Greg
Greg's diary
July 2004
Translate this page
Select day in July 2004:
Su Mo Tu We Th Fr Sa
1 2 3
4 5 6 7 8 9 10
11 12 13 14 15 16 17
18 19 20 21 22 23 24
25 26 27 28 29 30 31
Select month:
2004 Jan Feb Mar Apr
2004 May Jun Jul Aug
2004 Sep Oct Nov Dec
Today's diary entry
Diary index
About this diary
Previous month
Next month
Greg's home page
Greg's photos
Network link stats
Greg's other links
Copyright information
    
Groogle

Thursday, 1 July 2004 Echunga
Top of page
next day
last day

Finally I'm getting back to normal, but it's not easy. As I've said many times before, setting up a new machine is just too difficult. At least I now have the feeling that I've found a better approach than before: basically, keep all mods to the system and ports configuration under RCS on a separate file system. Then, along with a list of the real ports I've installed (not including the myriad dependent ports), I should be able to create a complete new system almost automatically. At least now I have time to debug the procedure.

Today it wasn't helped, however, by a couple of problems. It seems that FreeBSD 5.2.1 runs really badly on the new MSI K7N2 Delta motherboard , but 5-CURRENT runs well, so ended up doing the installation with the system disk from beeble. This is in fact probably the best way to do it, rather than installing from CD-ROM. In this case, it was somewhat confused by the fact that somebody had broken world, and I had to re-cvsup to get a consistent source tree.

All that was in the background; the good news is that I can get back to some kind of work on my program.


Friday, 2 July 2004 Echunga
Top of page
previous day
next day
last day

Still more work on the new system today. It's falling into place, but it never fails to surprise me how long it takes. Hopefully the work I'm doing now will pay off in the future.

To Grumpy's in the afternoon with a couple of bottle of beer for criticism. To my surprise, both Andrew and Thomas liked them. Damn. Now I don't have any clear way to improve them. Bought some English yeast for one of the next couple of brews, which will be roughly like Brew 30. We'll see how much difference that makes.


Saturday, 3 July 2004 Echunga
Top of page
previous day
next day
last day

Back to working on AUUGN today, in the process debugging some macros and working out an automatic way to build the download file that groff uses to decide which fonts to include in a PostScript document. In FreeBSD, the file is called /usr/share/groff_font/devps/download, and it contains a lookup from font name to the file which contains the description:

# List of downloadable fonts
# PostScript-name   Filename

Symbol-Slanted              symbolsl.pfa
ZapfDingbats-Reverse        zapfdr.pfa

The second entry on each line is the name of a file in the same directory. They start with a line like: %!PS-AdobeFont-1.0: Garamond-Bold 001.003, which gives the information needed for download. The following single line script adds to it:

grep AdobeFont *.pfa | sed 's/:.*://'|awk '{print $2 "\t" $1}' >> download

Apart from that, more work on the automatic system upgrade. Some of the tools are not really laid out for this kind of work: the install target in /usr/src/etc/ does nothing more than to build a couple of whatis files. The real install target is called distribution, and if you invoke it without any special options, it overwrites your /etc file system, provided that all the subdirectories it expects are present: there's a target distrib-dirs to create them, but distribution doesn't (yet) depend on it. Played around with that for a while and was able to install /etc on my virgin disk.

Also working on how to build ports, somewhat confused by the fact that some of them don't appear to work correctly with DESTDIR set:

  /mnt/usr/local/man/man1/s2p.1
/usr/bin/strip: /mnt/usr/local/bin/perl5.8.4: No such file or directory

I'm also still running into various problems with the new motherboard, so since it's apparently as good as identical with the motherboard in wantadilla, which has been running reliably for months, decided to swap the systems around. That was more work than I thought: apart from random issues like a monitor cable disconnecting itself from the monitor, old wantadilla didn't boot properly with the new beeble system: I got watchdog timeouts on both Realtek and a 3Com Ethernet boards, and they didn't go away until I disabled the APIC. Clearly we have problems with recent -CURRENT.


Sunday, 4 July 2004 Echunga
Top of page
previous day
next day
last day

A whole day spent brewing beer! The first couple of times round it was fun, but now it's just looking like a lot of work. Things went relatively smoothly, except that I forgot to read my notes from last time, and so made some minor mistakes (or “suboptimal choices”) which I could have avoided.

In March two of our video recorders died in rapid succession, and we sent them down to an el-cheapo place to have the symptoms (tape eating) fixed. This turned out to be a bad idea; yes, it didn't cost much, but one still had such bad tracking that we can only use it as a tuner for the TiVo, and the other one died again with the same symptoms this evening. It doesn't seem to be worth replacing it. We bought a special offer video recorder from Coles last month for only $129, with 6 head technology and NTSC playback. The firmware's a bit baroque, but we'll survive until I finally get my computer-based system together, so we'll look for another one.


Monday, 5 July 2004 Echunga
Top of page
previous day
next day
last day

Today I should have got on with my program, but I still hadn't finished the system upgrade, and decided that it would be more productive to finish that first. Somehow it took up the whole day. I wish I had a more accurate way of predicting the time things take.

The new motherboard does seem to be defective. Now it's in the old wantadilla, which has been stable for months, but today I had multiple problems: first Netscape crashed multiple times, not in itself such an uncommon occurrence, but it hasn't happened for a while; later I had two spontaneous system resets. How I hate flaky hardware!


Tuesday, 6 July 2004 Echunga
Top of page
previous day
next day
last day

Another hardware day. Finally decided that the new motherboard was flaky enough to send back, even though I couldn't demonstrate obvious problems. Also sent back the motherboard and disk from echunga that died last week. Of course they couldn't replace the motherboard imemediately—it's 9 months old and thus obsolete—so I'll have to wait a couple of weeks for a replacement.

Yvonne came back with the new hardware and also a Samsung cordless mouse that I had enquired about but not ordered—not the first time that they've sent things that I haven't ordered. Still, the reason I didn't order the mouse was because they said it only had two buttons and a roller; it turned out that it has “five buttons” and a roller, at least one on each side. Unfortunately, I don't have documentation, so it'll be a while before I can use the thing.

In addition, Yvonne brought back another video recorder, the same as the one she bought from Coles last month. The last one cost $129; this one cost $99, probably less than the cost of repairing the old failed one. Didn't have the time (or necessity) to install it today.

More interesting than either was my new laptop, a Dell Inspiron 9100 that had been left me by somebody else in the company. In general, hand-me-downs are less than successful, but in this case it was almost exactly what I want, especially since it was a top-end unit. Somehow it's typical, though, that the one thing that interests me—the display resolution—wasn't top-end: only 1680x1050, barely more than 4-year-old sydney. I suspect that the real issue is that Microsoft platforms can't use higher resolution.

Put the new motherboard into new wantadilla, and no longer had the stability problems I had seen before. Instead, I got an interrupt flood on the Ethernet cards. I had seen this before on the previous motherboard, too, but there I could get rid of it by disabling the APIC. This time, disabling the APIC didn't help. I had to reallocate the IRQ to get rid of the problem. Why does each motherboard have its own problems? And yes, of course I set the BIOS settings to the same values.

Still, the system upgrade procedure is working well. I think I've made enough progress now to start again from scratch and confirm that it works.

Despite all this, also found some time to work on my program. Things are now looking a lot more uniform, and I feel happier about it; but it'll still be a while before I finish ironing out the wrinkles.


Wednesday, 7 July 2004 Echunga
Top of page
previous day
next day
last day

How long these things take! The admin work, which I had hoped to have out of the way this time last week, is still with me. To get the system update thing working properly, decided to start again, this time with the new 80 GB disk that I had got yesterday in exchange for the defective /src. There's enough space on that to have two system partitions of 10 GB each and a separate /home partition. That way future upgrades use the alternate system partition, and you don't have to worry about changing hardware around: just boot from the other partition. If it doesn't work as expected, going back to the old version is as simple as rebooting from the other partition.

Also accepted the fact that I had to do something about my mail. My inbox was hitting the 10,000 message mark, and more importantly coming towards the magic 48 MB after which postfix stops saving messages. I thought I could get rid of 5,000 messages relatively quickly. Maybe I did: interleaved with other work, it took me 4 hours, or about 4 seconds per message. Still too long. I should unsubscribe from some lists.

Under those circumstances, it wasn't practical to work on my program, so turned my attention to the new Inspiron 9100, which runs Microsoft “XP Professional”. Getting that to run was quite an experience. Though the machine had already installed all the software, I had to add myself as another user, which required reinstalling the software. “Outlook 2003” required remote product registration, either via the Internet or by phone. I have no intention of taking the risk of connecting any Microsoft machine to the Internet, so I chose the phone method. Type in 42 digits, and get another 42 digits back. You'd think that people would make more of the advantage of Open Source software that you don't have to do this kind of nonsense. It has nothing to do with the cost of the software, just the time and inconvenience it takes to install it.


Thursday, 8 July 2004 Echunga
Top of page
previous day
next day
last day

Still more admin work today, and got to understand the Microsoft box better; it's running a version called “XP Professional”, and although it's pretty much as painful as other versions of Microsoft, it has one good feature: I can access the “desktop” from another machine with the rdesktop program. rdesktop has some strangenesses, and I couldn't get it to stop reversing the mouse buttons, but it means that I have almost a real keyboard and a reasonable sized display to work with.

Also continued work on the upgrade to wantadilla—the machine paniced once in bzero, which doesn't make me very happy. Maybe it's the memory after all. I'll have to keep an eye on it.

In the evening, the Digitrex DVD recorder hung itself up again. It's been less frequent, but it's still there. The only way to reset it is to pull the power plug and reinsert it: if ever a device needed a power switch, it's this one. This time it caused enough of a surge to blow the residual current sensor, quite a feat for a device without a ground pin. When I turned it back on again, the UPS continued to beep, although it had power. Took the thing apart (“no user-serviceable parts inside”) and found, well, no user-serviceable parts. After some investigation discovered that the power connector itself included a well hidden fuse:


https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20040709/big/fuse.jpeg

Image title: fuse          Dimensions:          1536 x 2048, 256 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Friday, 9 July 2004, thumbnails          All images taken on Friday, 9 July 2004, small
Diary entry for Friday, 9 July 2004

 

In case that's still not obvious, it's the black area between the connector and the inscription INPUT. It even contained a replacement fuse, quite a nice idea. Now if only that had been more obvious from the outset.


Friday, 9 July 2004 Echunga Images for 9 July 2004
Top of page
previous day
next day
last day

Finally things are almost back to normal again. I'm continuing my upgrade of wantadilla in the background, and I managed to sort out my problems with the Microsoft box as well: it seems that rdesktop problems were due to a strange feature of “XP Professional” that turns the mouse buttons round for USB mice only (well, also for rdesktop, it would seem). I wonder why anybody would want his buttons reversed only for specific mice. Also managed to install a newer license manager for the Oxford English Dictionary, which actually looks a lot better on rdesktop than it ever did on the machine itself.

Back to work on my program, and I'm roughly where I was two weeks ago, before I decided to rewrite the endian layer, and before I had the hardware problems. The difference is that things look so much clearer now.

Having a Microsoft box on the network means opening yourself to many more vulnerabilities, and I've been working on tightening up my firewall. Here's the most interesting output from iptables for a 24 hour period:

  463 74104 DROP       all  --  sm200d *       0.0.0.0/0            10.0.0.0/8
  147  7136 DROP       tcp  --  sm200d *       0.0.0.0/0            0.0.0.0/0          tcp dpt:80
 1304 63588 DROP       tcp  --  sm200d *       0.0.0.0/0            0.0.0.0/0
   85  6630 DROP       udp  --  sm200d *       0.0.0.0/0            0.0.0.0/0          udp dpt:137
  139 25449 DROP       udp  --  sm200d *       0.0.0.0/0            0.0.0.0/0
  681 43968 DROP       all  --  sm200d *       0.0.0.0/0            192.109.197.95
30454 1470K DROP       tcp  --  sm200d *       0.0.0.0/0            0.0.0.0/0          tcp dpt:135
 5015  243K DROP       tcp  --  sm200d *       0.0.0.0/0            0.0.0.0/0          tcp dpt:139
77563 3746K DROP       tcp  --  sm200d *       0.0.0.0/0            0.0.0.0/0          tcp dpt:445
25889 1261K DROP       tcp  --  sm200d *       0.0.0.0/0            0.0.0.0/0          tcp dpt:80
  203 23656 DROP       all  --  *      *       0.0.0.0/0            192.109.197.169
 1129 54992 DROP       tcp  --  *      *       0.0.0.0/0            0.0.0.0/0          tcp dpt:25
67572 3269K DROP       tcp  --  sm200d *       0.0.0.0/0            0.0.0.0/0          tcp flags:0x0212/0x022
 9738  760K DROP       udp  --  sm200d *       0.0.0.0/0            0.0.0.0/0          udp dpt:137
 4405 2177K DROP       udp  --  sm200d *       0.0.0.0/0            0.0.0.0/0
   21   840 DROP       all  --  *      *       0.0.0.0/0            0.0.0.0/0

There are other rules there too, but they're not as effective. This shows that in the course of 24 hours I have bounced in the order of 12 MB of traffic—not only unwanted, but indiscriminate. That's about a third of my total traffic. It's about time that this kind of abuse were made illegal, world-wide. I suppose it will be one day, but first the legislators need to understand what it's all about. And yes, probably most of the traffic comes from innocently affected Microsoft boxes. That's not an excuse, any more than driving an unroadworthy car is an excuse for having an accident.


Saturday, 10 July 2004 Echunga
Top of page
previous day
next day
last day

Back to working on AUUGN today, and again made relatively good progress. If this continues, I may be able to continue to edit it. But it would be a whole lot better if I could get more people to participate.

In the afternoon, Gareth Andrews came around with a computer on which he wanted to run my temperature control software. Installed FreeBSD 4.10 on it (and, as I later discovered, left the CD in the drive when he took it home), and then the temperature control software. That was worthwhile: I found numerous buglets which needed correcting. Still, all told, including installing FreeBSD, the hardware and the temperature control software, not to mention a bit of beer tasting, we were done in three hours. Gareth knows nothing about FreeBSD, though, and I fear he might have a bit of difficulty coming to terms with different concepts.


Sunday, 11 July 2004 Echunga Images for 11 July 2004
Top of page
previous day
next day
last day

Somehow another day where nothing much happened. Chris Yeardley showed up to pick up a laptop I had got for her, and spent some time looking at that; it's a hot-off-the-press Asus W1000. It has lots of multimedia stuff on it, including a remote control, and I was left wondering whether it was a computer or a TV. Spent some time playing with the multimedia stuff, and in the process brought out my new Inspiron 9100 and discovered it had much of the same stuff on it. I can't help marvelling, though, how bad all this stuff is. It has a thing called “Dell Media Experience™” which claims to have “the features and controls of a normal living-room DVD player”. I suppose that depends on your definition of “normal”, but I'd expect at least slow motion and search functions, all of which are missing. Comparing with Chris' machine, it appears that the “Dell Media Experience™” is really a front end to the equally emetic “Windows Media Player”, though I can't see where the interface is. About the only thing that the latter has in common with the living-room DVD player is the stupid button panel at the bottom of the window:

 

https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20040711/small/wmp.gif
Image title: wmp
Dimensions: 600 x 487, 78 kB
Dimensions of original: 688 x 558, 60 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 11 July 2004:
thumbnails    small images    diary entry

It looks like a toy, but if the “toy” impression isn't strong enough, it's relatively trivial to convert this stupid window into an even more stupid-looking “skin”:

 

https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20040711/small/skin.gif
Image title: skin
Dimensions: 600 x 292, 84 kB
Dimensions of original: 423 x 206, 32 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 11 July 2004:
thumbnails    small images    diary entry

On the other hand, I haven't found out how to display a DVD with it. I put a DVD in the drive, but it didn't find it. When I tried navigating the windows that Microsoft makes you use for finding files, they were there:

 

https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20040711/small/wmp-menu.gif
Image title: wmp menu
Dimensions: 600 x 458, 120 kB
Dimensions of original: 771 x 589, 100 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 11 July 2004:
thumbnails    small images    diary entry

I can also display them if I go via the “Dell Media Experience™”, so it's obviously just a matter of pressing the right mouse buttons. What a pain!

After that, to complete the joy of Microsoft, we tried to set up satellite networking with another machine that Chris had brought. We finally succeeded, but like with so many things about Microsoft, we don't know why. Very frustrating.

In the evening, worked on a new chicken curry recipe to cater to Yvonne's taste. With a bit of work, it won't be bad.


Monday, 12 July 2004 Echunga Images for 12 July 2004
Top of page
previous day
next day
last day

Up bright and early this morning to get back to work on my program. First, noticed that one of my nightly backups had failed because of lack of space, so decided to put in the new 200 GB drive that I had bought for that purpose. A couple of minutes into the backup, the system panicked with a divide fault. Tried to take a processor dump, but it didn't work. On rebooting, the system panicked again, this time with a different problem. Still no dump possible. Tried to do remote debugging and discovered that the sources were on the same system. But I didn't want to move them to where they belonged, because that was the file system I wanted to back up.

Tried a number of different things and established that it was definitely a software bug, presumably somewhere in UFS, triggered by the somewhat unusual parameters I set up for the drive, possibly the -g 1000000000 (“average file size is 1 GB”). Gave up on it for now; the backup is more important. Hopefully I'll be able to reproduce the bug when I have time; it could be an interesting one for my debug tutorial.

After that, thoroughly exhausted, back to work on my program. Didn't get very far. I'm coming to the conclusion that programming is like running: you can sprint for a short distance, but you can't keep that pace up for very long, and you need to find a (slower) sustainable pace.

In the afternoon received a parcel in the mail, apparently containing a dead rat:

 

https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20040712/small/dead-rat.jpeg
Image title: dead rat
Dimensions: 512 x 384, 30 kB
Dimensions of original: 2048 x 1536, 384 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Monday, 12 July 2004:
thumbnails    small images    diary entry

Further investigation showed that it was more harmless: a hop plant that I had ordered a week or two ago. It would have been nice for them to have identified it better.


Tuesday, 13 July 2004 Echunga
Top of page
previous day
next day
last day

More stability problems; zaphod hung up during a level 0 dump shortly after 9 pm last night, presumably the old NFS deadlock. I've been trying to do a dump for 24 hours now. Grrr.

More work on my program today, and finally got back on track with what I had been doing before my current spate of hardware problems. Things are looking good again.

In the afternoon, another dump failure, this time caused by kondoparinga panicking. I wish I knew why I can't get a dump on this machine. Started the dump Yet Again.

Later in the afternoon into town for a normal and a special general meeting of the IT Council of South Australia. The latter was to approve a new constitution, which limited the number of people on the board of directors to 9 (isn't that a popular number?), and elected new members, one of whom was me. It'll be interesting to see how things go now.

In the evening, zaphod hung again. Still no dumps!


Wednesday, 14 July 2004 Echunga
Top of page
previous day
next day
last day

My dump failed again! Only half an hour after I started it, kondoparinga (provisional name for new wantadilla) panicked in the rl Ethernet driver with a VM fault. That's certainly nothing to do with the disk system, but I still couldn't get a dump. I suspect a conspiracy of hardware and software. Addressed the rl problem by replacing the Realtek card with a 3Com, though it's not clear that it's a problem with the card or the driver.

One thing is becoming clear: dump is not the tool to use. It has a number of great disadvantages:

  1. It only works on real file systems.
  2. As a result, you can't make dumps of individual directories.
  3. You also can't dump a network mounted file system.
  4. dump has different formats on almost every system. Even in FreeBSD, restore on release 4 can't read dumps from release 5.

Years ago I recommended tar for these reasons; it has none of these disadvantages (well, there are still some minor format differences, but there are solutions to them). Time to return to tar. Today I started doing a backup of the top-level directories in /src, which took at least as long. It's possible that dump has a slight performance edge. Also, bzip2 must be one of the most CPU-intensive programs I know. It took over 2 hours of CPU time on kondoparinga (currently running as an Athlon XP 2500+) to compress a 17 GB tar archive. At least—by the evening—I completed the dump.

On with my program, and made good progress.


Thursday, 15 July 2004 Echunga
Top of page
previous day
next day
last day

On with my program today, and finally reached the first milestone: I can now store and retrieve data. Now to work on the details of storage.

More work with the air conditioner today, which has been icing up continually. Took the thing apart and put the sensor in the coolest part of the coil; hopefully we'll have some peace now.


Friday, 16 July 2004 Echunga Images for 16 July 2004
Top of page
previous day
next day
last day

Spent most of today looking through some code to add to my program, and finally decided that the easiest way to find out how it worked would be to look at it in action with other code that used it, not made any easier by the fact that my tree was out of date, and that I had unwittingly added updates since. Managed to get that sorted out and found another 5 levels beyond where I had got in the documentation. Debuggers are really useful, especially since the documentation can't say which parser is used. Spent the rest of the day analysing what I had found.

Another letter from Canon today, relating to my scanner. They're not prepared to do anything beyond refund the money for the scanner. I'll probably do that in the end, but I'm left with a bad taste in my mouth. It's clear that this scanner was brought to the market with inadequate software, and since they refuse to release the specs—by no means typical of the industry—they're effectively releasing non-functional products and just putting up with the not-too-frequent complaint. That's not the behaviour of a reputable company. I had in the past toyed with buying a Canon camera. They've certainly ensured that I won't buy anything else from them until I see evidence in a change of attitude.

My hopes to have fixed the air conditioner proved to be premature, as I came out and found the de-icing sensor probe covered in ice. Decided that it was obviously defective, so removed it from the machine for Yvonne to take and get a replacement. When I had done so, I saw the underside of the switch (where the contacts are) for the first time:

 

https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20040716/small/switch-top-1.jpeg
Image title: switch top 1
Dimensions: 298 x 219, 16 kB
Dimensions of original: 1192 x 875, 166 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Friday, 16 July 2004:
thumbnails    small images    diary entry

The third contact (between the other two) is unused; it's normally open. But what's that screw in the middle? It looked like an adjusting screw, and some experiments with the deep freeze and some iced water proved it was. Put it back in the air conditioner, this time with the screw facing upwards, and was able to adjust the thing so that it de-iced properly. What a surprise! It's been over six years since we first reported this problem, and it's all a documentation problem, one that obviously no air conditioner technicians know about. Spent some time adjusting the screw, which is surprisingly sensitive.

I'm also left with the question: should I turn on early, which results in a de-icing cycle of only a few seconds, but which happens relatively frequently, or turn off late, which can take up to 30 seconds to de-ice, but which doesn't happen very often?


Saturday, 17 July 2004 Echunga
Top of page
previous day
next day
last day

Quiet day. I had intended to finish my work on AUUGN today, but didn't make much progress. Bottled some beer in the morning, and spent repeated short periods of time adjusting the de-icing switch on the air conditioner.

I'm beginning to think that the de-icing switch sensor is still in the wrong place. First, some background about the machine and how it works: like all air conditioners and refrigerators, it compresses refrigerant, which heats it up, and then allows it to expand, which cools it down significantly. The trick is maintaining reference temperatures.

Normally, fridges and air conditioners first pass the heated refrigerant through a coil (think: “radiator”) to cool it down to ambient temperature; after further cooling through expansion, the refrigerant is used for cooling something, such as a room or the interior of a fridge.

It can also work in the other direction: after cooling, the refrigerant is warmed up to ambient temperature, again through a coil. Then it's compressed and gets a lot hotter, and that's used to warm things. This is the way we use it in winter.

There are two problems with heating:

  1. The cooled refrigerant enters the coil at temperatures below 0 °C, which causes atmospheric humidity to condense and freeze. That's why we need de-icing. In a completely dry atmosphere, it wouldn't be necessary, and the coil temperature could easily drop far below 0°.
  2. The switch measures temperature, not icing, and the temperature curve is very non-linear. It drops rapidly to 0°, but if enough ice forms, it then stays there: freezing water creates a lot of heat. I can adjust the temperature to just about anything I want, but it needs to be very exact.

All this makes me think that maybe the sensor is mounted in the wrong place, or mounted incorrectly. I can think of two ways to take advantage of the physical situation:

  1. Mount the sensor less closely in contact with the coil. It won't cool down until some ice forms, creating better thermal contact.
  2. Mount the sensor higher up the coil, where the temperature is also higher (the refrigerant enters at the bottom and leaves at the top), and wait for the ice to form there before de-icing.

The sensor was originally mounted higher up, so maybe I should do something about that.


Sunday, 18 July 2004 Echunga
Top of page
previous day
next day
last day

Spent most of the day working on AUUGN, and more or less got finished. A large part of the job was just getting a framework together; next time round should be a lot easier, but I suspect if anybody else takes over editorship, he'll change the tools again, and it'll all be a throwaway implementation. Still, since I'm using my standard macros, it gives me a chance to fine-tune them, particularly the macros for generating the table of contents.


Monday, 19 July 2004 Echunga
Top of page
previous day
next day
last day

Back to normal work today, researching some existing code that I have to put into my program. The main issue is that it's part of a library, and I need to disentangle it. Made reasonable progress with that.

Also paid some attention to mail; I've now passed the 10,000 message mark, and I'm coming dangerously close to the 48 MB above which postfix refuses to store messages.

Also spent more time working on the new system installation stuff. It's getting better, but it's still not done. This is a real pain.


Tuesday, 20 July 2004 Echunga
Top of page
previous day
next day
last day

More work on my program today. Incorporating the existing code into my program turned out to be easier than expected. I had planned for the rest of the week to do it, but by midday I had the first half finished, the half that I thought would be more difficult. That probably means that the other half will be a pig and will keep me busy until the end of next week...


Wednesday, 21 July 2004 Echunga
Top of page
previous day
next day
last day

Today spent most of the morning looking at the second half of this week's work, a library function in our library which I needed to extricate from the functions with which it's bound. Then I discovered that both FreeBSD and Linux have the same functionality in the library. By that time I was just about done with the extraction, so I wrote a test program with both versions and confirmed that they both did the same thing. That's the rest of this week's work done already.


Thursday, 22 July 2004 Echunga
Top of page
previous day
next day
last day

More tidying up work today; it's about time I finally cut over kondoparinga as the new wantadilla. Installing OpenOffice still proves to be a pain, not made any easier by some Ethernet problems: yesterday a card died on me, and today one started reporting ridiculously long packets (over 32 kB). This caused a panic in bcopy, called from rl_rxeof, presumably because the interrupt handler hadn't checked the length before calling it. Unfortunately, I still can't dump the system, so wasn't able to do much except put in another card.

Some performance measurements on the work side of things, with interesting results.


Friday, 23 July 2004 Echunga
Top of page
previous day
next day
last day

Spent some time debugging today, most of it spent debugging the debugging macros. Didn't find the main bug I was looking for.

Apart from that, decided that I should finally complete the transition to the new wantadilla. Yvonne brought back some hardware today, notably a new hard disk for my laptop, so I can now install FreeBSD on the Inspiron 9100. A surprising amount of preparation is necessary for that. Managed to install Microsoft “XP Professional” on the new disk—hopefully moving to a new disk doesn't constitute breach of their EULA—which will free up the big disk for a real operating system or three. Installation was surprisingly easy, and it didn't even overwrite my FreeBSD partition, but at the end of it the only network interface it could find was firewire.


Saturday, 24 July 2004 Echunga
Top of page
previous day
next day
last day

Spent most of today playing around with hardware. We've decided that we can't use the Digitrex DVD player any more, so spent some time looking for alternatives involving our TiVo, but beyond finding a vserver already installed on the machine, didn't make much progress.

Most of the time was taking up building software on laptops. On the Microsoft front, discovered that none of the Dell-specific drivers had been installed. The Microsoft install procedure doesn't cater for installing vendor-specific drivers; considering that Microsoft itself is only half an operating system (native hardware support is minimal), that's particularly bad. At the end of the install, I had to install each driver individually from the Dell CD-ROM, requiring multiple key clicks per driver, with almost no help as to whether I really needed the driver or not.

At the end, rebooted and discovered that, although I had asked Microsoft to install in “Drive D:” (the second partition on the disk), it had installed in the first two partitions; the first is a Dell diagnostic partition, only 39 MB in size. There seemed to be no way past that. In the end, deleted the partition (isn't it nice to have a dd backup of the partitions?) and got it to work. In the process, discovered that FreeBSD's fdisk can also work on copies of MBRs:

=== root@kondoparinga (/dev/ttyp0) /home/Sysconfig 27 -> fdisk /dumpb/adelaide/bootsector
fdisk: device /dumpb/adelaide/bootsector is not character special
******* Working on device /dumpb/adelaide/bootsector *******
parameters extracted from in-core disklabel are:
cylinders=0 heads=255 sectors/track=63 (16065 blks/cyl)

parameters to be used for BIOS calculations are:
cylinders=0 heads=255 sectors/track=63 (16065 blks/cyl)

Media sector size is 512
Warning: BIOS sector numbering starts with sector 1
Information from DOS bootblock is:
The data for partition 1 is:
sysid 7 (0x07),(OS/2 HPFS, NTFS, QNX-2 (16 bit) or Advanced UNIX)
    start 80325, size 30678795 (14979 Meg), flag 80 (active)
        beg: cyl 5/ head 0/ sector 1;
        end: cyl 1023/ head 169/ sector 63
The data for partition 2 is:
sysid 165 (0xa5),(FreeBSD/NetBSD/386BSD)
    start 30759120, size 27846000 (13596 Meg), flag 0
        beg: cyl 1023/ head 170/ sector 1;
        end: cyl 1023/ head 254/ sector 63
The data for partition 3 is:
<UNUSED>
The data for partition 4 is:
<UNUSED>

In the meantime, also installed FreeBSD on the new disk on the other laptop (eucla), using the system upgrade procedure I'm working on. Had a few additional hiccoughs, but it seems to work.


Sunday, 25 July 2004 Echunga
Top of page
previous day
next day
last day

Somehow got nothing done today. Continued installing FreeBSD on eucla, which went a little better than yesterday, and by the evening had the machine pretty much the way I wanted it. The big success story was the new X system from X.org, which installed without problem and started without a configuration file—this on a machine with a 1680x1050 wide screen display. Very impressive.

Spent some time burning CD-Rs for AUUGN. Given the failures we have had in the past, made two copies of each on different burners and checked that the other burner could read the results. They could, but there seems to be something strange about the way they're burnt: I can use dd to read in a CD-R, and it reads in correctly, but then it gets an I/O error trying to read beyond the end:

=== root@eucla (/dev/ttyp2) /var/tmp 14 ->  dd if=/dev/acd0 of=fedora2.iso bs=128k
dd: /dev/acd0: Input/output error
5093+0 records in
5093+0 records out
667549696 bytes transferred in 241.148143 secs (2768214 bytes/sec)

The same thing doesn't happen with CD-Rs burnt on other systems. I wonder what causes it.

Today was the 22nd anniversary of the day I met Yvonne—also a Sunday. A lot of water has flowed under the bridge in that time. Off to Maximilians in Verdun to celebrate.


Monday, 26 July 2004 Echunga
Top of page
previous day
next day
last day

Somehow didn't get much done again today. Spent some time debugging allocation problems on my program, and found a couple of bugs, but they weren't the cause of the problem. Mañana.


Tuesday, 27 July 2004 Echunga
Top of page
previous day
next day
last day

More work on my program today, and found a number of bugs, including the one I was looking for. Using gdb macros certainly makes things easier, but it also has its down side: at least one of the incidences of corruption that I thought I saw was in fact due to a macro misinterpreting data. Still, got past all the current problems and moved on to the next bug.


Wednesday, 28 July 2004 Echunga
Top of page
previous day
next day
last day

Still more debugging, and discovered more supposed bugs which were in fact due to bugs in the debugging macros. If only the gdb macro language weren't so obscure! Found all bugs and fixed all but one: there's one remaining endian issue in the index pointers. That's going to have to wait until tomorrow.

Creating the CD-Rs for AUUGN proves to be more of a headache than I expected: this quarter, amongst other things, we're supplying a CD-R of Fedora Core 2. In the past, we've always only supplied one the first CD of sets, and things have worked well. Not so this time: the first CD apparently doesn't even contain the base X window system. Looks like we'll have to supply CD 2 as well, which makes it look silly that we're not also doing CD 3. Grrr.


Thursday, 29 July 2004 Echunga –> Melbourne
Top of page
previous day
next day
last day

Finally finished download disk 2 of the Fedora Core 2 CD-Rs today, so burnt one and tried to install it on eucla. The installation failed in the middle: the installed system couldn't read the second CD-R, which I had burnt and tested on the same drive. The same thing happened with one burnt on adelaide. Spent a lot of time checking, without coming to any firm conclusion, but it looks as if there might be something about the way FreeBSD burns CD-Rs that doesn't appeal to Fedora's kernel. An installation of Red Hat 9, almost the same system, worked fine and was able to read the CDs. It looks as if both Fedora and FreeBSD have to share the blame here: Fedora should be able to read CD-Rs that other systems can read, and FreeBSD should not burn CDs that are less readable than the ISOs from which they were derived. Grrr. David Newall came to the rescue with the CDs he used to install Fedora, and sent them off to the replicators.

In the process, took a cursory look at Red Hat 9. I hate it! This is a Microsoft lookalike. It took me several minutes just to find a way to start an xterm. When is the computer industry going to get out of sign language mode?

Off to Melbourne in the evening for an AUUG board meeting. Stayed at the Duxton Hotel, not for the first time, but today had great trouble with the air conditioning, which seems to be a problem with the system: the temperature control is as good as useless. Why do people have so much trouble with thermostats?


Friday, 30 July 2004 Melbourne –> Echunga
Top of page
previous day
next day
last day

Up at the crack of dawn today, helped by the inclement temperature in the room, to start the board meeting at an atypical 8 am. As Immediate Past President, I'm taking a back seat this year. Things don't look good: David Purdue, the new President, only intends to stay in the position for one year, and there's talk of winding up AUUG due to lack of interest. Our membership has dropped 50% since November 2000, and there's no evidence (as we thought we had two years ago) that the rate of decline is slowing. One thing that was agreed on (somewhat against my will) was that AUUGN will cease publication as a paper magazine, though we'll continue with some electronic form at least for a while.

Back home in the evening with a feeling that I hadn't really achieved anything worthwhile.

On 10 August 2004, Abby Dinham of ZDNet Australia wrote an article based on these statements. She (he?) didn't contact me first, and as a result she got the wrong end of the stick. For example, she quoted the statement: “there's talk of winding up AUUG due to lack of interest.” Yes, that's true. There's been such talk for at least three years. We'll probably continue to have such discussions for the next few years; it's part of our duty to consider this kind of eventuality and to gauge whether it's going to pose a real threat.

Without the kind of background that regular readers of this diary have, it's easy to misinterpret such statements. Abby seems to have thought that the issue is immediate. I obviously need to be careful what I say in this diary.

As if that wasn't enough, on 13 August, Iain Ferguson, another ZDNet reporter has smelt blood and reinforced Abby's behaviour, performing no further investigation and deliberately ignoring the context or my request to contact me first. It's not surprising that he got things wrong. But then, he couldn't even read what I had written. Reluctantly, I'm going to have to put a copyright on this diary to minimize the number of people quoting me out of context. I don't want to stop people quoting me, but it doesn't help anybody if people go and invent things because they're too lazy to do their research. I wonder if this sloppy reporting is the kind of image that ZDNet intends to project.


Saturday, 31 July 2004 Echunga Images for 31 July 2004
Top of page
previous day
next day
last day

Another “new weekend” day, where I did no real work. Did spend some time looking for video software; it seems that most software is more concerned about silly-looking “skins” than functionality. mplayer seems to be the most popular one, but it doesn't have slow motion play, let alone reverse or slow reverse. After a while started installing xine, which at least promises slow motion, but I have this horrible premonition that I'm going to have to hack mplayer to do what I want. Somehow the free software scene is drifting off-topic.


Do you have a comment about something I have written? This is a diary, not a “blog”, and there is deliberately no provision for directly adding comments. It's also not a vehicle for third-party content. But I welcome feedback and try to reply to all messages I receive. See the diary overview for more details. If you do send me a message relating to something I have written, please indicate whether you'd prefer me not to mention your name. Otherwise I'll assume that it's OK to do so.


Top of page Previous month Greg's home page Today's diary entry Next month Greg's photos Copyright information

Valid XHTML 1.0!

$Id: diary-jul2004.php,v 1.54 2011/04/12 00:04:36 grog Exp $