Daynotes 1999 week 29

Daynotes: Week of 19 - 25 July, 1999

Daily notes and commentary -- Week 29

* Link to: last modified 23:15 GMT+2 on 25.07.1999

Hi, welcome to this week's daybook page.

himself See the update-link (above) that points where I last added some text, which should simplify your keeping up to speed. Of course, you may still have to scroll back a bit and see if I've updated more than once since you last visited, but that is easily done.

Associated links:

Write me at: bo@leuf.com -- if private, mark it as such!
Posted mail/discussion, see the WikiForum LeufNet
Occasional thematic articles, see "DisISay" LeufOrg

Monday 19 July

Yahoo.geocities follow-up. The legacy sites are removed, quickly replaced by other free hostings.

My former LeufOrg site at Athens/5272 is replaced by "Paradise", yet another of these Second Coming sites -- that's bound to shock some people following outdated links <g>...
The former Doctorbank site at Athens/5373 is now a Spanish/Portuguese serving about the Maçonaria Order, clearly a Freemason branch -- that's also bound to shock some people following the old links. Even though long using http://come.to/doctorbank redirection, there are abound to be many obsolete links and bookmarks.
And Edward's site is now a launchpad to music downloads. I moved his site back "under the fold" in LeufOrg, but since we've used a come.to alias for him from the beginning, that should not be a problem for his readers.

In a low blow to culture, a book-importing friend reports that the EU has changed the VAT rules . Previously, this worked so that companies always paid for imports within the EU without VAT, but of course had to pay full VAT within their own country. Individuals paid full VAT in country of purchase, even for mail-order. For books, this meant that the UK with 0% VAT on books was an excellent place to order books from. And amazon.co.uk built on that. Now the new rules state that book purchases in the EU must be VAT taxed in the buyer's country. We're back to square one, when imports went through customs and got levied VAT on top of book value+freight+fees. So it goes...

Aye, t'is true I sailed the binary C,
With roving apps in hoary past.
T'was a terrible GUI I did behold,
Whene'er the Captain screened to me,
"Make her load and make her fast,
Or mark my mail, I'll clobber ye cold."
We rounded the remains of trunc and byte,
Lookout hailed the clients about,
With POP and IMAP and LDAP and plain,
To no avail so we had to fight:
Custom forms, make toolbars sprout,
While task requests around us did rain.
Badly we were fragged and were leaking bits,
Yet still we fought and tried to RAM,
But helmsman he choose a corrupt folder,
From which to fetch his course, the twit.
The crew did rant, Capt'n spewed spam,
Neither made helm any bolder.
In circles he steered, in feedback and loops,
We limped to Help to make repairs.
But context was wrong, we had no command.
Seek here, sort there, send in the troops,
Captain wants grep, he despairs;
On Windows pane he beats a hand.
Lowly Boot does console, so too CleanView,
We clear the decks and hose the trash,
And wonder all that the ship has not rolled
From weight of code and lack of clue
Of how to avert the crash
When logic and hope both they fold.
-- Code fragment of the Binary Mariner, restored from disk recovered from MS Leviathan.

Well, it's 2 a.m. again. Tom agrees with me that my last chapter was very long. It surely was.

Tuesday 20 July

(Another late update.) Hot and humid day. Too many late working nights until 2 or 3 a.m. Even the kids are up until past midnight lately. Not to mention the kittens... Have to shape up; it's only 3 weeks until they're back at school.

I expected to spend some time working during the evening, but that came to nothing as a thunderstorm moved in. I follow a fairly strict rule about keeping computers off during thunderstorms (except for a lowly Atari Stacy that runs -- in idle standby with parked harddisk 99+% of the time -- 24/7 as Fido mailer and fax receiver). Granted, the risks are far less in the city given underground cables for electricity and telephone, as compared to the countryside, but there is always the factor of possible power outages or glitches. Less of a risk for me running a notebook with battery pack (effectively a UPS), but I still don't feel comfortable about working when electrical discharges are all around.

In my past I've had numerous experiences that taught me much about lightning and then later computers reacting to lightning.

When I was in my early teens we lived in an old house, up a hill, in an "off the side of a side road to a side road to a side road" location. Thunderstorms were not uncommon and frequently meant power outages of minutes to long hours. Nearer lightning strikes had this tendency to make frequent local calls -- the phone would pling and ring like crazy. Sometimes we could afterwards go out and see the evidence of strikes in trees and utility poles. At times, we had to replace the fuses.

Well, one night there was another dramatic storm with strikes all over the place. Interesting thing about lightning and thunder: the closer and more overhead lightning is, the louder and more sudden the clap of thunder can be. Up to a point. When a strike is real close you don't hear any thunder at all. Think taking the Sunday newspaper, supplements and all, rolling it up and dipping it in water. Then whack this good and hard against a solid (oak) table edge. Add an "electric spark" sound to the beginning. Anyway, something like that was what I heard that night as I lay in bed,. I must have lifted right off the bed in shock. This was followed by a strange afterglow in my wide-open eyes, and I heard a rattle as if from hail inside. We were all a bit shaken after that, but the storm passed quickly.

The full damage report in the morning was that a strike had hit the utility pole just outside the house, splitting it and taking out lines,. The strike had literally pulverized all the fuses in the attic fusebox -- it was just to sweep out the ceramic sand. A side strike had hit the wall outside my bedroom window, followed an unsuspected stump of wiring through this and into the electric wiring on the inside. The rattle-like-hail had been the remains of the ceramic cover over the T-connection there, opposite my bed.

Many years later, when I lived in Gothenburg, I enjoyed the spectacle of summer thunderclouds forming over the sea and then drifting in over land. These were remarkable things to see, especially in the evening. Small "knots" of cloud would form on the horizon -- you could see each "churn" and begin to spout lightning in all directions: up, down, sideways. Then these miniature thunderclouds would drift closer and transform into more normal thunderclouds as they moved inland, often following the river valley. Small but extremely active.

The relevance of this to computers is that we lived on top of one of the many small mountlets (80 meters or so above sea level) characteristic for the city. I could be working away on something on a computer when suddenly pfft! I would get a system reset. Some investigation when this happened often enough showed that it happened whenever one of these churning knots of cloud and lightning came into line-of-sight of the room I was working in. Interesting.

When I eventually started working with Atari ST machines, I never got that effect anymore. These were well shielded cases. In fact, I knew of several places that used them (part of the desktop signmaking system I had designed) right next to arc-welding equipment. Not locations that I would have recommended for computers, but it didn't faze the Atari system in the least, which kept right on cutting signs.

I had earlier experience with high density electrical discharges and electronics from when I in the mid 70s did some electronic design work for the medical faculties of the university. One of the projects involved study of how white blood cells could pass through capillary blood vessels much smaller than their own size. And why this ability was lost in forms of rheumatism. This meant a lot of high-magnification film and high-definittion video of living tissue, which in turn meant high-power flash. One entire wall of the lab was devoted to cabinets filled with banks of huge capacitors. Across the floor ran really heavy-duty wasp-striped cables. Shades of the mad professor. Whenever they ran film sequences (up to 25 high-power flashes per second), they got in complaints from all over the building about electrical interference and power brown-outs. I have no idea what kW usage this lab had, but transient drains must have been something fierce.

My part was the design of a frame-counter and date/time display that was optically merged with the images. One version of this had integrated counter/display LEDs that were kind of nifty, and in theory resilient to transient noise because both counters and display drivers were on the same chip. Shield the box, put in good regulation and filters, and all's well... Well, we still got invalid displays -- not only that, we got undefined segment combinations. Again, it turned out to be a line-of-sight problem -- if the flash unit and display were in particular positions relative to each other, the rf pulse would apparently enter the chip via the visible segment display and randomly trigger the internal driver circuitry. Curiously, the actual counters were never bothered and kept accurate values.

Wednesday 21 July

I plan to submit my chapter today. Am about half-way through the "final" tweaking pass.

(much later) And so came to pass... I "celebrated" by leaving the computers to the kids and becoming a normal mortal for the duration of the day. In fact, my wife and I told the kids to be good and mind the kittens, and we then went out for a couple of hours. Haven't had enough quiet walks together just the two of us lately, so it was needed and much appreciated.

Thursday 22 July

(update posted Friday due to problems)

It looked to be simple enough. I had the final code for Office on a CD and wanted to replace the beta installation I've been using in my working NT4-SP4 system. Little did I know...

It started innocently enough. My cautious instinct suggested that I uninstall the current incarnation and do a regclean before installing the new. (Tom agreed with that assessment when I mentioned it to him. Clean installs are the best.) No problem there. Regclean found considerable litter as is always the case with these massive MS installations. Of course, maybe I should have rebooted after each step here...? (I forget my own mantras sometimes.)

In with the new CD and -- oops, what's this? The autoboot routine informed me that it has detected an older version of Windows Installer and is upgrading it. Oh well... I go to the setup and pick all my options and point to where the old installation used to be. Strange. I'm told that I'm lacking 63 Mb space on the system partition, even though I'm installing the same component configuration I just removed in beta version. I decided that I did not have the time to figure this out (in itself an innocent, but bad move) and in any case had been intending to expand the system partition size. Good enough excuse to do this. So PartitionMagic got to do its stuff.

Trouble is, PM refused to resize the partition, claiming structural problems. Since this was in the Boot code pass (I started from NT), I exited and discovered that NT no longer boots. I instead went to Win95 and started PM from there. PMs internal check partition routine terminated however when the fault table got full, so I ran Scandisk (Everything except Linux is FAT, so there's no problem seeing the NT partition, which looked ok in casual inspection.) The Scandisk result was very disheartening indeed: massive cross-link problems in the system32 and profile folders, and 48488448 bytes with data in 470 lost file fragments! Urk...

I looked through the dire list and spotted SETUPDLL.DLL among the fallen, which seemed significant somehow, along with a few files like MÚÕ?7ÔÝ+.ÛRX found in "invalid locations". This was not good. Bad timing too, so I had to leave it for a while.

When I came back some hours later, I zipped and tried to rescue some of the partition contents, applying triage in the sense of only salvaging what could not easily be restored from zero (my collection of desktop images for example). Oh yeah, I remembered to move out the Outlook PST files too -- will see later if they survived the cluster chaos. The structural errors were localized enough that this went well enough (though I may yet discover broken files here due to cross-linking -- still, that was why I was copying out files before fixing the errors, because then I get warnings).

In any case, because I keep data well separated on separate partitions, there was not all that much to salvage that could not simply be re-installed.

After the vital copy-out was done, I let Scandisk fix errors. My intent was to restore the NT4 loss from backup and see how if registry and configuration could be retained.

Here I had the next surprise. I could not restore a workable NT from backup. And install would not even identify this as a valid NT partition. My choice was apparently only one: a clean re-install -- in other words lose all configuration and application registry information. (One does not ever restore a registry from one installation into another. Too many variables.) Plus of course the utilities that were on the partition -- while apps are elsewhere, many tools get put on the system partition.

Pause: The wife and I went to the movies, and saw Entrapment. And I thought I had problems with my hobby :)

The moment of truth came: to nuke the partition. I let PM reformat it as NTFS, but stupidly enough this then made my former D come up as the last partition in the chain (K, the former E-K being bumped down). Which I of course discovered first after a long period of setup file-copying got to the "install on what partition" selection. Geez. I exited and did another reformat as FAT, which being visible in Win95 lets install also identify it as D.

So, go for broke. I decided then to instead of installing NT4, to install NT5 as my main working system. I've had a trial set-up on K for a time, and have generally liked the feel of it.

Ok, fine, except now I discovered somewhat later that all the setup file-copying went mostly to D, leaving me about 50 Mb shy of what install insists on having free. I figure out that this lack is mainly due to the initial swap file, but I have no control during install where this goes.

Right, I was going to resize D. Clean forgot about that. Resized, there was more than enough space, but my, oh my, how time flies when you're having a good time...

Installing takes time, but NT5 is much more hands off than NT4. There were a few differences from the first NT5 install (K). After a false start or two when I was reminded that set-up got stuck on not being able to find the product catalog from the CD. Copying the file to an accessible place just hangs everything to hard reset. Oh bother, I copied I386 (all 4975 files, plus setup.exe) to where there was still some space on another partition, and installed from there. Then it finds. Go figure.

Getting tired, I inadvertently at one point somehow got set-up restarted from CD when I was expecting configuration after 2nd reboot. Not clear either how I got into, or how I got out of that. Really only a few niggles after that, e.g. when set-up couldn't find a couple of component files. Browse located them one folder lower in the temporary directory than set-up was searching. (Again, that did not happen for the first trial K-partition install.)

Everything PnP-detected perfectly. By now 2.30 a.m. I was pretty done for the day, but I had to see if I could set up minimal mail and ISP. Pegasus installed without a hitch on top of the previous incarnation in my apps partition, keeping all settings intact. I plugged in my DIVA ISDN modem and went to "Install new hardware". Smooth as silk, it found an appropriate driver and I had connectivity as soon as I defined a dial-up for the ISP. I sent off a short report to my co-author Tom, and then went to bed at 3.30 a.m. Whew.

Friday 23 July

(see details yesterday) Today started very slow, though I was up by noon. Sort of. Apart from a mail check, I stayed away from the computer for a few hours at least.

During the afternoon, I've been re-installing existing software (to get registry up to speed) and working towards the point when I go for Office again. Then I will be back to full productivity. Early candidates otherwise have been my various Internet tools, including my html editor. Hence this update.

Separate system, application and data partitions prove invaluable at times like this. "Inconvenient" as the NT4 crash was, the working data was safe, along with most application configurations. And the apps that don't care about registry (run-anywhere) weren't affected at all, apart from some lost desktop shorcuts.

I've speculated about backup failing, and figure it must be due to the dynamic assignment of 8+3 aliases being different than what is specified in registry. This can easily happen with churning of folder name order as installations come and go. The MS apps are so integrated into the system as a whole, that installing and removing these components can I suppose seriously change deep registry assumptions.

I. Had. No. Idea.

In the new NT5 setup, I poked around some more and found the "Radio" toolbar in Explorer windows. Built-in streaming radio content. People with LAN or cable 24/7 connectivity will love this... I'll have to try this with MP3 content too.

Saturday 24 July

(again late update)

Nose in the yoghurt, sniff;
Shoo cat, shoo!
Chin on the tablecloth;
Shoo cat, hiss!
Claws on the curtains, shred;
Shoo cat, shoo!
Paws at the flowerpot;
Shoo cat, scat!

I've gotten the NT5 setup to productive status now, what with Office installed and the major applications properly into registry. Overall, I like NT5 better than NT4. Of course it seems excessive that the "bare" OS takes a minimum of 500 Mb, not counting swap files. It also seems excessive that the closest thing to a "context" resume for e.g. Word is to use the "hibernate" shutdown. If Word has any way to restart with the last document window and cursor placed where I left off -- a feature found in many by now ancient apps, both windows and other platforms -- I have yet to locate this.

Tom's tip about the My Document context menu option to fully relocate the default save path for all applications is a good one. I concur with his reaction: "finally"! You have to wonder sometimes how much the programmers have actually used the products themselves.

Evening was devoted to family and friends.

Sunday 25 July

Wash day. Start the web-day by posting an update to yesterday. Then we'll see what the rest of the day gives.

* Now where did that week go to?

I've been noodling around trying to get Apache 32bit to serve locally under NT5, and so far failed. Something is different in the current setup and I can't figure out what.

Actually, there are several things different under NT5. Opera sizes the form dialog for wiki differently. Although this could be some preference setting shift due to reinstallation. Oh well...

A larger chunk of the day went to reviewing some writing that Tom had done. I find it equally enlightening with regards to my own writing, since we often do similar things that we may not always spot in our own writing after working x hours on a chapter. After an edit pass like that, I usually find things to phrase better in my own work.

Back to top -- Week list