Monday, July 27, 1998

Back into the typical Monday morning grind. The dogs woke us at 6:30 a.m, asking to go out. A bit too late, unfortunately. One of them apparently wasn't feeling well in the middle of the night, and we found numerous accidents on the library and dining room floors. Thank god for hardwood floors and polyurethane. Dogs hate to break training, so the poor thing slunk around the whole time Barbara and I were cleaning up.

Checked my email, and found a message waiting from the publisher with feedback about the preliminary outline/TOC/sample pages I'd sent him. He generally liked what he saw, but made quite a few suggestions about how to improve it. I'll work on those this week while he's on vacation. I hope to have a more-or-less completed chapter by the time he's ready to look at the material again.

I checked the NECx and Insight web sites this morning to see if they reflected the new pricing for Pentium II processors. They still had the old pricing - about $350 for a PII/300. I checked the CSO site, and they'd already posted the new prices - $259 for a PII/300. That's not bad, considering the new quantity 1,000 bulk Intel pricing is $205. The $259 price is for a boxed processor with fan.

CSO was also showing the PI/266's at $199. I'd have preferred those to the PII/300's. The PII/300's cost about 30% more, and run about 12% faster. Not much of a deal, but CSO can get the 300's, and can't promise to get the 266's anytime soon, if ever. Apparently, all of their suppliers but one are saying that the PII/233's and 266's are no longer available. The one exception claims that they will have a few 266's arriving 7/31. That means I wouldn't be able to get them until early next week, and I need to get this box built this weekend.

CSO doesn't actually have the PII/300's in stock yet, but can get them in a couple of days, so I put in an advance order to make sure they have two of them set aside for me. I also had the salesperson put aside a 128 MB registered DIMM, at less than a buck a meg, no less.  She also talked me into an Intel 2X/8MB AGP card instead of the Matrox, which no one seems to be able to get. She said she just installed the Intel AGP board in her own system, and prefers it to the Matrox. I'd wanted the Matrox to make it easier to shift the hard drives from the existing NT box (it has a Matrox Millenium II/PCI), but it's no big deal. I'll just boot NT in vanilla VGA mode and change drivers.

I'll pick all this stuff up no later than Friday afternoon or Saturday morning, because our friend Robin is coming over this weekend to help me build the box. We had to postpone it last weekend, and I don't want to postpone it again. She's interested in PC hardware and has never built a computer before.

Tuesday, July 28, 1998

Sick dog again. This time at 6:15 a.m. Nothing like awakening to liquid-sounding excretory noises and finding puddles of unmentionable stuff on the bedroom floor. It's hardwood, too, thank god. We gave up on the Imodium and started feeding him special dog pills that supposedly will fix the problem. We'll see. Nothing but rice for him for the next couple of days.

The salesperson from CSO called this morning to say that everything had arrived except the Toshiba 24X IDE CD-ROM drive, which they don't expect to have back in stock for several weeks. She offered the Teac 24X drive as a substitute, and for six bucks less. After checking it for compatibility, I told her that'd be fine. I'll head over there tomorrow morning to pick the stuff up. Barbara works from noon until 9:00 p.m. on Wednesdays, and so will be around to keep an eye on the dog. He seems to be getting better. One additional episode this morning after Barbara left for work, but none since then.

Spent most of the day writing sample pages for this proposed book. My "sample pages" are going to turn out to be a full chapter or more by the time the publisher gets back from vacation, but that's okay. I don't think it's going to be wasted effort.

A day seldom goes by without something weird happening. Here's the latest: I use Word97 (with SR1 installed) running on NT4/SP3. Generally, it's rock-solid. However, with a very few specific documents, Word GPF's reproducibly when I exit Word after editing that document. No data is ever lost. If I saved the document manually before exiting, the saved version is fine. If I didn't, Word prompts me to save, saves the changes, and then GPF's. When this happens, perhaps one time in ten the Winword.exe process remains active, consuming nearly 100% of the processor, although the application is no longer visible on the Applications tab of Task Manager. This is reminiscent of the Word/Excel zombie problems before SR1 and SP3, which was supposedly fixed.

These are documents I've created myself from scratch. There's no sign of anything weird about the document or of any virus infection. There's nothing strange about the type of content. Doing a Select All and then pasting the results into a new document doesn't solve the problem. The new document then generates GPF's reproducibly. I've tried editing these documents on a couple of other computers that have their own local Word97 installations, and the documents GPF there as well. Although I can't say that this will happen on every computer, one of my editors had the same thing happen to him, so it looks like it's native to the documents themselves.

What's even stranger is that the document appears to "poison" the running instance of Word. I can edit one of these documents with no problem. If I then close the document without exiting Word, everything is still fine. I can call up and edit an ordinary document and save it with no problems. When I finally exit Word, it GPF's. The second (ordinary) document does not become "poisoned". Apparently, editing any of these problem documents notifies Word to die when exiting.

This has been going on for months. I put up with it because it happens on very few documents and because no data is ever lost. However, it gets very aggravating when I have to work all day with a problem document, watching Dr. Watson come up each time I save and exit. I can't find anything about this on the Microsoft web site, anywhere else on the web, or in TechNet. Today, a new twist manifested itself: On my sample pages document, Word now GPF's not just when exiting, but when printing. It finishes sending the print job, which prints fine, before it GPF's, so I'm not going to worry about it. Sometimes, though, I think some of Microsoft's products should star in Unsolved Mysteries...

Wednesday, July 29, 1998

The dog appears to have recovered. No surprises since yesterday morning. Still, we're not taking any chances. He'll be getting his anti-diarrhea dog pills and eating rice and chicken for a couple more days.

While Barbara was still home this morning to keep an eye on the dog, I headed over to Computer & Software Outlet pick up the processors, memory, and other stuff I need to build the box. Robin called this morning to say she had some free time Friday if I wanted to build the system then instead of waiting until this weekend. So, we're on for Friday. Here's the current status of the component list:

Not bad. About $1,800 with sales tax for a dual processor PII/300 box with 128MB of RAM. Actually, I cheated a little bit. I didn't actually have to buy the hard drive and floppy drive. They came from the stock room, but I added them back in to the cost to keep the comparison valid.

On the way back home from CSO I was listening to NPR news and heard a story about a serious security flaw in Outlook 98. It's possible to include a file attachment in an email message that will cause the recipient's copy of OL98 to crash. Worse still, after crashing OL98, the attachment can cause rogue code to execute on the recipient's computer. You don't even need to open the message for this to happen. Microsoft has posted a hotfix for the problem. It's only 151KB, and takes about 30 seconds to install. You can download it from the Microsoft web site at http://www.microsoft.com/outlook.

Problems again today with my main ISP, BellSouth.net. It started in the morning when Outlook kept displaying time-out messages when I tried to grab my email. I finally telnet'd over to port 110 on mail.lig.bellsouth.net (my POP3 server), and found that OL98 wasn't causing the problem. The POP3 server itself was pathetically slow - literally 3 minutes from the time I entered the username until I got a password prompt. I finally set the OL98 timeouts to the largest value available (5 minutes, I think), and just let OL98 work in the background.

The slow mail retrieval I could live with, but around lunchtime the whole Internet appeared to be bogged down. Web pages were taking forever to load, and even DNS lookups were timing out. I figured it was BellSouth.net having yet more problems, but that turned out not to be the case. I connected to my alternate ISP, and the same thing was happening. Apparently, the whole Internet (or at least the part of the backbone I'm connected to) was bogged. Makes it tough to work.

Thursday, July 30, 1998

I noticed when looking at my web stats that the Northern Light search engine finally got around to indexing my whole site. It's the first one to do so. I also registered my URL with AltaVista, HotBot, Excite, and a couple of others two or three weeks ago. AltaVista and HotBot grabbed the index page immediately, but neither of them has gotten around to indexing the site yet. The others haven't even hit the index page yet. I guess now that I'm being indexed by at least one engine, it's time to get serious about getting some content up on the other pages of the site. Most of them are still just placeholders.

Well, the older dog (Kerry) seems to have recovered completely from his accidents-in-the-house problem, but now the younger one (Duncan) seems to have caught it. Naturally, he managed to hold it all morning while he upstairs on the hardwood floors. When I went down to fix lunch for my mother, he lost it on the carpet. Sometimes taking care of dogs is worse than taking care of kids.

This week is rapidly winding down, I'm scheduled to spend tomorrow building a new PC, and I haven't gotten anywhere near where I want to be on the chapter I'm working on. Typical week, so far.

Friday, July 31, 1998

Robin showed up at 10:00, and we spent most of the day building the new dual-processor computer. The only minor problem was that the Micronics motherboard documentation (on CD only - no printed manual) nowhere stated whether to use brass (conducting) standoffs or nylon (non-conducting) standoffs. There were solder pads at each mounting hole, which made me 99% sure that I was supposed to use brass standoffs, but I've seen motherboards smoked from using brass where nylon was the correct choice. After only 8 minutes on hold, a nice Micronics tech listened to my question and put me on hold again. When he came back, he said he'd gone to look at a system they had torn down, and that all brass was the correct choice.

The thing actually went together in an hour or so, and appeared to work fine. Until, that is, I attempted to install Windows NT 4.0. The text-mode portion of Windows NT Setup worked fine. At the first reboot into GUI mode, however, the system started an endless loop of rebooting. On the blue screen that appears when the system shifts into graphic mode, there are ordinarily two lines. The first gives the version and build number, and the second gives the number of processors and the amount of memory. The second line never appeared.

My first thought was that the hard disk might be the problem. It had originally been formatted under MT5, and I'd had enough problems with that OS that I was hoping it was causing the problem. I downloaded a low-level format utility from the Western Digital web site and used it to wipe the entire contents of the drive. I then re-ran Windows NT 4.0 Setup, but the same problem occurred.

After verifying all system board jumpers and BIOS settings, I finally concluded that either the memory was too slow (I'd installed non-PC100 SDRAM) or that the AGP video card was causing the problem. While messing around late in the evening, the system appeared to boot successfully. The second line on the first GUI screen appeared, although after a slight delay. NT4 then did the normal conversion from FAT to NTFS, and forced the normal, expected reboot. Unfortunately, it started looping through reboots again. At this point, it's obvious that the problem is not the video card - it would work always or not at all - so the problem must be memory. I'll head for CSO tomorrow and get some PC100 memory. I'll also pick up a vanilla PCI video card just in case.

In the mean time, we had to call the vet to schedule an emergency visit. Our older dog, Kerry, was chewing his tail head area and had horrible patches of broken skin around it. I called the vet, and fortunately our friend Sue was working. We got a 3:00 appointment, and Barbara took Kerry to see Sue.  They had to use a muzzle, because Kerry was in so much pain that he was showing signs of biting. He never actually bit anyone, but he was making serious growling noises and making threatening lunges. When Kerry returned, he had shaved patches, but Barbara says that Sue told her that Kerry'd be fine.

Saturday, August 1, 1998

CSO opens at 10:00, so I headed over there at opening time. I brought along the non-PC100 128MB SDRAM DIMM I'd bought, and swapped it for two 64MB PC100 SDRAM DIMMs. They didn't have any 128 MB SDRAM DIMMs. Apparently, the price of RAM had taken a major jump recently, because the 64MB DIMM's that had been listed at $78 each were now $103 each. Aggravating, but there it is. If I'd known that before I handed my 128 MB DIMM across the counter, I'd probably have kept it, or at least held out for a trade at the new price instead of the old price. I also picked up a Matrox Millennium PCI card, just in case.

I got back home and installed the PC100 RAM. At first, I thought it had worked. NT4 booted normally, although with a short pause before the second line of the first GUI screen appeared. It went into the first GUI phase of NT4 Setup. I went to tell Barbara that it appeared the problem was solved, and when I came back the system boot screen was showing. No joy. I guess I'm going to have to talk to Micron tech support about this when they open on Monday.

I still have parts strewn all over the kitchen table. Our friends Sue and Robin came over for dinner tonight. We ended up having to eat in the dining room, of all places. The dogs recognize Sue, of course, and they know she's the vet, but they treat her as a social visitor when she's here. It's interesting that they can differentiate between vet-mode and friend-mode.

Sunday, August 2, 1998

I'm going to take today off. I'll talk to Micronics Monday. If I can't get this thing working as a dual processor box, I'll turn it into a single Pentium II/300 workstation, buy another PII system board, and use it and the other PII/300 processor to upgrade kerby (a 486/66) to a PII/300 box. Then I'll swap that in to my wife's office for thoth (her Gateway Pentium/133 box).

After talking with my friend David Rowe, I took his suggestion and removed one of the processors, substituting the terminator card supplied with the Helios board. No joy. Now the system spontaneously reboots even sooner. I hope I've just done something stupid, and that Micronics tech support has a quick and easy fix for this. I'd hate to have to remove the motherboard and ship boards back and forth.

Updated: 31 May 2002 14:22

Copyright © 1998, 1999, 2000, 2001, 2002, 2003, 2004 by Robert Bruce Thompson. All Rights Reserved.