultimately that content degrades if people do not voluntarily pay additional monies, but what you have there is the Tragedy of the Commons and in particular the Freeloader problem. * * * * *
We actually lived in Ardmore (on Miller Street and Lyndhurst) for several years. During that time, Barbara was initially head of the Business/Science Department at Main Library, and later became head of the Reynolda Manor branch library. We bought a house in Town & Country (near Reynolda Manor) in 1987. She continued working for a while at Reynolda Manor. One day, she and Karen Robertson, who was head of Rural Hall, sat down and agreed that they were both bored with their branches, so they traded. * * * * *
Thanks. I'll give that a try. I'm surprised that Linux doesn't automatically detect DMA-capable hardware and enable DMA. * * * * *
Unfortunately, using software RAID on NT Workstation requires a bit more than a registry tweak. You also need to pirate a bit of software, because NT Workstation does not come with the NT Server Ftdisk.sys (with RAID1 and RAID5) and the other files necessary to enable RAID. Having stolen those, however, RAID1 and RAID5 run fine on NT Workstation, or so I've been told. I've never done it myself. For $100 or so, you're probably better off going with the FastTrak IDE RAID controller anyway. * * * * *
Well, of course, existing Unix interface standards are meaningless for Linux. 99.9% of the potential Linux users out there are the ones running Windows, and they don't know or care about pre-existing Unix interface standards. If Linux is to become a successful mainstream desktop operating system, the Linux GUI needs to look and work as much like possible as the Windows GUI. I mean, it's not like Microsoft is likely to complain. All the Linux folks would need to do would be send Microsoft a copy of their own testimony during the Apple look-and-feel mess. Maybe Linux renames the Start button the Begin button, and the Wastebasket/Trashcan/Recyling Bin the Circular File. Ideally, the GUI would also not display stuff like /dev/hd0, instead aliasing stuff in the familiar C: style. And so on. I'm not arguing that folks like us can adapt. Of course we can. But anyone who wants Linux to succeed as a mainstream desktop OS had better realize that they'd better go to the mountain, because the mountain ain't gonna come to them. * * * * *
Thanks.
|
|
wpoison
Search [tips]
|
Saturday, 22 April 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Things I stumble across while looking up other things. The ants are my friends and other misheard song lyrics here, some of which are truly hilarious, often because the misheard versions make at least as much sense in context as the actual lyrics. There's even a name for this phenomenon. Misheard song lyrics are called Mondegreens after the misheard lyrics from a Scottish folk song: Oh, they have slain the Earl o' Morray and laid him on the green, misheard as Oh, they have slain the Earl o' Morray and Lady Mondegreen. Tomorrow Barbara's parents and sister are coming over for dinner in celebration of the Vernal Equinox (or something like that). I suppose the actual holiday is Eastre, but that sounds too much like Easter for my tastes. I've decided to cut back my posts on weekends, both because I need a break from time to time and because the traffic here is very light on Saturday and Sunday, typically 500 page reads or less per day. But then, as Ralph Waldo Emerson said, "A foolish consistency is the hobgoblin of little minds," so I may post from time to time on weekends anyway, if only for the loyalists who visit every day regardless. Just don't expect posts of the usual size, and don't be surprised if there's nothing here on any given weekend day. I'll hold most mail for Monday, when more people will see it. Pournelle called to tell me he's going to France next week. Not that he wants to go to France, mind you. Roberta wants to go to France and, as Jerry says, "when you've been married for fifty years, you learn to grovel." He asked if I'd keep an eye on his web site while he's gone, because he doesn't know if he'll have any connectivity at all. Even if he does, Roberta likes him to stay away from the computer while they're on vacation. At best, he may have email access. If so, he may send me short reports that I'll post for him on his site. I have a chapter to work on, and this dual-processor Pentium III box has been sitting here unused for far too long. But Barbara reminds me that her family is coming for dinner tomorrow, so we'd better schedule working on the dual-CPU box for later this week or next weekend.
|
|
wpoison
Search [tips]
|
[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Phil Katz is dead at age 37, found in a motel room surrounded by empty liquor bottles. What a tragedy. His name won't mean anything to most people whose on-line experience began with the web, but it means a lot to those of use who ran bulletin board systems in pre-web days. Back in those prehistoric days of 15 years ago, 20 MB or 30 MB hard disks were considered large and most people had 1,200 bps modems because you had to spend big bucks for a 2,400 bps modem. It took forever to transfer files, and a 100 KB file was a monster download. A class of utility programs called archivers made its debut in the BBS world to address these problems. An archiver combined many related files into a single archive file, at the same time compressing the data. At first, most people used an archiver called ARC. Phil Katz wrote a better version of ARC that he called PKARC and began to distribute it on BBSs. PKARC was faster than ARC and provided better compression and more options. It wasn't long before System Enhancement Associates (SEA), the developers of ARC, threatened a lawsuit against Katz. There was a long, acrimonious, and public debate between Katz and his supporters on one side, and Thom Henderson, the president of SEA, and his supporters on the other. Eventually, Katz was compelled to withdraw PKARC. Far from giving up, Katz sat down and wrote an alternative archiver that he called PKZIP. PKZIP was faster and provided better compression than either ARC or PKARC. BBS sysops, disgusted with the actions of SEA in suppressing PKARC, abandoned the ARC format almost literally overnight. Many of us took the time to convert every one of the ARC files we had posted for public download into ZIP files. Within weeks of its release PKZIP had become the overwhelmingly dominant archive program, and ARC was withering on the vine. There was a lesson there for anyone who cared to see it. Don't annoy sysops, because they determine what the standards will be. US Robotics learned that lesson well. They cultivated sysops, giving them deeply discounted prices on their modems, and by doing so became the standard in modems. Hayes, formerly the gold standard in modems, wasn't paying attention and eventually went bankrupt because of their inattention. Unfortunately, Phil Katz wasn't paying attention either. When he took PKZIP commercial, Katz made a huge blunder. Part of the culture in those early shareware days was the pay-once concept. Those who registered a shareware program like PKZIP were usually promised that they could use all subsequent versions of that program without paying again. Unfortunately, a lot of shareware companies broke that promise. DataStorm Technologies, for example, had promoted the pay-once concept with their overwhelmingly popular ProComm terminal program. When DataStorm went commercial with ProComm Plus, they claimed that PCP was a "different" program, and refused to allow those of us who had registered and paid for ProComm 2.4.2 to use ProComm Plus without paying again. Big mistake. When Katz brought out the commercial version of PKZIP, and later the Windows version, he made the same mistake. I attempted to upgrade for free, based on the promise made when I originally registered and paid for PKZIP. I was told that my original registration didn't qualify me for the new commercial version. Thousands of other people who had registered and paid for PKZIP were apparently told the same thing. Many of us were BBS sysops, and once again the wrath of the sysops was a terrible thing. History repeated itself. Robert Jung had written a competing archiver named ARJ. Many sysops banished ZIP files from their download directories, converting everything to ARJ format. So how is it that ZIP files are common today, while few people have heard of ARJ files? Something called the World Wide Web arrived, making BBSs obsolete almost overnight. ZIP was still in wide use outside the BBS community, and survived by default as the standard archive format. Unfortunately for Mr. Katz, although the ZIP format survived and thrived, PKWare became an also-ran in the utilities market. Many alternative archivers were released that could read and write ZIP files, most notably Nico Mak's WinZIP. Although PKWare released a Windows version of their archiver, very few people used it. And so Mr. Katz is found dead in a motel room, surround by empty liquor bottles. A sad tale. I was going to post that image of US government thugs armed with automatic weapons kidnapping that little Cuban boy, but I didn't. Probably just as well, because I'm sure it's copyrighted, although it is posted on many web sites. I was going to try to figure out how to link it so that clicking on the thumbnail displayed the image and also played a MIDI recording of Die Fahne Hoch (aka Horst Wessel Lied). Enough people have posted the picture and commented on the outrageousness of sending jackbooted thugs to break down the door that my comments would add little weight. Reichfuhrer-SS Janet "Heinrich" Reno must be happy now, although I suspect she would have preferred to use tanks. I am reminded of Jeff Cooper's observation that men dressed in black, wearing masks, and carrying automatic weapons are terrorists and should be shot on sight. I couldn't agree more. [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] |