| Greg's diary |
| September 2006 |
| Today's diary entry |
| Greg's blog |
| About this diary |
| Previous month |
| Next month |
| Greg's photos |
| Greg's mail queue statistics |
| Greg's home page |
| Greg's links |
|
|
|
|
Somehow I got nothing done today, despite working flat out to do so. On advice from Ben Herrenschmidt, downloaded the Ubuntu “alternate” CD image, but that didn't give me much more control over the installation. In particular, no choice of what software to install, certainly not whether to start GNOME or not.
The manual disk partition menu (text mode) is marvellous: it has to have a frame, which makes it one line shorter than it should be. It scrolls, but there's nothing to tell you that. Instead, you're left with a menu that looks complete, but is missing the “Next” button: it looks as if there's no way to continue.
Somehow didn't have the courage to try to continue beyond the basic installation. First I need to fix my “new system” scripts to work with Linux as well, then I'll be able to do things much more quickly.
Instead, back looking at the drivers for the VIA chip. What they supply is supposedly tailored for Fedora Core 4 (now happily obsolete), but part of it at least (marked “kernel”) is intended to be incorporated in a build of X.org as a replacement for the via driver. Started that, but ended up with no accessible header files and even less desire to continue, so put it off until next week. I should take a break from computers over the weekend.
Managed to more or less keep to my intention of not working today, and instead off with Yvonne and her new horse Carlos to Kuitpo forest. Spring has come with a vengeance, and the temperatures were round 25°. Found some interesting wildflowers, and this time took some photos, in the process letting Darah go free. People do that all the time in Westerns, of course, but in real life it's not so easy, and I was surprised that she didn't move from where I left her:
After closer examination, noted that I had let her reins hang in one of the bushes. I suppose that to to her that was like being tied up, so she stayed where she was. Certainly an idea to think about.
The warm weather made itself visible in another manner, too: there's a good reason for the expression “to sweat like a horse”, and Darah usually sweats when riding; but today the sweat had the consistency of whipped cream:
Back home and relented a little with my intentions of not doing anything with computers, and tried once again to install SANE, on the new CVR machine because of the proximity to the scanner. In the past, I had claimed that the name“Scanner Access Now Easy” had a typo: it should have been not, not now. Today it worked much better, though, and I was able to scan in an image relatively easily, though the preview function was broken and only showed a chopped-up SANE logo. Tried again to install on wantadilla (like tv2 running FreeBSD 6.1-RELEASE), and it couldn't find the scanner device, though dmesg showed the same output. I wonder what went wrong there.
Also did some work in the garden, planting a number of new seedlings on the north side of the house, and planned some cooking. We've eaten Husfrusill a couple of times now, but there are so many different Swedish herring recipes that it seems a pity to always make the same one. Found a number by David Weinehall. The one that we chose included the recipe item 1 burk röd stenbitsrom (50 gram). Rom is roe, and clearly stenbit is turbot (German Steinbutt). Looked in the fridge and found a 50 gram jar of Lumpfish (whatever that is) roe, made in Sweden and clearly marked in six languages: in French it's lompe, in Italian and Spanish lompo, in German Seehase (at least I know that name, but not the fish), and in Dutch snotdolf or snotolf. And in Swedish? No mention. Fortunately I had a dictionary, and to my surprise discovered that the Swedish name is stenbit—so it has nothing to do with turbot, which in Swedish is called piggvar. Just goes to show how easy it is to jump to conclusions; but what a surprise that we actually had such an unusual ingredient in the fridge.
Quiet day; spent some time putting yesterday's photos together, and did a little more garden work. Also played around with the new TV machine, without achieving much.
Woke up this morning feeling like I had been hit over the head with a hammer; no idea why, but I was only half awake all day long, and as a result had to cancel my attendance at the planning session for ICT Council for South Australia, which annoyed me somewhat.
Spent some time looking at why SANE didn't work on wantadilla, and discovered that it was, indeed, a device permission problem (on /dev/uscanner0), which doesn't seem to be created and go away when the scanner is attached or detached. Changing the permissions solved the problem. I suppose that, in addition to the bugs in SANE, it begs the question why the device has 644 permission in the first place.
Spent most of the day working on the video device, mainly database and web stuff. Things are looking better, but there's still plenty of work ahead.
Also managed to finish reinstalling Microsoft on pain.lemis.com, which turned out to be less pain than I had expected. Now if I could only find out how to turn this stupid double click selection into a single click. I managed it on the last machine, but working through the maze of twisty little menus is really confusing.
Only one thing is missing from my Microsoft reinstall on pain: the Oxford English Dictionary. I bought mine, software edition 2.0, in December 2001. As I commented at the time I got it, it uses some emetic form of copy protection, and I later discovered that it doesn't work under Microsoft “Windows” XT. In September 2003 I upgraded my Microsoft to this platform and had to contact OED to get updated software; strangely, I didn't mention it in my diary. This time I went to the same place and found that the information was no longer there. So, though the software works, the license manager doesn't. Sent a message off to OED and got a reply by the evening—once again suggesting that I buy the updated edition, this time without specifying a price. Three years ago it was “only” £ 51.06, more than most CD-ROMs cost new.
The AUUG annual conference is due to start on 8 October, less than 5 weeks. We appear to have no sponsors, no programme and no possibility for registration. Sent a message to this year's board and got... no answer.
More programming, and made reasonable progress.
Followed up on the Oxford English Dictionary software today. The information is currently here (not the URL I was given), but basically the instructions are:
Later tried Yet Again to install a version of MythTV on the new CVR box, which I'm calling tv2 for the moment. First I started with the version of Ubuntu that I installed last week. I hate Ubuntu! It installs nothing of value, and it doesn't even have a valid apt/sources.list: it appears to have only the CD-ROM, and that doesn't include everything. In particular, I wasn't even able to apt-get Emacs. I also couldn't find a menu item for mounting NFS file systems. To make matters worse, the xterm replacement in the menu misinterprets all Meta-key functions, so I couldn't even type into it. There is an xterm, but I didn't find it in the menus; and like always (it seems) in Linux today, the normal pointers had been replaced by a symbolized hand with truncated fingers. The whole thing looks more and more like a poor copy of Microsoft.
In addition, had more NFS mount hangs. I can't necessarily blame them on Ubuntu, though in this case the lack of correct DNS configuration might have helped. Possibly I should be installing Debian, but there seems to be no way to install the unstable version directly, and nobody I know uses the stable version.
So I went back to installing KnoppMyth again, having first ensured that my disk was in the right place. That worked better, though still not correctly. On reboot, it asks for the root password which I had had to enter, and then rejects it. On further investigation discovered that the password hadn't been set—and that the login in that particular window wouldn't accept an empty password. So ended up logging in on /dev/tty2, setting the password, and returning. The next step is to configure MythTV, which I'll leave until tomorrow.
On with the KnoppMyth configuration. The “automatic installation” asks very few questions: user name and password, “administrator” password, time zone. As I saw yesterday, It doesn't set the administrator password, and today during setup I discovered it doesn't set the user password or the time zone either (though it does create the user name). It also had the same mutilation of IP addresses in the network setup screens that I had noticed (but not reported) earlier (dropped the first digit, so 192 became 92).
Setup is interesting. Here's a detailed description of what I did.
Also—finally—got round to setting up my beer kegs, something that I started two months ago. The big thing I had been waiting for was Yet Another fridge, which arrived on Wednesday. I now have three: one for fermentation, one for storage (“lagering”), and now one for finished beer. The storage fridge should run at 1°, and it will as soon as I fix my temperature control software to handle more than one fridge, while the serving fridge should run at about 7°.
Drilling a ½" (yes, really) hole in the front of the fridge was a matter of the right tools. The hole cutter I had proved to be significantly less hard than the door, with the result that I lost all the teeth and didn't make much of a dent in the door:
Finally used a jigsaw, which went much more easily than I thought, as did the rest of the installation:
The only problem is learning how to serve it, as seen by the glass on the top of the fridge. It looks like it'll take a while to get used to how to do it.
Living where we do, we often find spiders on the lounge-room floor. Usually it's the cats who find them and chase them. We don't kill them—we catch them and put them outside. Anything that eats flies has to be our friend. Tonight, though, the “spider” moved in a strange way—it hopped:
Lots of work on MythTV, and made some progress; most of the information is in the log. It's not clear whether KnoppMyth thinks the installation is complete or not; rebooting the machine still brings the setup menu. This time I didn't kill it quickly enough, and... it killed my configuration! It completely blew away the mythconverg database, and I had to start again! It's a good thing that I've been writing down what I did.
One seemingly unrelated problem I had yesterday was poor picture quality when recording from the new DViCO card. Spent some time playing around and found that it didn't even complete initialization in ceeveear:
hiddev96: USB HID v1.10 Device [DVICO DVICO USB HID Remocon V1.00] on usb-0000:00:10.0-2 usbcore: registered new driver yealink drivers/usb/input/yealink.c: Yealink phone driver:yld-20050816 usbcore: registered new driver usbhid drivers/usb/input/hid-core.c: v2.6:USB HID core driver Linux video capture interface: v1.00 cx2388x dvb driver version 0.0.5 loaded CORE cx88[0]: subsystem: 18ac:db10, board: DViCO FusionHDTV DVB-T Plus [card=21,autodetected] TV tuner 4 at 0x1fe, Radio tuner -1 at 0x1fe ACPI: PCI Interrupt 0000:00:06.2[A] -> GSI 17 (level, low) -> IRQ 17 cx88[0]/2: found at 0000:00:06.2, rev: 5, irq: 17, latency: 32, mmio: 0xdf000000 cx88[0]/2: cx2388x based dvb card DVB: registering new adapter (cx88[0]). DVB: registering frontend 0 (Zarlink MT352 DVB-T)... CORE cx88[1]: subsystem: 18ac:db10, board: DViCO FusionHDTV DVB-T Plus [card=21,autodetected] TV tuner 4 at 0x1fe, Radio tuner -1 at 0x1fe ACPI: PCI Interrupt 0000:00:0c.2[A] -> GSI 18 (level, low) -> IRQ 18 cx88[1]/2: found at 0000:00:0c.2, rev: 5, irq: 18, latency: 32, mmio: 0xdc000000 cx88[1]/2: cx2388x based dvb card cx88[1]: frontend initialization failed cx88[1]/2: cx8802_stop_dma ACPI: PCI interrupt for device 0000:00:0c.2 disabled cx88-dvb: probe of 0000:00:0c.2 failed with error -1Suspecting a newer hardware revision, I compared the PCI parameters. They were identical for both cards, both under Linux and FreeBSD . Linux says:
0000:00:05.0 Multimedia video controller: Conexant Winfast TV2000 XP (rev 05)
Subsystem: Unknown device 18ac:db10
Flags: bus master, medium devsel, latency 32, IRQ 17
Memory at de000000 (32-bit, non-prefetchable) [size=16M]
Capabilities: [44] Vital Product Data
Capabilities: [4c] Power Management version 2
0000:00:05.2 Multimedia controller: Conexant: Unknown device 8802 (rev 05)
Subsystem: Unknown device 18ac:db10
Flags: bus master, medium devsel, latency 32, IRQ 17
Memory at df000000 (32-bit, non-prefetchable) [size=16M]
Capabilities: [4c] Power Management version 2
FreeBSD says:
none4@pci5:4:0: class=0x040000 card=0xdb1018ac chip=0x880014f1 rev=0x05 hdr=0x00
vendor = 'Conexant Systems, Inc'
device = 'CX2388x TV Capture Card'
class = multimedia
subclass = video
none5@pci5:4:2: class=0x048000 card=0xdb1018ac chip=0x880214f1 rev=0x05 hdr=0x00
vendor = 'Conexant Systems, Inc'
device = 'CX2388x TV Capture Card'
class = multimedia
So do I have a dud card, or is there some undocumented change to the cards that the current drivers can't handle? It would be nice to know what frontend initialization failed means.
While investigating the problem, tried to install FreeBSD on the machine; it failed consistently with disk access problems. Presumably it doesn't understand the new hardware. I've seldom had so many compatibility problems with a motherboard.
On the suggestion of Jan Jones, decided to install MythDora, a Fedora-based MythTV system, on the machine. As usual, finding the documentation is not easy, and managed to download an ancient ISO image (release 2.1; current is 2.32) at a snail's pace (took nearly 3 hours). By that time I had found the correct version and started another download, but it wasn't ready by the evening. The documentation promises a step-by-step pictorial illustration of the installation process, but in fact only gets as far as the end of the standard Fedora installation. I'll have to wait until tomorrow to find out what the installation really looks like.
While waiting for the MythDora download, also tried installing Microsoft “Windows XP Media Center Edition” from the MSDN DVD set. It got about half way and asked for CD 2, whatever that may be. Looks like a bug in the packaging.
To keep boredom away, did some work in the garden, planting some of the many seedlings we have raised over the last 12 months. Things have changed a lot since then. Photos to follow.
Finally burnt an ISO for MythDora and booted it. The results weren't encouraging. It did confirm that the sound hardware is working, the first distribution to do so, but that was about the only good news:
So currently I'm dead in the water. There doesn't seem to be any reason why the X server won't start—on booting it recognized the vesa driver, and this particular image is designed for the VIA chip sets anyway. As I said: not very encouraging.
It's becoming clear that I need to split this problem into hardware and software issues; probably the best is to swap the old and new machine and get the new machine to do the video recordings until I can get MythTV running on the old one, and then address the display hardware issues on the new one.
More work on the black box today, and discovered that there's still a surprising amount of work to do on the build infrastructure. Got that pretty well sorted out, but there's still the can of worms of the startup scripts to look at, and there's plenty to keep me busy there.
In the afternoon, swapped the disks and tuners between ceeveear and tv2 and did a test recording. A good thing too; the “old” card, which works perfectly in ceeveear, produced numerous error messages in the VIA motherboard:
Sep 12 16:09:47 ceeveear kernel: cx88[0]/2: cx8802_restart_queue Sep 12 16:09:47 ceeveear kernel: cx88[0]/2: cx8802_restart_queue: queue is empty Sep 12 16:09:47 ceeveear kernel: cx88[0]/2: queue is empty - first active Sep 12 16:09:47 ceeveear kernel: cx88[0]/2: cx8802_start_dma w: 0, h: 0, f: 2 Sep 12 16:09:47 ceeveear kernel: cx88[0]/2: setting the interrupt mask Sep 12 16:09:47 ceeveear kernel: cx88[0]/2: [d375cb80/4] cx8802_buf_queue - first activeThe resultant recording was full of artefacts that roughly paralleled the error messages, so I put the hardware back again and had no further problems.
Somehow I've not had luck with this board (an MSI K9VGM-V with the Via K8M890 chip set):
Looks like time to investigate alternatives. It seems that the nVidia nForce 550 would be a good choice, but I haven't found any in MicroATX form factor.
More work on the installation today, mainly setting up the web server and MySQL. Ran into an unexpected problem: to install PHP into Apache, you need to add some lines to the httpd.conf file:
--- httpd.conf 13 Sep 2006 02:48:01 -0000 1.1 +++ httpd.conf 13 Sep 2006 05:29:59 -0000 1.2 @@ -100,6 +100,10 @@ LoadModule userdir_module libexec/apache22/mod_userdir.so LoadModule alias_module libexec/apache22/mod_alias.so LoadModule rewrite_module libexec/apache22/mod_rewrite.so +LoadModule php4_module libexec/apache22/libphp4.so + +AddType application/x-httpd-php .php +AddType application/x-httpd-php-source .phps <IfModule !mpm_netware_module> # @@ -208,7 +212,7 @@ # is requested. # <IfModule dir_module> - DirectoryIndex index.html + DirectoryIndex index.html index.php </IfModule> #
So I did that, installed some PHP files, and got the message “ You have chosen to open index.php, which is a: PHP file... What should Firefox do with this file? ”. This message is indicative of the brain death that associates file names with their contents, but also that Apache doesn't seem to have recognized the file type. Spent quite some time discussing the matter on IRC, and finally discovered that the message is completely misleading: the file in question was created by phpMyEdit, and it seems that I had not copied something important across. But what a misleading message!
Finally the sun stopped shining, so I could get some photos of what we've been doing in the garden. We started on garden beds about 16 months ago: here's what's happened in the meantime.
The AUUG 2006 annual conference is due to start in 27 days. As of this morning, no information beyond a preliminary programme had been published. This is somehow typical of the way things have been going in AUUG in the last few years, and I wrote a letter to the membership suggesting that we should close AUUG down. That generated a fair amount of discussion on the AUUG-talk list, and also some in the Linux Australia circles: in particular, Jeff Waugh seriously discussed whether AUUG and Linux Australia should merge, something I've been playing with for years. I wonder if the time has finally come.
More work on installability. I still have trouble with phpMyEdit. It installed so easily the first time, but now it's not working. I suspect that I have missed some dependency, but I can't find out which.
At least yesterday's mail message has stirred up a lot of discussion, both on the AUUG-talk and the Linux Australia lists. The latter seems to be getting involved in nomenclature discussions that are probably not going to lead anywhere. It also got the AUUG board to finally release some information about the AUUG 2006 conference. I hope it's not too late.
And, of course, zdnet published an article in their usual inaccurate fashion: “However, the group has suffered a dwindling membership over the years, leading to a discussion kicked off in mid-2004 about finally closing its doors.”. This “discussion” was their last lot of inaccurate reporting about the contents of this diary. As on that occasion, I wasn't contacted first (in fact, I found out about the article from somebody on IRC), and the author obviously didn't read what I had written there: “There's been such talk for at least three years.” I wonder if zDnet think that it's good journalism not to follow leads.
More work on getting phpMyEdit working on the black box, without success: when I select the editor page, I just get an empty screen. No error messages, nothing. About the only clue I have is that the installation information is wrong: “phpMyEdit requires web server (we recommend Apache), PHP interpreter without any special modules and MySQL relational database management system”. I've already established that, at least under FreeBSD, it also needs the devel/php4-pcre port. I still suspect I'm missing something else. This is really frustrating.
Over to Hahndorf in the afternoon to try out my freshly installed disk. Not a total success; I need to reconsider how I do this work. Driving 35 km there and back is inefficient if I have to give up after an hour.
Quiet day, spent mainly in the garden and out riding. We went to the Chookarloo part of Kuitpo forest, which has been turned into a campground over the last few years (and had the spelling of its name changed; it used to be called Tjukarlu; I don't know if the new spelling is better). Horse floats aren't allowed into the parking space, so for a long time we had assumed that it wasn't open to horses, but in fact it seems that only the floats are prohibited (a good thing considering the space available). It's quite pretty there now.
Over a wooden bridge to get there; for some reason Yvonne was concerned about the horses getting excited, but it worked fine, and after a couple of times I was able to ride over with no problems:
While there, tried out my hypothesis that Darah would stand still if her reins were held only a little bit. It proved to be incorrect. I laid the reins over a signpost, and she just walked off. Even when I hooked them round a post, she wandered around:
So I still don't know what the reason was two weeks ago.
What a terrible year it's been! I've already reported on the accidents that Yvonne has had, but that's nothing compared to other developments. Her father died, her brother has been in hospital for nearly two weeks while incompetent doctors do a good job of ignoring his condition, his brother-in-law died and wasn't found for a week, his mother-in-law has broken her knee, and this morning I heard the news that Rob Levin, my ex-boss at Linuxcare, was hit by a car while riding his bike a few days ago and died of the head injuries he sustained. What a sad couple of months!
To make things worse, people on the BSD IRC channels had nothing better to do than to drag Rob's name into the ground. He was never a very popular person, but his heart was in the right place, and it's sad to see people picking on him now that he's dead.
Didn't do much work in the day; the weather has been unseasonally warm, and there was work to do in the garden. While mowing the lawn, a piece of wire got caught in the chopper. While pulling it out, found a grille that looked like part of an old mobile phone. Wondered who would have left a phone lying around here, reached for my hip and... no phone. All that I could find was the leather cover (almost undamaged) and parts of the front:
Went out a couple of times later to look for the rest (in particular, it seemed reasonable that the SIM card might have survived), but didn't find it until it occurred to me that the phone could have been slung out of the mover with some force, so went looking for traces, and found them:a tongue of chopped grass leading out past the fence into the adjoining paddock:
On a hunch, installed mplayer on my old Dell Inspiron 5100, now over three years old. It has a 2.4 GHz processor, but that's faster than the modern crop of Athlon 64s, and it shows; I could display 720p MPEG-2s with almost no problem, something that's still an issue with the Athlon. Looks like I may have to go for a faster-clocked Pentium.
Called up the insurance today about my phone, and discovered I could spend up to $200 on a replacement, so into Mount Barker to take a look. Somehow mobile phone manufacturers have lost the plot: they all have cameras, and Nokia now sells phones with MP3 players and FM radio—but only the most expensive have tri-band capability or Bluetooth. After considerable comparison, hampered by lack of documentation and the refusal of the salesman to let me see the manual, decided on a Motorola L6, which came with a Bluetooth headset (and Bluetooth capability, of course), for a total of $150. The cheapest Nokia machine with Bluetooth costs more than double that.
Back home and was able to marvel once more at Motorola's poor user interface. I never know when the machine has been powered on, and end up holding the off button down for 10 seconds until it finally comes on. The Bluetooth dongle is even worse. The Quick Start instructions state: “With your wireless module off, press and hold the MFB until the LED remains on (about 5 seconds)”. No explanation of what the MFB is—that's in the full instructions, which makes the Quick Start rather pointless. It stands for “Multi-Function Button”, and means the big button in the middle of the dongle. Tried that, and the light went on, but only flashed and then went out again. Finally called the Motorola hot line, was assigned a customer number, and told to hold it even longer (in fact, about 7 seconds). No mention anywhere that it first flashes a bit and then goes out, let alone why.
It turns out that my salvaged SIM card is dead after all. The phone came with a prepaid SIM card, so investigated that: it's not very attractive, because they don't work outside Australia. So I'll need to get a new SIM card. In the meantime tried activating the card via the web, which failed because the software insisted that I have a mailing address with a street number, which I don't have, whichever way you want to look at it. Spent some time on the phone with Abby, who also wanted me to live in a suburb, and was genuinely surprised to hear that suburbs only cover a diminishingly small part of Australia. Ended up entering the name of the house and the road, something that the web forms wouldn't have allowed me to do. What a pain!
Also converted Yvonne's phone to prepaid; she ended up with $1 of credit, and to get more I had to go to a web site. I've done this before in Sweden, where it worked well. Not here: Telstra has surpassed itself with confusing, badly rendered and badly functioning pages. After finding the page for Recharging, I had to log in, pick up a “PIN” (really a one-time password) from the phone, and then log in (there's a tab My Account at the top of the home page). That didn't work for a number of reasons: first of all, it was so badly rendered that the Login button overlaid the password box:
When I tried to select it I got a message that the search failed. Tried it on the Apple machine, where I got almost illegibly small text (recall that this is a 140 dpi monitor), but discovered that it was login for something else.
At this point I was randomly “selected” for a questionnaire about the web site. Spent some time on that, giving correspondingly bad marks; it's a good thing they didn't ask at the end of my experience, where it would have been much worse still. Left contact details in the tiny little window for comments (one line, 15 characters wide, but scrolls; what are people thinking when they design these forms?). It'll be interesting to see whether anybody contacts me.
Finally found the correct login page by following the Recharging link and selecting the Login to My Pre-Paid Online link on the right. The resultant screen is impossibly broken, and I was not able to select anything from the menu:
The lines are part of the poor rendering; further images here. Leon confirmed that yes, indeed, the recharge was successful. Presumably it's been deducted from my credit card. So what is this nonsense? My best bet is a database problem. Asked Leon to put in a complaint, but they don't issue complaint numbers in that part of Telstra. It'll be interesting to see if anything comes of it.
Another day with little to show for it. In the morning followed up on a lead from Ken Tom about my phpMyEdit problem. As suspected, it was a missing port: database/php4-mysql. I suppose you could forgive the documentation for that, since it's needed to use PHP and MySQL together at all. After installing the port, you also need to add the following line to the PHP extensions file /usr/local/etc/php/extensions.ini:
extension=mysql.soI still can't find a good way to make a port out of phpMyEdit, since you need to have somewhere to install the scripts.
Telstra still kept me busy today. Into Mount Barker to get a new SIM card, which went surprisingly quickly, and was even free—yesterday I had been told it would cost $33, but today I was told that, since it had failed, it would be replaced. It wasn't as if I had hidden the fact that it had been through a lawn mower. Also found a possibility of finally getting the phone bill in my own name instead of Linuxcare's—all previous attempts have failed because I need confirmation from Linuxcare, and they don't seem to exist any more. Now I should be able to present a statutory declaration en lieu. I had just got home again when I got a phone call: my accessory kit for the phone had arrived, and was doubtless already there when I left the shop.
That left us with two prepaid SIM cards, each with $10 of credit. There's a scheme (childishly called Credit Me2U) to transfer credit to another phone, so did that, and discovered I could only transfer an integral number of dollars, and there's a $0.25 fee, so I ended up transferring $9 and having a remaining credit of $0.75. Oh well.
The second card was another problem: it hadn't been registered yet, and Telstra's site is broken, so I couldn't register easily. Even when I put in a dummy address in Echunga, it refused. Called up the phone line, which required me to enter the number of the phone via the touch pad. For the first time in my experience it got the number wrong, though it was clear from the display that I had entered it correctly. Spoke to Josh, who first told me that there was no such scheme as Credit Me2U; when I told him the name, he checked, confirmed the existence of the scheme, but told me repeatedly that the transfer wouldn't work anyway, and seemed to ignore my statements that I had just performed such a transfer. These people infuriate me.
He transferred me to his supervisor, Terry, who insisted the same thing, but registered me, this time at Lot 145, Hundred of Kuitpo (if he got the spelling right), SA 5000. He repeated his insistence that the transfer wouldn't work, because it was a bonus balance. At least got a snicker out of him when I told him that it was about the only thing that did work. Finally got off the line and transferred the balance with no difficulty. I wonder if this is not intended to be possible, so they insist in the face of proof.
What a waste of time mobile phones are! Spent some time trying to work out how to create my own ring tone and download it. The manual mentioned almost nothing about how to use the digital interfaces (Bluetooth and USB, the latter supplied without a cable). Fortunately it's a standard mini-USB connector, so I connected it up to the Apple, which said it couldn't power the unit, and that I should connect the power cord too. Problem: it's powered by the USB connector. FreeBSD just reported an I/O error. Searching the Motorola web site just brought a picture of a CD, and it gradually became apparent that the software is sold separately—in some cases I found on the web, for more than I paid for the phone. What does Motorola expect to gain by that? The advertising is so poor that they can hardly sell any of them. Was not able to find alternative software within a reasonable time, so it looks as if that will be shelved.
Over to Hahndorf to talk to the Peters—Peter Jones is now back from Europe. Came back with lots of work to do.
Since starting using my beer kegs a couple of weeks ago, I've been having serious trouble with foam: to paraphrase a mail message I sent to the Australian Craftbrewers' web site,
I end up with a half litre of foam and no beer (initially). That's presumably a common fault, but I've tried all the remedies:
- Pressure too high. I've gone as low as 30 kPa with little or no improvement. I'm currently running at about 60 to 70 kPa.
- Hose between keg and tap too short. I have 2 metres.
- Open tap flat out and shut quickly. Not quite as bad, but still not good enough.
So what else can I try? I'm beginning to think that it's my beer, which has a pretty good head. And then I'm reminded of the joke that I heard in Germany some time ago (as you can tell by the names):
Ronald Raygun invites world leaders to Camp David: Maggie Thatcher, Leonid Brezhnev and Helmut Kohl. He shows off his new toy, a swimming pool.Maggie looks at it and says, “But it's empty”.
Raygun: Right, that's the trick. Get up on the diving board, call out the name of your favourite drink, and jump. By the time you hit the pool, it'll be full of it.
So Maggie gets up, shouts “Gin!” and jumps in. Pool fills with gin, Maggie swims around for a while until they have to fish her out.
Brezhnev get up, shouts “Vodka!” and jumps in. Stays a little longer before he's fished out.
Kohl gets up, shouts “Pils!” and jumps in. Falls flat on his face in the empty pool. Everybody knows it takes 5 minutes to pour a Pils.
I thought that my problem could be related.
Today, however, I found the reason: following the instructions I got with the keg, I had initially set the pressure to 250 kPa before dropping it to a more reasonable pressure. It seems that it takes a very long time for the carbonation to drop. I finished the keg yesterday, and started the new keg with 50 kPa; it was completely flat. After some experimentation, 100 kPa is reasonable, and I can probably go a bit higher.
Diane Shelander from New Zealand (really Oakland CA) over in the afternoon. She bought Serafina some time ago, and was passing through on the way to Alice Springs.
Spent most of the day looking at DVD authoring software and confirmed my past observations that the coding standard in this area is terrible. Beyond the function name, I have no idea what this code is supposed to do (no comments, of course), but it must be possible to make it more readable:
int readscr(const unsigned char *buf)
{
return ((buf[0]&0x18)<<27)|
((buf[0]&3)<<28)|
(buf[1]<<20)|
((buf[2]&0xf8)<<12)|
((buf[2]&3)<<13)|
(buf[3]<<5)|
((buf[4]&0xf8)>>3);
}
Doubtless a little documentation of the underlying data structure would work wonders.
We've seen enough limitations of phpMyEdit to make it worth investigating another similar product, Dataface. Downloaded and installed the software, which was pretty similar to phpMyEdit. Running it was different, though, a command line interface that is definitely inferior to the web-based interface of phpMyEdit.
That last line is indicative of a missing \n at the end of the output. There's not as much choice as with phpMyEdit, but it seems to work. At least until I try to access the page, which seems to be empty. I've seen that particular issue with phpMyEdit as well, but this time the actual code appears to be empty, and so were the config directories. Should I continue? The screen shots look more professional than phpMyEdit, so maybe I should. But I wish I didn't have to spend so much time debugging these things.=== grog@echunga (/dev/ttyp3) /web/data/dface 125 -> ./makesite $W/testit grog:@wantadilla/expenses .MySQL password: (in fact, I don't have one for local access) Copying conf.ini file to '/web/data/testit/conf.ini'...Found table: expenses . Adding to application menu... Found table: yvonne_expenses . Adding to application menu... Copying .htaccess file to '/web/data/testit/.htaccess'... Copying index.php file to '/web/data/testit/index.php'... Creating tables directory at '/web/data/testit/tables'... Creating config directory for table 'expenses' at '/web/data/testit/tables/expenses'... Creating config directory for table 'yvonne_expenses' at '/web/data/testit/tables/yvonne_expenses'... Site successfully created at '/web/data/testit'.=== grog@echunga (/dev/ttyp3) /web/data/dface 126 ->
Also did some further searching for means of communicating with my new mobile phone.
Spent most of the day learning about DVD mastering. Pioneer has some relatively useful documentation on the subject, in particular a Technical guide that is unfortunately somewhat out of date and doesn't mention DVD+R at all (only DVD-R). But somehow none of these documents give me the kind of overview I want. The way I see it, there are two different structures to understand:
The DVD is divided into titles. You don't need to investigate the data format to know that, but it's useful to know that the data structures include a single Video Manager Group (VMG) and several Video Title Sets (VTS). The latter represent the individual titles.
The payload data for each of these components is stored in VOB (video object) files, which as far as I can tell are always a specific kind of MPEG-2.
Metadata for each of these components is stored in Video Title Set Info (VTSI) files. They contain things like chapter markers (currently of particular interest to me), subtitles, camera angles and things.
The VMG is divided into the Video Manager Information (VMGI) and the Video Object Set for the VMG menu (VMGM_VOBS). At least, that's what the Pioneer docco says; it looks pretty clumsy to me. Based on my current understanding, these are basically the DVD menus.
The file system structure. DVDs have both UDF and ISO 9660 file systems (they use the term UDF bridge in a manner which is not defined clearly enough). There's a relatively strict hierarchy, since these files need to be readable by DVD players and other systems with low intelligence. There are up to three top-level directories:
Here's an example from a DVD and a DVD+RW, taken from a mount of the DVDs as an ISO 9660 file system under FreeBSD. First, the DVD:
total 1 dr-xr-xr-x 1 root wheel 2048 Jan 1 1970 audio_ts dr-xr-xr-x 1 root wheel 2048 Jan 1 1970 video_ts /cdrom/audio_ts: total 0 /cdrom/video_ts: total 7736 -r-xr-xr-x 1 root wheel 14336 Jan 1 1970 video_ts.bup -r-xr-xr-x 1 root wheel 14336 Jan 1 1970 video_ts.ifo -r-xr-xr-x 1 root wheel 36864 Jan 1 1970 video_ts.vob -r-xr-xr-x 1 root wheel 18432 Jan 1 1970 vts_01_0.bup -r-xr-xr-x 1 root wheel 18432 Jan 1 1970 vts_01_0.ifo -r-xr-xr-x 1 root wheel 52074496 Jan 1 1970 vts_01_0.vob -r-xr-xr-x 1 root wheel 24506368 Jan 1 1970 vts_01_1.vob -r-xr-xr-x 1 root wheel 49152 Jan 1 1970 vts_02_0.bup -r-xr-xr-x 1 root wheel 49152 Jan 1 1970 vts_02_0.ifo -r-xr-xr-x 1 root wheel 51853312 Jan 1 1970 vts_02_0.vob -r-xr-xr-x 1 root wheel 1073709056 Jan 1 1970 vts_02_1.vob -r-xr-xr-x 1 root wheel 1073709056 Jan 1 1970 vts_02_2.vob -r-xr-xr-x 1 root wheel 1073709056 Jan 1 1970 vts_02_3.vob -r-xr-xr-x 1 root wheel 821295104 Jan 1 1970 vts_02_4.vob -r-xr-xr-x 1 root wheel 51200 Jan 1 1970 vts_03_0.bup -r-xr-xr-x 1 root wheel 51200 Jan 1 1970 vts_03_0.ifo -r-xr-xr-x 1 root wheel 63410176 Jan 1 1970 vts_03_0.vob -r-xr-xr-x 1 root wheel 1073709056 Jan 1 1970 vts_03_1.vob -r-xr-xr-x 1 root wheel 1073709056 Jan 1 1970 vts_03_2.vob -r-xr-xr-x 1 root wheel 1073709056 Jan 1 1970 vts_03_3.vob -r-xr-xr-x 1 root wheel 655984640 Jan 1 1970 vts_03_4.vobThere are a number of things to note here:
DVD+Rs (and presumably also DVD-Rs) also have a directory video_rm:
Example 2: DVD+RW dr-xr-xr-x 1 root wheel 2048 Sep 4 02:46 video_rm dr-xr-xr-x 1 root wheel 2048 Sep 4 02:46 video_ts ./video_rm: total 1 -r-xr-xr-x 1 root wheel 116736 Sep 4 02:46 video_rm.bup -r-xr-xr-x 1 root wheel 8192 Sep 4 02:46 video_rm.dat -r-xr-xr-x 1 root wheel 116736 Sep 4 02:46 video_rm.ifo ...
I haven't found out what it's for yet, though clearly it follows the same file naming convention, and from that naming it's presumably global in scope.
Non-computer day today, mainly. Tomorrow is to be brew day, and Tom Maynard had asked me to show some photos of what I do then, so started today with crushing the grain. That didn't quite work out the way I expected: in the middle of it, the drive shaft of the mill broke off in the socket:
Fortunately Grumpys were able to help, so Yvonne went over there to get the rest crushed.
A couple of weeks ago Yvonne had bought a surprisingly cheap venison roast in the Central Market, and we had roast it as recommended; that wasn't a complete success. The meat was OK (at $8 a kilogram, obviously not first choice), but the method of preparation wasn't appropriate. Next time we'll try something like a German Wildgulasch. On this occasion, we decided to prepare the last 300 g remaining as a terrine, and that's what I started today, severely hampered by the lack of certain basic ingredients in Australia (not even fat!). I'll write this up when I'm done, but here's the current state of my terrine de chevreuil:
Woke up in the middle of the night and looked at the (LED) clock in the bed and—it was out. That's particularly bad news: it means that not only have we had a power failure, but that the main UPS has run out of power, taking all my main machines with it. echunga had been up for nearly 300 days. Only the laptop and brewer, my brewing machine on a separate UPS, survived.
When I got up, took the opportunity to upgrade echunga to the most recent version of FreeBSD. Somehow I get these things out of phase:
Sep 24 09:26:18 echunga kernel: FreeBSD 6.1-PRERELEASE #0: Sun Feb 5 14:15:02 CST 2006 Sep 24 12:10:45 echunga kernel: FreeBSD 6.2-PRERELEASE #1: Sun Sep 24 11:49:40 CST 2006
The PRERELEASE tag exists only for a short time (about a week or two) before each release.
Upgrading went relatively smoothly, but I seem to have hardware problems with wantadilla: when I first tried, it wouldn't power on, and I started preparing to change the power supply before it decided to work after all. When it did, it took a while to recognize the primary disk, and even spontaneously rebooted. But then it settled down—for now. Also, ceeveear disappeared off the net. ping showed:
=== grog@wantadilla (/dev/ttyp1) ~ 20 -> ping ceeveear
PING ceeveear.lemis.com (192.109.197.140): 56 data bytes
64 bytes from 192.109.197.140: icmp_seq=8 ttl=64 time=10009.458 ms
64 bytes from 192.109.197.140: icmp_seq=9 ttl=64 time=9008.767 ms
64 bytes from 192.109.197.140: icmp_seq=10 ttl=64 time=8008.009 ms
64 bytes from 192.109.197.140: icmp_seq=11 ttl=64 time=7007.248 ms
64 bytes from 192.109.197.140: icmp_seq=12 ttl=64 time=6006.486 ms
64 bytes from 192.109.197.140: icmp_seq=13 ttl=64 time=5005.724 ms
64 bytes from 192.109.197.140: icmp_seq=14 ttl=64 time=4004.965 ms
64 bytes from 192.109.197.140: icmp_seq=15 ttl=64 time=3004.205 ms
64 bytes from 192.109.197.140: icmp_seq=16 ttl=64 time=2003.448 ms
64 bytes from 192.109.197.140: icmp_seq=17 ttl=64 time=1002.685 ms
64 bytes from 192.109.197.140: icmp_seq=31 ttl=64 time=0.996 ms
64 bytes from 192.109.197.140: icmp_seq=19 ttl=64 time=12011.031 ms
64 bytes from 192.109.197.140: icmp_seq=20 ttl=64 time=11010.272 ms
64 bytes from 192.109.197.140: icmp_seq=21 ttl=64 time=10009.509 ms
64 bytes from 192.109.197.140: icmp_seq=22 ttl=64 time=9008.747 ms
64 bytes from 192.109.197.140: icmp_seq=23 ttl=64 time=8007.981 ms
64 bytes from 192.109.197.140: icmp_seq=24 ttl=64 time=7007.217 ms
64 bytes from 192.109.197.140: icmp_seq=25 ttl=64 time=6006.462 ms
64 bytes from 192.109.197.140: icmp_seq=26 ttl=64 time=5005.690 ms
64 bytes from 192.109.197.140: icmp_seq=27 ttl=64 time=4004.925 ms
64 bytes from 192.109.197.140: icmp_seq=28 ttl=64 time=3004.164 ms
64 bytes from 192.109.197.140: icmp_seq=29 ttl=64 time=2003.404 ms
64 bytes from 192.109.197.140: icmp_seq=30 ttl=64 time=1002.645 ms
64 bytes from 192.109.197.140: icmp_seq=32 ttl=64 time=6699.138 ms
64 bytes from 192.109.197.140: icmp_seq=33 ttl=64 time=5698.364 ms
64 bytes from 192.109.197.140: icmp_seq=34 ttl=64 time=4697.611 ms
64 bytes from 192.109.197.140: icmp_seq=35 ttl=64 time=3696.843 ms
64 bytes from 192.109.197.140: icmp_seq=36 ttl=64 time=2696.150 ms
64 bytes from 192.109.197.140: icmp_seq=37 ttl=64 time=1695.390 ms
64 bytes from 192.109.197.140: icmp_seq=38 ttl=64 time=694.629 ms
64 bytes from 192.109.197.140: icmp_seq=39 ttl=64 time=13645.678 ms
(etc)
Those ping times come in groups with times descending by almost exactly one second. Since they're issued a second apart, this means that groups of replies came together. In other words, the interface only worked briefly every 10 to 15 seconds. I've seen this numerous times over the years, both under FreeBSD and under Linux. I wonder what causes it. In many cases, disconnecting the Ethernet cable and replacing it fixes the problem, but it didn't work today. In all cases, I've been able to fix it with:
ifconfig eth0 down ifconfig eth0 up
Spent most of the day brewing. Today was different: Tom Maynard had asked me to show some photos of what I do, so took many photos for a Brew Day page. That took up most of the day.
Back to work today, spending some time considering our application structure, which could do with improvement. Not for the first time, I'm left with the trade-off between a quick fix and “doing it right”. Obviously I'd prefer the latter, but time constraints often make it difficult, if not impossible.
I seem to have problems with my network in the living part of the house: yesterday there were problems with ceeveear, and today brewer, on a wireless network, stopped responding. It doesn't have a monitor, so my normal method is to turn off the UPS (because that's more accessible than the reset button) and let it go again. That didn't work today, and I had to drag in hardware:
But there's something more sinister going on here. I suspect I haven't seen the last of it.
More problems with the networking today: brewer disappeared off the (wireless) network, and teevee off the wired network, both connected by the same access point/switch combination. Changed the switch (I have two of them), but the other one was set up for a different wireless channel, and for some reason neither brewer nor tvremote (also on the same wireless network) found the new channel. Why wasn't the new channel they found automatically? Or is the other one defective too? Ended up putting the original switch back and hard rebooting both brewer and teevee. High time we fix the problem.
We're still having problems with authentication for the CVS server, so revisited the whole thing and wrote a HOWTO for our specific situation. That worked fine, but Peter Denton tells me he still can't authenticate. I wonder what's going on there.
Spent the afternoon looking at DVD formats. It's a real can of worms. The following is an attempt to structure it in a way that's easy for me to understand. It may contain errors; please point out any you see.
Started eating the terrine in the evening. Tasted good, but I wonder if it wouldn't have been better without the venison.
Yvonne off to Olivaylle today, possibly for the last time: Chris is leaving and moving Dereel, near Ballarat. I also headed off, first to visit Patrick Forster JP to get a statutory declaration signed: after over 5 years, I can finally get my mobile phone transferred from Linuxcare to me.
Then to Hahndorf to talk to the Peters about CVS, which was fruitful. It brought home to me, however, how painful it is to work in a default environment: one screen, only 1024x768, fvwm2 in its default configuration (I don't know the mouse bindings), and a keyboard with Alt and Ctrl in the wrong place.
On the way home, dropped into the Telstra shop in Mount Barker and finally performed the transfer of contract, which was successful on the third attempt: the first time a software error asked for a field that didn't exist, the second time round it got the wrong name (mine rather than LEMIS). This software also wants a “suburb” and a street number; and it won't accept the only number I have, 969092, because it's limited to 5 digits. At least the people here know how to work around the bugs.
Also in Mount Barker, bought a power point for the new fridge that we got a couple of weeks ago. The prices were strange:
For what looked like an identical power point, I could buy one in a polythene bag for $7.19, or in a firmer packaging for $49.95. Found that strange enough to get the cashier to check: they were, in fact, both the same price. The more expensive one was also flagged as being 15 A, which it wasn't; but even then, the price would be ridiculous.
Back home, and did some debugging work. Peter is using replex (/usr/ports/multimedia/replex), which by default gets built without debugging symbols. Tried building it from the work directory with -g, and ended up with complaints about O_LARGEFILE not being defined (a bug, not a feature, but indicative of the fact that something had gone wrong in the build). It proved that building from the ports directory sets different CFLAGS from what you get if you build in the ports directory. That's got to be a port bug.
While working on this, tried out a new approach: watch is a program that enables you to share a terminal. With this entry I was able to watch what Peter Denton was entering on his screen in Hahndorf, and also type things in where appropriate:
=== root@blackbox (/dev/ttyp3) ~9 -> watch -W /dev/ttyp4
The -W flag enables writing to the terminal. That makes debugging even easier than
both of us sitting in front of one keyboard.
Received a project document, generated by Microsoft but in PDF form, this morning; it rendered fine in acroread, but printing it out produced gibberish. Spent most of the morning investigating and discovered:
Tried to scan in a page and send it back. That didn't work either: two weeks ago this scanner worked fine on two different machines. Now I get a message /dev/uscanner0: Invalid argument. That's a file system error, but ktrace didn't show it happening. The fact that the problem persists on two different machines suggests that the scanner (many years old) may have failed. But to test that I'll have to reinstall the driver on Microsoft, still more joy.
Message from Tom Maynard:
> the terrine...Tasted good, but I wonder if it wouldn't have been better > without the venison. <LOL> You didn't like it roasted, either. Perhaps you're just not a venison man.In fact, I do like venison. But what is venison? The Oxford English Dictionary states:
The flesh of an animal killed in the chase or by hunting and used as food; formerly applied to the flesh of the deer, boar, hare, rabbit, or other game animal, now almost entirely restricted to the flesh of various species of deer.
According to this definition, the meat isn't venison at all, since the animals in question are domesticated. That also has a slight effect on the flavour; more importantly, though, it's a very broad category, contrasting with domesticated meat (i.e. pork, mutton, goat or beef). Clearly there's room for significant differences in taste. In this case, the meat is rather boring in taste, and also a little tougher than necessary: it just doesn't improve the dish.
More restructuring of the CVS repository today. Thinks are looking cleaner.
After that, to a meeting of the ICT Council for South Australia. The web site's a mess, and one of the items on the agenda was updating it. The prices that have been quoted to us—as a non-profit organization, at that—blew my mind, with the result that I'm now involved in seeing how we can do it more cheaply.
Somehow didn't get very much done today. Did some more work on synchronizing our CVS repo, and also some restructuring work, but didn't reach my real goal of work on a program that can take an incoming MPEG stream and burn it to a DVD directly.
My ancient Technics amplifier (about 25 years old) has a problem: the “Tape monitor” switch has become noisy. That wouldn't make any difference, since I don't need that functionality, but it means that the right channel drops out from time to time. Today took the thing apart for a look. There's no way of getting into the switch without destroying it, but the underside looks more promising:
The cables go to a second switch (“Rec Mode”) to the right of the switch, which shows here as two rows of 4 solder connections just above the sole red wire. All I need to do is to connect a wire between the correct connections and remember never to connect in external tape inputs. Of course, I couldn't confirm this, because the switch started working again, so for the while the amplifier is upside down waiting for the problem to happen again.
Quiet day; spent half the day writing up about my brew day, and finished most of it, though I'll probably still have to do some touch-ups.
I have always liked the Malaysian curry puffs, sort of a hybrid between a Cornish pasty and a mincemeat curry, but making the individual puffs has proved to be more work than we wanted to do. Today Yvonne had the idea of a giant ravioli, filled with curry puff filling, and baked in the oven:
Achim Patzner took objection to calling this thing a ravioli. He's right, of course: first, there's only one of them, so it would be a raviolo; more importantly, ravioli are made from pasta. This is a pie.
| Previous month | Greg's home page | Today's diary entry | Next month | Greg's photos |
| $Id: diary-sep2006.php,v 1.39 2008/02/10 23:51:24 grog Exp $ | ||||