September 2002 Part Two


15th September

I've been having problems cleaning the Perspex windows on the new case, but have finally found something that works really well - Renovo Plastic Window Polish. It's designed for the windows on boats, or open-topped cars, but neatly removed the host of fingerprints (and smears from other cleaning products!) and left everything shining and glowing prettily. It was about 5 all-in for a little bottle, but from the look of it will keep me going for quite a while. Recommended.

This afternoon was spent patiently sleeving cables and threading them into the chassis of the new case. There's still a handful more to do, but I managed to wrap all the power supply leads in black split loom (a flexible, corrugated plastic pipe) and it turned out quite well. I prefer the braided finish to the corrugated, but I hust couldn't face teasing and wiggling the braid over all those PSU connectors!

I think I'm nearly there, now - at least for the first build. I'll fit a third neon tube, I think, to brighten the insides a little more, and I still need to find a good way of mounting the controls for the neons and variable-speed fans, but those can wait until round two. Phew!

14th September

Fifty percent of stalkers are ex-partners, according to a recent report. Yes, I expect that they probably are...

Elsewhere, Sideshow reports that Warren Zevon has been diagnosed with inoperable lung cancer. I was only introduced to Zevon's music fairly recently, but "Mr Bad Example" is rapidly becoming an all-time favourite, and I know that his talents will be sorely missed by a certain French Inhaler.

Bah! David Blunkett, our fascist bully-boy of a home secretary, has condemned civil liberties campaigners in the UK for daring to complain about government surveillance measures. He cites the Labour party's record on data protection, freedom of information, and human rights - "Taken together, all this amounts to more protection for the British citizen against the state than virtually anywhere else in the world." Oh, for fuck's sake!

Spectacular new images from the Hubble telescope, this time of a interesting area of the Large Magellanic Cloud named N11A, a relatively young star formation in the shape of a rose. Pretty...

[Later] Rude words! I've just wasted a lot of time fixing a problem that didn't exist! One of the side effects from my registry repair, last week, was the failure of the performance monitor counters for the network card. Re-installing the card seemed the obvious thing, and so I grabbed the latest version of the manufacturer's driver suite and set off. Now, the last time I tried using 3Com's own drivers, I lost all contact with the network until I uninstalled them and replaced them with Microsoft's defaults - so I wasn't particularly surprised when I couldn't connect to the LAN after I'd installed them and rebooted. I was willing to persist this time, though, as I wanted the additional tweaking and tuning settings that the "official" drivers provide, so I spent a few hours trying various tricks and work-arounds (with a notable lack of success!) before uninstalling them again with the intention of reverting to the Microsoft ones. More fuss ensued then, as the original drivers seemed to have mysteriously vanished from my system, and attempts to copy them over from Ros's PC revealed that her floppy drive appears to be failing! Aargh! Eventually I burned them to a rewritable CD, but it still didn't help - no network! More fiddling, and close examination of INF files... nothing... until I suddenly thought to set a fixed IP address in stead of requesting one from DHCP, and everything burst into life again - and then I remembered that I'd left the DHCP server paused while I was testing the Axis camera last night... <slaps forehead>

Ah, well - I'm running with the 3Com drivers now, and at least I have my performance counters back.

13th September

I've been drooling over each new version of the Axis network camera for at least ten years, so today a little dream finally came true. Boy, was it a bitch to set up, though. For some unfathomable reason it's TCP/IP stack is written so that having requested an address from a DHCP server, it reboots to enable the new address and then... requests an address from the DHCP server, reboots, requests, reboots, requests... It's a very odd way of doing it!

The DHCP server in Windows 2000 doesn't seem to allow a MAC address to be excluded from the scope, somewhat to my surprise, but with some judicious enabling and disabling of the service I managed to keep the camera tied down for long enough to assign it an address manually... at which point I had to perform a pair of firmware updates, each of which cleared the network configuration and required the DHCP dance again. Once that was out of the way, though, everything since has been smooth - image quality is good, and it's ten frames-per-second maximum is as smooth as most of the expensive corporate video-conferencing systems I've used in my time... A reasonable image size and compression ration at that frame rate fits nicely into our ADSL pipe, too, and I think it's going to be a very workable configuration. Oh, and it's brought another three multi-coloured LEDs to the network, too, prompting me to finally update the Fan Counter.

These were taken in really low light, and at the default image settings... As usual, the PC's reflections and internal lighting are making it awkward to capture - but the second picture, of the gadgets on one side of my desk, is clearer and easier to make out than under the current lighting conditions in real life!

I seem to have a veritable plethora of cameras, right now - Ros had been planning some world travel when I started shopping, and even though her plans have been postponed somewhat, now, the idea had acquired a momentum of it's own... So I currently have a pair of 3Com USB cams and the Axis all staring at me with unfathomable little electronic eyes - I'm sure it's not healthy.

12th September

Ros bought me a DVD player for my birthday, and a fancy new multi-remote control to go with it... Knowing geeks as she does, of course she let me choose my own, and I've picked up a nice Toshiba, pre-modified to be region-free. It was dead simple to install, and probably the hardest part of the whole process was deciding whether to get the silver version or the black... Sound and image quality are both rather better than playing DVDs from my PC, but that is likely to be at least partly because of the 20m s-video cable connecting the PC to the television. It has some neat audio enhancement modes, too, including one that boosts the volume in dialog sections and cuts it during explosions etc... Very useful!

The remote control is a neat little gadget, too - it has some conventional or "hard" buttons, together with a big blue LCD touch screen that can be customised to show different buttons for different devices. It will take a little work to teach it all the signals from the old remotes, but I think it will be worth the effort. I found an extremely useful resource while I was researching the state of the art in remote controls, too - Remote Central, which has extremely voluminous reviews of all the current models of interest.

Elsewhere, Buzz Aldrin shows that he still has, if not the right stuff, at least some kind of stuff...

11th September

I've just been playing with another online "Real or Fake Breasts" quiz, and as usual I can spot implants with almost 100% accuracy. Having paid a little more attention today, though, I think I've figured out what I'm picking up on: the upper surface of natural breasts, where they curve out from the body, is nearly always concave - in other words, a curvature inwards towards the body before curving back out to the fullest part of the breast. Enhanced breasts, on the other hand, seem to have a convex shape to this surface, becoming fuller as they curve out from the body.

There are other indicators too, I suspect (and, after all, sometimes it's just blindingly obvious!) but I'm sure that this is the most significant one for my subconscious. It seems extremely repeatable, whatever the size and shape of the organs in question, as long as they are seen at least partly in profile. It's usually harder to see this curve straight-on, especially in a picture, but from that angle the artificial separation is often another dead giveaway... Well, I don't think that I've ever been up-close-and-personal to a pair of enhanced boobies, but although I love big breasts I'm sure that my preference would be for the genuine article: apparently many women with implants don't think that they feel very natural (they can be very cold, too, I've just read); and if they can tell then I expect that I could, too!

I hope that the recent health scares over silicone will eventually turn the tide against implants of all types, but it seems that the media is still firmly leading many women around by their noses, and I'm sure that the medical industry will be only too happy to provide "safer" and more exotic techniques. It's a damn shame...

10th September

I've just finished reading "Inside Linda Lovelace", the original 1974 version of her autobiography, written just after the movie "Deep Throat" had catapulted her to fame. In it she describes the (then) upcoming sequel, "Deep Throat 2", and it turns out that the leading male actor was to be a computer geek named "Dilbert"!

I'd assumed that this was common knowledge, as I'm normally the last to discover these 'everyone knows' things... but a quick search of the web suggests otherwise - and I find that peculiar to say the least.

I haven't seen the film, so I suppose that the character could have been renamed, or never named on-screen, or even cut altogether... and, of course, Lovelace herself later claimed that both of the first two autobiographies were penned by ghost-writers, so it's possible that the entire description of "Deep Throat 2" was just a fabrication... But still, the book itself must have had a huge readership over almost thirty years and I would have expected somebody else to have noticed!  <laughs>  Hmmm. With a little work, this could easily become a 21st century equivalent to the old "Captain Pugwash" myth, only in reverse!

9th September

I'm speaking with crossed fingers, but I think I've fixed the registry problem that has been plaguing me for the last ten days. The only authoritative documentation I've found so far, Microsoft tech note Q269075, suggests three possible (and plausible!) problems - the presence of a Promise Fastrack 66 IDE RAID adaptor, corruption in the SYSTEM hive of the registry, or a SYSTEM hive that is too large for the available memory at boot time. Their solution to issues with the former is removing the adaptor, which isn't really possible here as the Windows installation in question is running from it... And so, knowing that I'd installed device drivers that could be problematic with SMP hardware, I've been focussing (without any success at all!) on the possibility of a corrupted registry hive.

While browsing yesterday, though, I stumbled on an attempt at "translating" that tech note into less technical language - and while the article as a whole was amateurish, misleading, ill-advised and poorly written, something that he said about registry size struck a chord. A second look at the tech note reveals that the SYSTEM hive has to share the 16Mb boot environment with the loader, the kernel, the HAL and the boot drivers - and Microsoft warn of potential problems if the hive reaches 13Mb or more.

Now, my hive was only around 10Mb at the time, but the rest of my startup environment is relatively fat and complex, and I could see that it might still cause problems - the typical time for a registry-related boot failure was just after some hardware object had been re-detected, and it seemed possible that any new drivers or devices would push me over a hypothetical 10Mb-ish working limit on my system.

I looked at various ways of slashing-and-burning within the HKLM subkeys themselves, but everything in there is the proverbial maze of twisty passages full of inter-related objects, and it's very dangerous to work blindly... In the end I restored an extremely old SYSTEM key from shortly after the OS was first installed - it redetected all it's hardware very nicely from the existing INF and driver files, and as far as I can see everything is working well. The drawback is that a fair amount of my application software has components that install as a service or virtual device driver - and these will all need careful repair or re-installation. So far I've spotted and corrected the WebTrends firewall monitor, the anti-virus and defrag apps, Internet Explorer's history function, and a couple of CD mastering utilities - but there are more that I've forgotten, I'm sure, and doubtless issues with missing services or settings will suddenly pop out of the woodwork when I least expect them. On the other hand, if I'd bitten the bullet and re-installed the OS, I'd be looking at re-installing everything 'round about now, so I consider myself ahead.

So now I can get on and finish the hardware installation, at last - those brightly coloured PSU cables have been nagging at me through the side window for days, but I have black split loom and I'm not afraid to use it!

8th September

I've just watched a rare and unusual video, a performance by John Otway and Wild Willy Barrett at the Tricycle Theatre, London, in 1981. Otway and Barrett are legends in their niche, presenting a set ranging from tender love songs through solid R&B and bastardised covers of classic rock, together with a stage-show unsurpassed as a source of cheap thrills - Otway jumps off high things and turns somersaults while playing the guitar, head-butts the microphone until the blood flows, flails around on his back like an epileptic break-dancer to the accompaniment of synth-drum pads strapped to his limbs, and generally has a whale of a time just performing.

Barrett retired in the mid-eighties, but Otway has doggedly continued as a mainstay of the pub, club and student union circuit - I saw him play a blinding gig at my college, Polytechnic of Wales, sometime around 1986, following a strong recommendation from my older brother - who saw him at Bristol University at least eight years earlier... and he's still doing it now, heading for his fiftieth birthday and towing a succession of young, enthusiastic, straight-men-cum-guitarists who inevitably discover that they just can't keep up with his pace...

I've been on the Otway mailing list for years, but don't usually read it - however, a quick scan through a while ago pointed me both to the source of the video and Otway's own online shop, and also to the startling news that he is once more trying to butt his way into the limelight.

Although he's released a number of albums and singles, mainstream success has mostly eluded him - it's been twenty-five years since his only chart hit, "Cor Baby, That's Really Free" with Wild Willy Barrett, and although Otway's charming, manic grin never slipped, one could see that the musician inside him was shouting and stomping it's feet for recognition. The story goes that Otway's fans (and they do seem to be exceeding loyal fans) asked him what he wanted for his upcoming 50th birthday. The reply - "Another hit?".

And so that's what they're attempting - a ruthless and cynical manipulation of the music industry, with a carefully planned campaign of bulk-buying of the new single, Bunsen Burner, that would put the Hubbardites to shame - culminating in a birthday gig at the London Palladium on the night that the new chart listings are released.

I'll be there, and I think it's going to be an excellent night... The word is that Wild Willy Barrett might even be coaxed out of retirement and kicked severely around the body and head until he accompanies Otway in at least one rendition of "Racing Cars"... Oh, that guitar...

7th September

My birthday - 37 today.  [Correction - Apparently I'm 36... How time flies...]  I'm older than Ethernet (1973), the microprocessor (1972) and Unix (1970), older than the integrated circuit and the bad reputation of the GOTO command (1968), and older even than object-orientated programming (1967). I've been browsing an excellent computer timeline at an offshoot of the IEEE, which appears to list every significant development from 4000 BC onwards although, just to be contrary, there is nothing listed for 1966, my birth year. I should be mentioned there, dammit, at least!

Meanwhile - Dan is writing about mouse mats: "They're more interesting than you think. Well, they'd pretty much have to be, wouldn't they?".

5th September

Something of a tiring day at work, today, as I'm taking a day off tomorrow and needed to tie up some loose ends. The last loose end turned out to be field-stripping the CD/floppy drive assembly from a Dell server on the glass reception desk in the foyer - it was a faulty unit being replaced by the manufacturer, and rather than providing the entire module they had sent the separate assemblies without the frame or mountings, and the courier was adamant that he had to take away exactly the same parts that he'd delivered. I hadn't dismantled that particular subsystem before, and it was the usual Chinese puzzle of tiny black screws and fragile plastic clips - I was hurrying somewhat, unfortunately, and ended up having to scrounge a little plastic eject button from a dead floppy drive later on, having inadvertently sent the original away still attached to the faulty drive. I hope that this unexpected precision in supplying replacement components isn't a new trend for Dell...

No rest for the wicked, though, and during some abortive speculation on whether to watch a DVD tonight, I remembered that I hadn't yet connected the CD and DVD audio cables. So, out with the heat gun ("More power!") and two hours later J.R "Bob" Dobbs was my uncle. They certainly look pretty, and if I remembered the various flavours of socket correctly they may even work!

4th September

I've managed to find a UK supplier of black IDE VXA tape drives, so with luck one will be on it's way to me next week. I'm currently ahead in the eBay auction for the SCSI version, too, and having a spare drive (if a second-hand one) for emergencies will probably help the frown lines considerably during the next crisis...

I'm also bidding on an Axis 2100 network camera, and as that is looking promising too, I'm looking for a way to hook it up to the wireless LAN so that it can be moved around the house without leaving a trial of CAT5 cable. Wireless Ethernet bridges are readily available, but most are designed for linking different buildings on a campus and so have a range measured in miles and a price tag measured in thousands... Fortunately there are also several manufacturers at the low end of the wireless market, and the Linksys WET11 appears to fit the bill for a little over $100. They currently seem to be unobtainable in this country, though (as so often - most of what I need at home seems to be sufficiently rare and unusual to discourage importation) and thanks to the the DTI's insistence on subtle changes in the 802.11b "standard" for the UK, an overseas version may not be an option. The search continues...

3rd September

I've been looking at tape drives for the home LAN, today, after a growing feeling of dissatisfaction with the Seagate Hornet Travan drives that I'm currently using in three of the PCs. As well as the comparatively high failure rate (although periodic overheating won't have helped that), their 10Gb native capacity is now starting to become a major issue - a weekly cycle of the two desktop systems takes four tapes each, and apart from all the fuss of changing tapes the chances of a serious failure somewhere in the cycle is significantly greater.

Seagate do make a higher capacity Hornet (20Gb native) but my confidence in the whole Travan format has been somewhat shaken and, besides, it's not really a significant increase: what I would like is a central backup server with a drive that could hold the entire network on a tape or two, as although there are advantages in having each system responsible for it's own backup, I'm coming to see that even on a small LAN they are often outweighed by the disadvantages...

There are several suitable candidates (LTO, AIT and SDLT spring to mind, in order of merit) but they all have price-tags in the thousands. DAT is certainly more accessible, but also has a fairly high failure rate, and although the latest DDS4 drives have a similar 20Gb native capacity to the Travan 40, it still wouldn't be roomy enough for a single server to backup all the workstations... and, dammit, I just don't like the format!

Fortunately there seems to be an easy answer, in the shape of Ecrix's VXA-1 packet-writing drive. I've been using a couple of these at the office for backups of the DMZ etc and they certainly seem neat enough; their 33Gb native capacity should cope with a weekly cycle of both desktops and the server on that sought-after couple of tapes... although, actually, I'm still intending to get a pair of them - a SCSI version for the server (I have my eye on one at eBay) and a black ATAPI version for my own PC. I've been in the industry for far too long to feel comfortable without a spare drive at hand in case of emergency...

It's going to be an interesting game fitting the ATAPI drive in the new case, though - it's a 5" device rather than the little 3" Hornet, and I may finally have to find out if IDE cables will work "the wrong way around" to connect it up... <laughs> We shall see.

2nd September

I shall do the odd bits-and-bobs of wiring over the next week, I think, but the bulk of the final cosmetic touches will have to wait, as wrestling with that registry problem has drained my resolve somewhat. It's an annoying problem, too - Adaptec offer no comment apart from a one-line handwave buried in a README file, and it's a touch nerve-wracking knowing that the next reboot might reveal another fried registry. That threatened OS rebuild is still scary, but becoming rather more appealing.

Also scary is my continuing failure to work out why the backups on both Ros's and my systems are failing - and possibly those on the server, too, just to make a hat-trick! I think that my problem is a defective tape drive, as I installed a new one to go with the new case (I have rather a lot of that particular model, thanks to an odd series of events) but it's certainly not clear... it could be a faulty cable, which might give the same symptoms, or it could be a firmware incompatibility affecting the software compression, or an IDE bus contention issue, or something else that I haven't thought of yet! I need much more time to test and check and replace, and that's problematic while simultaneously trying to debug mysterious failures on Ros's system (the backup runs, but the verify pass aborts immediately with a cryptic "Job cancelled by \" message and no data can be read from the tape) and the server's apparent inability to read any tape from any system. Bah!

There was much head-scratching when all that started to come to light, I can tell you...

1st September

Many more games with the registry today, unfortunately - something is corrupting the SYSTEM hive that stores all the information about the PC's hardware and drivers, and all evidence seems to point to Adaptec's USB2 drivers being the culprit. To be fair, if you take a magnifying glass to a microdot buried in the small print of one of the driver 'README' files, it does mention that the current drivers "don't support" multi-processor systems... But this is not mentioned anywhere else in their product documentation or online support documents, which I think is rather shabby. The advice seems to be to use the generic Microsoft USB2 drivers (I found them via the New Hardware Wizard's "Windows Update" option) but unfortunately I'm still having intermittent boot failures even after carefully removing all the Adaptec drivers and INFs.

So - It looks like I may have to bite the bullet and install a fresh copy of Windows, which would be traumatic indeed considering the complexity of my system. It's long overdue, though, as I'm still using the original installation of Windows from when I built the PC in the spring of 2000 - it's been carefully cleaned and groomed when necessary, but even so there must be an incredible quantity of orphaned data in the registry (and elsewhere!) after more than two years of active use. Performance is likely to be suffering as well as stability, and barring a miracle cure I think it will be the best option...  :-(

There's still a lot of wiring to do inside the PC, but at least the hardware is working well. The noise levels are quite comfortable considering that there are still ten fans running, and once I've laid the temperature probes my gut feeling tells me that I'll be able to clock some of the fans down to an even slower, quieter speed (they're about half-way, at present) as all the components feel comfortably chilly right now.

As well as placing the thermal probes, I have to sleeve nearly all of the power supply cables, install the connections from the CD drives to the sound and TV cards, find a niche for the light switches, and another for the fan speed controllers. There's a fair bit of work still to do, but the bulk of it is done and after the last few days I feel like taking a break for now. Phew!






Weblog Archive