|
| Greg's diary |
| October 2011 |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
This view of the diary is limited to these topics: technology. There may be lack of continuity in the text, and some days may be completely missing. In case of doubt, please enable the complete display.
| Saturday, 1 October 2011 | Dereel → Clunes → Ascot → Dereel | Images for 1 October 2011 |
| Top of page | ||
| next day | ||
| last day |
|
We're too lame for reverse lookup
|
Topic: technology, opinion | Link here |
Chris Yeardley is flying to Queensland with Tiger Airways, an el-cheapo subsidiary of Singapore's el-cheapo airline of the same name. It's so el-cheapo that the Civil Aviation Safety Authority suspended their license for 5 weeks earlier this year due to safety concerns. Presumably they're now satisfied of their safety, but there seem to be other issues. It seems that you need to get your ticket by email and print it out yourself (or, presumably, find another method that costs more). Chris tried that, but the ticket didn't come through:
It used to be common that companies were too lame to have reverse DNS, but it's been a long time since I've seen that problem. Given that the company encourages email contact, they could at least play by the rules.
| Sunday, 2 October 2011 | Dereel | Images for 2 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Microsoft making life difficult
|
Topic: technology, opinion | Link here |
As I've noted on many occasions, I don't have a completely functional laptop. I must have about 7 of them, but the most recent is 6 years old, and only two seem marginally functional: the Dell Inspiron 5100 which seems to be shutting down more and more frequently, and the Inspiron 1150 with the dead USB bus. Currently only the 5100 has a functional disk. But for things like visiting the Friends of the Ballarat Botanical Gardens I don't need USB: I can just move the disk from the 5100 to the 1150.
Tried that, and FreeBSD came up with almost no problems. Just not a mouse. I've seen that before: there's some flag to be set, and I knew that I had mentioned it here somewhere. But Google seemed a better choice, and led me to this article, which had the advantage of referring to identically the same laptop. Just add
After that, things worked as if nothing had changed. Conveniently, the article also had a date on it, and that gave me a pointer to the diary article on the subject.
But I normally run Microsoft on this box (thus the system name pain). Tried that too. The system spontaneously reset, even in “safe mode”. This machine has only 384 MB of memory, compared to 512 MB on the other machine. I wonder if the system is expecting to find all the memory, or whether it just doesn't want to work on even marginally different hardware. What a pain this Microsoft stuff is! And how aptly named the system!
| Monday, 3 October 2011 | Dereel | Images for 3 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More Hugin building
|
Topic: photography, technology | Link here |
I've heard from Carlos Eduardo G. Carvalho, better known as Cartola, who's very active on the panorama scene. He's also yet another FreeBSD user in the Hugin community—all the more surprising that the latest version doesn't build under FreeBSD. Spent some time looking at the instructions that I had put up on the Panotools wiki. They're completely wrong. And despite the volume of this diary, I didn't describe the method in sufficient detail when I did it. Spent some time working on the description, but I'm still not done, especially since they've released a new stable version since then, one that is missing code necessary for the FreeBSD version.
| Tuesday, 4 October 2011 | Dereel | Images for 4 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
To the Friends again
|
Topic: gardening, technology | Link here |
The Friends of the Ballarat Botanical Gardens have a problem: their printer isn't working. And I have taken it upon myself to look after their computer stuff, though it's all old Microsoft-space stuff. Spent the morning preparing: collecting seedlings for Hebes and Betula pendula. Into town, got rid of my seedlings, and to look at the printer:
Presumably you don't need to be a computer expert to see that the thing is out of ink. And for that I came into town. Went looking for Mike Sorrell, but was told that he had left round midday. Spent some time looking for plants, and while discussing his disappearance with Yvonne, he came back. Spent some time looking at the equipment and discussing web pages with him: he wants to create his own content, which makes sense. I'm trying to make it as easy as possible to write web content, but I can't see my way past using an editor, and that seems to be a no-no in the Microsoft space. In any case, we weren't able to complete the discussion: the doors shut at 16:00, and we hadn't got much done by then.
It wasn't a complete waste of time, though: the Gardens have apparently just completed an exhibition of Cyclamens, and they had dozens of plants to give away, still mainly in bloom. Took 7 pots with me, all but one still blooming better than I have ever managed myself:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The one that wasn't blooming so well is interesting because it has a pure white flower.
|
Panasonic: we don't
do digital cameras
|
Topic: photography, technology, opinion | Link here |
Finding a definitive link to the DMC-FT1 proved almost impossible. Even Google couldn't help on the Panasonic web site. Finally ended up at a page that tells me:
|
|
||||||||
That's with no restriction on the search. What a broken site! And of course their lists of digital cameras show only current models. Finally tried their support pages, but even there I couldn't find it. Maybe this is a perspective-dependent model number, like Canon's: it differs depending on where you look at it from. Certainly it's no advertisement for Panasonic.
Of course, part of the issue is the non-deterministic nature of Google's search engine. After writing this, Jashank Jeremy did almost the same Google search that I had tried, only on the Australian site and came up with a link on the Australian web site.
| Wednesday, 5 October 2011 | Dereel | Images for 5 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Another net outage
|
Topic: technology | Link here |
I've had pretty good network connectivity over the last few months, but today I ran into more trouble: the modem went into UMTS mode and stayed there even when I was trying to download large quantities of data. We've seen that before. It seems to need physical removal and replacement of the modem in the USB connector. Did that, didn't get connected until the third attempt, and it dropped the connection immediately. Further investigation showed that the antenna connector had become dislodged. What horrible flimsy things these USB modems are!
After that, it ran for a few hours, then disconnected for 10 minutes (while I was outside, so I couldn't check what was going on). And later again things went to hell:
That shows me barely able to contact the next hop, and packet loss was round 50%. Called up Internode Support and was connected to Tristan (with whom I have spoken before) almost immediately—a far cry from my recent experience with TransACT. He wasn't able to do much: things cleared up almost immediately. He did note the incident and explained how they allocate dynamic IP addresses: they try to give the same address again, assuming I get connected to the same server. But there seem to be two servers they could connect to in Melbourne, and today I got connected to the other one, so clearly I didn't get the same IP address.
|
Firefox: starting profiles again from scratch
|
Topic: technology | Link here |
My firefox configuration is the result of years of bumbling, made easier by the lack of documentation and clear migration paths. Lately I've noted that the size of headings has changed, for no apparent reason. And then Jashank Jeremy told me about the “awesome” bar on the new firefox, something that I don't have.
Do I want it? The name sounds so stupid that it's reason enough not to use it. But it occurred to me that maybe I'm missing out on something useful too. So how about starting reconfiguring it from scratch?
Did that, with a new profile called Cold-turkey, and in deviation from firefox's obfuscatory directory names (can you remember 7v0n6ir5.Default User, with a space in it?), called the directory Cold-turkey too. Stopped the instance of firefox, restarted, and... no Cold-turkey profile. Yes, the directory was there, but not the profile. Further investigation showed that the file ~/.mozilla/firefox/profiles.ini was set to be writable only by root. No idea why, but I suppose it's to be expected that firefox would ignore such a problem when trying to write the profile information.
So I set permissions correctly, updated profiles.ini and stopped the instance of firefox. Started again:
|
|
||||||||
The message is Just Plain Wrong, at least in this case: for some reason, it had removed the directory. This might have been a race condition with stopping the instance of firefox, which rewrites profiles.ini. Did it also check the old version and remove directories that were “no longer needed”? Hard to say, but it fits in with the “there can only be one” mentality.
Tried again, and got the same message:
|
|
||||||||
This time it was right. It seems that the prior attempt actually created the directory. OK, that's what I wanted, so tried to select it:
|
|
|
||||||||
I've seen this before. But this time it has a different meaning. How do I get past it? Somebody gave me this URL, but by that time I was so fed up with this mess that I went out into the garden to do something useful.
| Thursday, 6 October 2011 | Dereel | Images for 6 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Goodbye Steve
|
Topic: technology, opinion | Link here |
It seems that everybody in the world is commenting on the death of Steve Jobs. Yes, it's sad, to me especially since he was considerably younger than I am. But some of the accolades I've seen tend to confirm the opinions I voiced a couple of months ago: the computer industry is no longer technology-driven but market-driven. Steve Jobs managed to market things that others had failed to do. He made a cult out of mobile phones, something that was beginning to stagnate. He brought out business models with things like iTunes, really a front end for the iTunes Store, something that upsets both free software advocates and people like me who just want a program that doesn't tell you how to live your life. But he didn't manage to bring out an operating system that looks the same from the (overly gaudy) GUI and under the covers: Mac OS X really lives a double life. But who cares? It sells.
In short, the Steve who turned Apple around and will go down in history as one of the great innovators of the computer age was the businessman Steve Jobs, not the tech wizard Steve Wozniak. The days of real technological innovation are gone.
| Friday, 7 October 2011 | Dereel | Images for 7 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Compressed air cleaning
|
Topic: technology, photography | Link here |
We've established that Yvonne's camera problems were due to dust, and I suspected that the old camera had the same problem. I'm also wondering if the spontaneous shutdowns on pain, my Dell Inspiron 5100 laptop, were not due to overheating because of dust buildup. To the garage with both camera and laptop and tried blowing compressed air through them. Nothing very exciting happened. The camera works most of the time, but it did give up on focusing on one occasion and had to be power cycled—which presumably could dislodge dust—and the amount of dust that came out of the laptop doesn't suggest that it was a problem. We can only be sure it has made a difference if it doesn't shut down after being powered on for days.
|
More panorama software
|
Topic: photography, technology, opinion | Link here |
I've been in contact with Cartola (Carlos Carvalho), who is very active in panorama photography, especially 360° panoramas. He was (probably rightly) surprised that I didn't have any web-based software for interactive display of my panoramas, and pointed me to a couple of programs that help: Salado Player and Panini. While searching, came across another, krpano. Decided to take a look today. None are in the FreeBSD Ports Collection, so the first step was to build them.
krpano proved to be a dead loss: it's commercial software. That in itself wouldn't be reason for rejection, but it's not available for FreeBSD. There's a Linux version, but that seems to be limited to “command line tools”, whatever that may imply, and if I don't have the source, I don't want to even try.
Next was Panini. It seems to be part of the pqvt project, and like so many projects on SourceForge, it doesn't seem to have any documentation. About all I could find was a file panini-build.txt in the top-level directory, written one line per paragraph and terminated with \r\n. I had to reformat it just to read it. When I did, it wasn't very helpful. It uses Qt, in particular qmake to build the project. According to panini-build.txt:
In the panini directory, type "qmake panini.pro"
I suppose “panini directory” means the top-level directory. At least there's a panini.pro there. OK, did that:
=== grog@dereel (/dev/pts/21) /usr/ports/graphics/panini/work/panini-0.71.104 24 -> qmake panini.pro
What's that? I really don't want to have to learn [a-z]make to build projects. It's bad enough that we have BSD make and GNU make, but things like qmake and cmake just don't seem to be worth the trouble. Did a bit of looking around:
=== root@dereel (/dev/pts/13) /usr/ports/graphics/panini/work/panini-0.71.104 66 -> man qmake
I suppose that's to be expected. There is web-based documentation, installed on my machine as http://wwww.lemis.com/portsdoc/qt/html/qmake-manual.html. But it's all too much work.
Moved on to Salado player. I can't complain about the volume of the documentation here, though it's clearly written by a non-native English speaker. Judging by the lack of articles, I suspect an Eastern European origin. But the “Quick Start” needs a lot of digestion, and somehow I don't have the patience for that any more. Put it off until some other time. I suppose my real concern is that I'll invest a lot of time setting the thing up, only to discover that there's some showstopper which makes it all a waste of time.
| Saturday, 8 October 2011 | Dereel | Images for 8 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More panorama experiments
|
Topic: photography, technology, opinion | Link here |
More investigation of Qt's qmake today. The manual states:
The following is a list of environment variables available to choose from when setting QMAKESPEC:
aix-64 hpux-cc irix-032 netbsd-g++ solaris-cc unixware7-g++ aix-g++ hpux-g++ linux-cxx openbsd-g++ solaris-g++ win32-borland aix-xlc hpux-n64 linux-g++ openunix-cc sunos-g++ win32-g++ bsdi-g++ hpux-o64 linux-icc qnx-g++ tru64-cxx win32-msvc dgux-g++ hurd-g++ linux-kcc reliant-64 tru64-g++ win32-watc freebsd-g++ irix-64 macx-pbuilder reliant-cds ultrix-g++ win32-visa hpux-acc irix-g++ macx-g++ sco-g++ unixware-g hpux-acc irix-n32 solaris-64 unixware7-ccThe environment variable should be set to qws/envvar where envvar is one of the following:
linux-arm-g++ linux-generic-g++ linux-mips-g++ linux-x86-g++ linux-freebsd-g++ linux-ipaq-g++ linux-solaris-g++ qnx-rtp-g++
It doesn't say so, but it looks as if for FreeBSD I need to set QMAKESPEC to freebsd-g++. Or do I? Maybe it's linux-freebsd-g++? Yes, looking at the names, freebsd-g++ makes more sense. But what's the other one for? Why is there no explanation of the two sets of values? And, in particular, why is there no default? This way every user who wants to use qmake needs to set QMAKESPEC—to the same value almost every time. What a mess!
Tried that and found:
=== grog@dereel (/dev/pts/21) /usr/ports/graphics/panini/work/Panini-0.71.104-src 42 -> export QMAKESPEC=freebsd-g++
=== grog@dereel (/dev/pts/21) /usr/ports/graphics/panini/work/Panini-0.71.104-src 43 -> qmake panini.pro
Is that an error or “normal”? I didn't have time to continue. Today was photo day, and Cartola's discussion of 360° panoramas had me thinking. There's very little missing on my verandah panorama to make a 360° panorama, so set to taking a zenith shot. That didn't work as well as I thought: in the process I made the second row too high, so there was no join, and I had to make a third row in the middle, a total of 111 images of 37 different viewpoints. On stitching them together, it became apparent that I didn't need all those images: I could have done them as before with just the additional zenith image. The result also showed how imperative it is to put the camera directly below the beam in the middle, which here looks as if it is bent in two to the right:
|
|||||||||||||||||||||||
And the nadir? No idea. What I have at the moment can be summed up by the three panosphere images. The first is horizontal, roughly corresponding to the image above. The second is looking down from above, and the third is looking up from below, with a hole in the base through which the other side is visible:
|
|
||||||||
|
|
||||||||
|
|
||||||||
The problem, of course, is that the tripod is there. How do I get around that? I suppose I could try removing the camera from the tripod and shooting freehand. In some cases, that might work, but with the decking it seems unlikely to work here. But I won't know until I try it—next week.
|
Another net outage
|
Topic: technology | Link here |
Once again I've been disconnected from the net. It's standard procedure now to pull the modem out of the USB slot (more carefully than last time) and replace it. And today that worked. How I hate flaky USB devices!
|
Backup causes system crash
|
Topic: technology, photography | Link here |
After processing my photos, backed up to the USB disk. More problems. After some time of backing up, got a lot of these:
Yes, those enormous offsets are valid: they're just shy of the 2 TB size of the disk. But that proved fatal. lagoon crashed completely. fsck worked—it thinks. I really need to get my eSATA interface working. How I hate flaky USB devices!
| Sunday, 9 October 2011 | Dereel | Images for 9 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Hugin on FreeBSD amd64
|
Topic: technology, photography | Link here |
Compiling hugin on FreeBSD is currently a minefield. The latest version has removed the code for one of the dependencies, tclap. While that's almost certainly the correct thing to do, FreeBSD currently doesn't have a tclap port, and I don't particularly feel like making one. So I had to replace the bits in the build tree. The problem there is that a make clean removes it all again, so it's very fragile.
In addition, Cartola had asked for a binary for amd64, which I'm still not running on dereel, so brought out the old teevee, the one with the Ethernet interface damaged by a power surge three months ago. Put in the disk for defake (the amd64 upgrade machine) and an Ethernet card, but out of habit plugged the cable into the on-board interface, and—it worked! It's amazing how often “damaged” equipment can recover after a few months. It's also irritating that I now won't be able to prove to Powercor that the thing was damaged at all.
It turned out that one of the dependencies, libpano13, needed to be updated, which mercifully didn't take too long. And after a lot of waiting—why does C++ compile so slowly?—finally got it built. This should be easier.
| Monday, 10 October 2011 | Dereel | Images for 10 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Testing hugin
|
Topic: technology, photography | Link here |
I finished my amd64 build of the latest version of hugin yesterday, and was going to send it to Cartola when I remembered the old joke “It builds! Ship it!”. Guilty as charged. So I tried it out:
=== grog@defake (/dev/pts/1) ~ 1 -> hugin
Not exactly the success story I was hoping for. But this was a remote display, and hugin uses lots of libraries in which I have dubious confidence. So I tried it on defake:0, and it started up normally. But I couldn't do anything with it: the Ethernet interface on this motherboard may be functional again, but the USB system isn't, and I would have to reboot to get a PS/2 mouse recognized. Not a problem in itself, since this is a test machine, but I also wanted to bring the system up to date. So I spent the rest of the day (asynchronously) doing that. Another day's delay in the delivery of hugin. It's a good thing I'm not being paid for it.
|
New GPS navigator
|
Topic: general, technology | Link here |
Yvonne returned from walking the dog with a small electronic box that she had found in the forest. It proved to be a Navman GPS N196, a model so old that Navman no longer want to know anything about it, not even for map updates (in its turn a good reason to avoid Navman). But it works, and I was able to find a “Home” POI in the maps. That's in Ballarat, so we'll drive past next time we're in town and see whether it's really theirs. Otherwise Yvonne needs a GPS receiver, so this could do the trick.
| Tuesday, 11 October 2011 | Dereel | Images for 11 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
St. Ignucius vs. St. Steven
|
Topic: technology, opinion | Link here |
I'm truly amazed by the number people who take the death of Steve Jobs as a personal loss. On Facebook many people have replaced their photo images with a sad Apple face, and acccolades continue to pour in. But many of these are the same people who complained about Apple's predatory behaviour, both against its customers and against its competitors. So many people complained about the locked-down nature of Apples smaller devices. Now rms pitches in, in his typically tactful way, writes (quoting Harold Washington) “I'm not glad he's dead, but I'm glad he's gone”. Predictably, many Apple fans are up in arms, and a number have quoted this article with the truncated URL by Joe Brockmeier, titled “Why FSF Founder Richard Stallman is Wrong on Steve Jobs”.
Why? He doesn't say. Instead, you can read:
Though I've often disagreed with the tone and language of Stallman's commentary on closed devices, he makes good points about software freedom.
...
While I'd love it if Stallman would retire, or at the very least improve his social skills, I hope he lives to be 120. As long as he's alive, there's hope he might change. I'd never be glad that he's gone.
Those are my excerpts, of course, but I can't see anything else in the article that accuses him of more than bad taste. Strangely, none of the people agreeing with this article seem to have noticed this. The title has little to do with the article.
Elsewhere, of course, The Register has commented, also effectively agreeing with rms' criticism of Apple's attitude to software freedom. And even the Los Angeles Times publishes:
Yet Stallman's critique of Jobs' business model has merit. For all Jobs' focus on user-friendly devices, Apple's buttoned-down approach to its software and apps, along with the way its mobile devices facilitate violations of their users' privacy, should be the subject of much broader concern. Stallman's eulogy may get wide distribution because of its tone, but his underlying point about the digital world deserves to be heeded.
So why are people so up in arms about the statement? I really don't know. Clearly nobody will ever convince rms to shut up or to be tactful. And I'm not sure I want him to. As long as there are people like him and Theo de Raadt way out there on the fringes, people like me can feel normal.
|
Resource problems
|
Topic: technology | Link here |
While watching TV, tried to look up something on Google, as you do. But this time I got an error message telling me that it wasn't available. Tried my (local) home page, and was told that it didn't exist. Panic time?
Into the office, where I found lots of messages like this:
The processes gave the lie to what had happened: I only use pkbzip2 for backups. And today I had tried to repeat a backup (level 1 dump) that I should have done on Sunday, but which I had forgotten. I do a daily backup to an external USB disk: a level 0 dump on the first day of each month, level 1 on Sundays and level 2 on other days. I have a script for this which does other cleanup functions, mounts the dump disk, calls the dump script and umounts the disk. This time I called the dump script directly without first mounting the disk, so it dumped to the root file system.
Problem identified. Remove the /backups/* hierarchy, mount /backups and repeat. Simple. And ineffective.
Further investigation showed other messages:
That's the maximum number of files that can be open in the system. On dereel it normally looks like this, with the current number of open files for comparison:
=== root@dereel (/dev/pts/12) /usr/local/etc/rc.d 25 -> sysctl kern.maxfiles kern.openfiles
And yes, indeed, kern.openfiles was at the maximum. But what was doing it? The kernel's not so good at reporting that. The obvious idea is that it had something to do with the full file system. Spent a bit of time looking with lsof, and finally found the culprit: yreport, part of my weather reporting software. Killing that process and restarting it solved the problem. Is this the cause of the problems I reported a couple of months ago? In any case, it's clear that it had a file descriptor leak (in particular, if a write fails), and it's fixed.
| Wednesday, 12 October 2011 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Outsmarting web crawlers
|
Topic: technology | Link here |
Stephen Rothwell and more recently Martin Schwenke have been complaining about the number of hits I get on our communal web server. One of the big issues are the web crawlers that index this diary in particular. The images are the issue: each image contains a link to a different sized version of the same image, which might look like http://www.lemis.com/grog/diary-oct2011.php?imagesizes=11111111111112#Photo-13. That indicates that it's the 14th photo on this month's diary, that it should be size “2” (“small”), and that all before it should be size “1” (“tiny”). If I follow it, I get a new link to http://www.lemis.com/grog/diary-oct2011.php?imagesizes=11111111111113#Photo-13, and so on.
That's fine for humans, but web crawlers follow these links, as Martin finally managed to make clear to me. And if they do their job exhaustively (they don't), things can really go crazy. Last month's diary has 164 photos in it. Each has 4 sizes, so theoretically there are 4¹⁶⁴, or 5.5E98 combinations of image size.
The result has been hit rates of several per second from the crawlers, all looking for the same page with different image sizes:
What should I do about it? I want the crawlers to index my pages, of course, and I also want the higher-resolution images to be indexed. The first step is obviously to recognize the crawlers, which currently I'm doing on an ad-hoc basis:
And then? I need to stop the crawlers thinking each of these pages is a separate, indexable page. So that requires a redirect. That could be as simple as http://www.lemis.com/grog/diary-oct2011.php?size=4, which returns all images at maximum size, but then the search results would point there too, and that would not be what people want to see. The search result must be http://www.lemis.com/grog/diary-oct2011.php, but the crawlers still should get the large images.
So: I need to lie to the crawlers. If a crawler asks for a page and specifies sizes, it gets redirected to the same URL without a size specification. And in that case, it always gets size “4”. The cached page will reflect that, but the link will return thumbnails.
And the results? Lots of 301 (moved permanently) returns, here in the 7th field of the log file:
Martin pointed me at apachetop, which monitors real-time performance of the web server. It's interesting to note the third line, which breaks up the return codes. 17.4% are 2xx, 82.5% are 3xx (mainly 301, moved permanently).
It'll be interesting to see how long it takes for the redirect ratio to drop. I note that the current version of http://www.lemis.com/grog/diary.php on Google is about a week out of date, so it might take at least that long.
|
More network problems
|
Topic: technology | Link here |
While working on the HTTP redirect stuff, noticed a significant drop in network performance: the modem was stuck in UMTS mode again. Again popping the modem “fixed” it, but it's getting irritating.
| Thursday, 13 October 2011 | Dereel | Images for 13 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Goodbye dmr—a real giant of the computer world
|
Topic: technology, opinion | Link here |
I've made a number of comments about the recent death of Steve Jobs, and expressed my surprise about the personal sadness that many felt. Also the reactions to rms' comments, which my wife Yvonne found so good that she sent him a personal congratulation. But what got me the most was how people claimed that Steve was a technical innovator, a “giant”.
And then today I heard the sad news that Dennis Ritchie died on Sunday. And it has taken the world this long to find out. Rob Pike seems to be the first to have reported it, less than two hours ago if I interpret Google's time-zone-less time specifications correctly.
What a comparison! If ever there was a “giant”, it was Dennis. I can't think of any part of modern computing environment which he hasn't influenced. Most modern compiled programming languages are either his own C programming language, or they're derived from it. His file system design is alive and well today and forms the basis of almost all modern file systems with the exception of Microsoft. Everybody who uses a web browser (in other words, everybody) sees the file name conventions that he developed 40 years ago. And as one of the two main developers of Unix, he has influenced all operating systems, even to some extent Microsoft. There's not a single product that Steve Jobs marketed in the last 20 years that doesn't depend on Dennis' work.
Now there's a giant. Isn't it sad that his passing has gone almost unnoticed? If ever I needed proof of my claim that the computer world is now market-driven and no longer technology-driven, the reactions to the deaths of Steve Jobs and Dennis Ritchie demonstrate it. Atypically, I feel a deep sense of loss.
|
dmr interview
|
Topic: technology | Link here |
Call from Susan Atkinson of ABC 702 Sydney radio, asking if I would do an interview about the death of dmr on this evening's “Evenings” program with Robbie Buck. It seems that she found me by quite a roundabout way: for some reason they rang UniSA and were connected with Ben Close, who put them on to me. Nothing to do with the messages I sent to various newspapers.
“Evenings” runs from 19:00 to 22:00, and my part was at 21:30, but at the last minute—apparently after announcing the interview—they postponed. As far as I can tell, some important sports news had cropped up, and now they're planning it for tomorrow between 20:00 and 21:00.
|
Death of teevee
|
Topic: multimedia, technology | Link here |
I wasn't kept idle during that time. Started watching TV, and then found something extraordinary:
|
|||||||||||||||||||||||
A SIGSEGV out of ls looks highly suspicious. At the same time, I got repeated reports on the (remote) system log:
Shortly after that, the system froze and I wasn't able to reboot. Took it in to the office and ran fsck on a different machine, where it ran. But I tried to shut down before the background fscks were finished, with the longest buffer sync output I have ever seen:
|
|||||||||||||||||||||||
It looks as if shutdown doesn't stop background fsck.
Got it running again, and we watched TV for about 15 minutes before the system died again. Despite the log messages, everything is pointing to something on the motherboard—and all the components are only two months old. Looks like another trip to Geelong tomorrow.
| Friday, 14 October 2011 | Dereel → Geelong → Dereel | Images for 14 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
To Geelong for hardware
|
Topic: technology, general | Link here |
Did some further tests on teevee this morning, and got it to a point where it wouldn't get through the POST with nothing connected to the motherboard. There was an extraordinary amount of dust in the heat sink, so blew that out with compressed air—finally an impressive cloud of dust—but how did it get in there in only two months? Checked the memory and the display card in Yvonne's computer, and they were both OK. So: processor, motherboard or power supply. Dragged out the receipt for the hardware, was about to put the machine with assembled motherboard into the car when it dawned on me. This wasn't the machine I bought in August: a week later I gave it to Yvonne and used her old motherboard for teevee. So not much hope of getting a replacement, though I should check when I bought the other one.
|
MSY: strangling their outlets?
|
Topic: technology, opinion | Link here |
At MSY in Geelong, discovered a very poor selection of hardware. In the end bought a AsRock N68VS3 motherboard, a Sempron LE-145, exactly the same components that I bought two months ago. Also Yet Another UPS, and a power supply for $48, more than either the processor or the motherboard. Why so expensive? That's all that the branch can get. According to the pricelist, they have at least 12 cheaper suitable power supplies, starting at $17. But the Geelong branch can't get them. And even if customers order in advance, they don't get delivered.
I sensed a bit of frustration from the salesman. MSY makes their money by being cheaper. But if they're not, why should I go there? I had really wanted a dual core CPU, and they had a couple on the lists, but only the most expensive was available—exactly as I recall it being the last time. I had already noticed that their disks were more expensive than at OfficeWorks. Why do MSY do this? Poor organization? Are they planning to shut the Geelong branch? Certainly counterproductive, in any case.
|
New teevee
|
Topic: technology | Link here |
Back home, put together the new machine. This time the processor is really a Sempron LE-145 and not the Athlon II that I got last time. You couldn't tell from the packaging. Here the chip of two months ago, then today's:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Brought the machine up, ran fsck and—it hung in exactly the same way as it did last night. Further investigation showed that it was the old soft updates bug with background fsck. Running fsck in the foreground fixed it. All that work for nothing!
Well, no, probably not. That didn't explain the SIGSEGV last night, nor the fact that the machine didn't get through the POST this morning. It looks very much as if there were two problems: the failing motherboard and the failing fsck. In any case, the machine ran, though for some reason the NIC is only running at 100 Mb/s. To be investigated Some Other Time.
|
Dennis Ritchie's legacy
|
Topic: technology, opinion | Link here |
The accolades for dmr are pouring in, as well they should. Today I had my moment of glory with the postponed interview on ABC 702 Sydney radio, and of course they asked me about Dennis' importance. How do you answer that, especially to an audience of people with no particular computer background? Yes, I mentioned Unix, I mentioned C, like everybody does.
But for me, the outstanding legacy of Unix is the file system, or at least the underlying concepts. It's such a masterpiece of simplicity, clarity and power. And nobody really seems to appreciate it. Certainly those who came after didn't understand the concept of a directory, or they would never have come up with this silly word folder. Who wrote it? I had always thought it was Dennis, but not necessarily at the top. But looking at the sources of the Fourth Edition, it seems more likely to have been Ken. And at the end of the interview, to bring things home to the audience, I said “Every time you see a slash in a URL, think of Dennis”. Not as appropriate as I thought, but I still think it's a nice sentiment.
Apart from such errors on my part, though: who were the really great pioneers of computing? I had always thought of Ken and Dennis being up on the list, but who's more important? John von Neumann? He pretty much defined computer structure, and it hasn't changed that much until today. Alan Turing? He was a great thinker, but it's not clear to me what legacy he has left behind.
Who else? There are plenty of people who have been pivotal in computing, but Ken and Dennis had not only ability but also the luck to be present at a pivotal time in the evolution of computing, and to be allowed to do things right rather than do things by the end of the quarter. I can't think of anybody else who has had so much influence. It's nice to see that gradually modern systems are gradually discovering facets of their work that had eluded them before. To quote Henry Spencer:
Those who don't understand UNIX are condemned to reinvent it, poorly.
| Saturday, 15 October 2011 | Dereel | Images for 15 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Ironing out the wrinkles in teevee
|
Topic: multimedia, technology | Link here |
So teevee is working well enough with the new hardware, it seems. But I still don't have the gigabit Ethernet running. Did a little investigation, and it seems that FreeBSD doesn't believe that the chip set can do gigabit Ethernet:
According to the motherboard instructions, though, it should do gigabit:
|
||||||||||||||||||||||
It wasn't until I checked the specs, though, that the truth came out: Asrock make two very similar motherboards, the N68-VGS3 and the N68-VS3. The instruction manual applies to both, but I have the N68-VS3, and the main difference between the two boards is that only the N68-VGS3 does gigabit Ethernet. OK, the motherboard only cost $45, but I'm still surprised that the gigabit technology makes any significant price differential.
| Sunday, 16 October 2011 | Dereel | Images for 16 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
“Open source” class assignment
|
Topic: technology | Link here |
Chris Yeardley over this afternoon to talk about installing FreeBSD on her new laptop. She has a class assignment to contribute to some “open source” project, and of course I encouraged her to do it with FreeBSD.
Chris isn't ready yet to take the plunge and install the ultimate anti-virus, FreeBSD alone on the disk. So how do you install FreeBSD as one boot option on a laptop? Well, I wrote the book, but that was years ago, and in general I don't do dual boot. It seems that Microsoft now allows partitions to be resized, but I don't know the details.
It proved that she had four MBR partitions on the disk already. Microsoft reports:
What does that all mean? Two partitions are “primary”, which is clear enough. D: is a “logical drive”. But then, C: is too: it isn't a real drive. Is this Microsoft's obfuscation for “logical partition”? Probably; at any rate, it was a logical partition (in an extended partition). That's a problem, because it's half the disk. Sure, I could remove it and replace it with a primary partition, but that's all that's left on the disk, and Chris doesn't really need 400 GB for FreeBSD.
Did some thinking and decided, at least for the moment, that Chris could connect an external USB drive and use that as the FreeBSD partition. Played around with that, enough to discover that the laptop has two different kinds of USB connectors: the ones on the right didn't seem to work, but the ones on the left did. Later noted that the ones on the right have blue markings on the socket, which indicates USB 3.0. But I thought they were compatible. I wonder if this is an issue with this particular laptop or its design. Or maybe it's just the bug that I swear crawled out of the laptop (on the right in the first photo):
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
More teevee pain?
|
Topic: multimedia, technology | Link here |
While watching TV in the evening, teevee froze again. I've replaced everything except the memory, which I had tested separately. Is it really the memory? Or something on the disk? The latter seems unlikely, since I wasn't using it. To be observed.
| Monday, 17 October 2011 | Dereel | Images for 17 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Nickel-Zinc batteries: first impressions
|
Topic: general, photography, technology | Link here |
Finally my Nickel-Zinc batteries have arrived. I bought them nearly 3 weeks ago, and received the confirmation:
Why so long? The envelope explains:
|
|||||||||||||||||||||||
That's a delay of nearly 2 weeks.
And the voltage? As delivered, they had a voltage of 1.73 V, significantly higher than the nominal 1.6 V or 1.65 V, but interestingly identical with the value in Wikipedia. But first I had to charge them. The charger came with instructions of above-average length and typical accuracy:
|
||||||||||||||||||||||
On the left “Safety features: Will not charge Non[sic]-Nickel Zinc rechargeable batteries”. In the middle: “ATTEMPTING TO CHARGE OTHER TYPES OF BATTERIES MAY CAUSE PERSONAL INJURY AND DAMAGE TO THE CHARGER” (no mention of the presumed destruction of the batteries).
Charging was made more difficult by the fact that the charger came with an American-style plug which didn't quite seem to adhere to the standards. I have lots of chargers like that, along with lots of adapters, but this one didn't fit most of them. Found one that sort of fitted, but the charge light didn't go on. Searched further and finally found one that had a snug fit. Still no charge light. I was about to give it up as a defective charger when I found a strip of plastic between the batteries (which were supplied in the charger) and the positive side of the charger. Now why didn't they document that? After removing it, they charged properly. And the voltages? It seems that all batteries have a nominal voltage that is really only nominal. Here NiMH (nominally 1.2 V), normal alkaline batteries (nominally 1.5 V) and NiZn (nominally 1.6 V or 1.65 V, depending on whom you ask):
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Can that be right? This is a cheap multimeter, and it would be reasonable to question its accuracy, though my suspicion is that the chips in modern multimeters are accurate enough. On a suggestion of Peter Jeremy, located a Mercury battery (in my Pentax Spotmatic). They have very stable voltages, but what? Most sources say 1.35 V, which seems too round a figure to be acceptable. In one place I read 1.355 V, which at least has more decimal places. My multimeter showed 1.360 V, less than 0.4% higher than the latter value, so I assume that the other voltages are also correct.
Is that too much? I put some in my power-hungry Nikon “CoolPix” L1, and it seems happy with them. The voltage of the NiZn battery in the example is shortly after charging. I put it in my old Garmin GPS receiver (bought in 1997) and let it search the sky for satellites. It took half an hour: the internal clock was off by 40 minutes, and there's no way to reset it. This device has a particularly heavy battery consumption—it uses a set of alkaline batteries in about 3 hours—so the 30 minutes should have shown a significant consumption. But at the end of that time, the voltage was still 1.809 V, a far cry from the nominal voltages. Of course, looking at the voltages of a normal new alkaline cell, it's clear that all devices should be able to accept a voltage of 1.6 V, but this is higher still. I'm still not game to put them in my Mecablitz 58 AF-1. I should find a couple of other things to try first.
|
Still more teevee pain
|
Topic: multimedia, technology | Link here |
In the course of the morning, discovered I couldn't access teevee—again! Brought it into the office and recovered things and watched it for a while. No particular problems. But I've changed everything except the memory, so swapped that too with Yvonne's computer. We'll keep an eye on it, but so far there have been no further crashes.
There's another irritating thing with teevee: since the new motherboard, I no longer get remote syslog messages. They're very useful if something crashes, and the only way I saw the disk errors last week. Why did they stop? Did a lot of playing around and discovered that it would work if I restarted syslogd (newsyslog wasn't enough). Somehow it must be related to the Ethernet interface, but I don't see how.
Then I discovered that cvr2 was down too! To the cupboard where I keep it, and discovered it happily running—only the network switch was powered off. It seems that that happened while I was playing around trying to charge the Nickel-Zinc batteries. teevee is on that switch too, so probably this time it didn't crash; but it did yesterday, so that's not overly relevant.
|
Backup pain
|
Topic: technology | Link here |
As if that wasn't enough, my weekly backup yesterday failed:
=== root@dereel (/dev/pts/14) /var/tmp 39 -> mount /backups
The device node was there, of course, which in FreeBSD (with devfs) means that it has been detected. And I was able to look at it, but it still claimed “not configured”:
=== root@dereel (/dev/pts/14) /var/tmp 41 -> bsdlabel da0s1
=== root@dereel (/dev/pts/14) /var/tmp 42 -> mount /backups
Tried running fsck:
=== root@dereel (/dev/pts/14) /var/tmp 49 -> fsck -y /backups
Why is it still dirty? It didn't find any errors. But I re-ran fsck, with the same results. Finally I discovered that the file system was already mounted:
=== root@dereel (/dev/pts/21) /usr/src/sys 11 -> df
But why an ENXIO? If the file system is already mounted, I should get EBUSY (“device busy”). And fsck should have at least said NO WRITE at the beginning. And none of that explains why the file system was mounted in the first place. The fact that fsck didn't find any errors at all suggests that it had been umounted cleanly.
| Thursday, 20 October 2011 | Dereel | Images for 20 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Playing DVDs
|
Topic: multimedia, technology, opinion | Link here |
Once upon a time TV was easy. You turned on a TV, found a channel and watched it. Then video tapes came, and you could record things, so you were independent of the broadcast time. Then pre-recorded tapes came, and with them licensing restrictions, notably the “don't copy” restriction which I personally find stupid. I can understand that the license holder wants to sell as many copies as he can (though nowadays they don't necessarily try), but that's only indirectly related to copying.
Nowadays it's DVDs and Blu-ray with all their stupid copy protection, which really only upsets the innocent, while the pirates know enough to circumvent the issues. I generally stick to the license agreements, but it's not always easy. Last week I borrowed some DVDs from the Geelong regional library and tried to play them (on my computer). No go. It didn't play in a normal DVD player either. Looking at the DVDs, it's understandable why:
|
|||||||||||||||||||||||
That's the worst of them, of course, but it looks like some previous borrower has given them to their very young children to play with. And yes, there's no hope for this DVD. But others are recoverable. I've found that the FreeBSD program recoverdisk does a reasonable job of recovering a disk image, if you first try to play it with mplayer to unlock the CSS keys. After that, mplayer can play the DVD image.
What are the ethics of this? The tiny print on the back of the DVD package (black on dark green) states:
WARNING: The copyright proprietor has licensed the film (including he soundtrack) comprised in this digital video disc for home use only. All other rights are reserved. Any unauthorised copying, editing, exhibition, renting, exchanging, hiring, lending, public performance, diffusion, and/or broadcasts[sic] of this digital video disk or any part thereof is strictly prohibited. This DVD disk is compliant with applicable DVD specifications. Some of its features may not be compatible with all DVD players.
Now what does that mean? On careful examination, I conclude:
There's no definition of “unauthorised”. In particular, I got this DVD from the library, which definitely comprises “lending”. Since this is done all the time by the libraries I know, I assume it's “authorised”. What else is authorized?
What is “copying”? They don't actually mention “playing” in this license, though arguably “playing” is a form of copying. The only issue is whether it's copied directly to the output or via the intermediate step of storing the data on disk. My application is “home use”, so I assume that this is acceptable.
There's no mention there of any restriction on what kind of device the DVD may be played on, just what kind it can be played on. So computers should be acceptable.
What features may not be compatible with all DVD players? Don't purchasers have a right to know if some obscure feature may stop it from playing in a standards-compliant DVD player? Reading elsewhere (this time in marginally more legible white on dark green) I see:
This NTSC Region 4 DVD will play in ALL DVD video players in Australia but the viewing television monitor must have NTSC & PAL (multi-standard) viewing capabilities.
NTSC? Why NTSC? This is a region 4 DVD sold in Australia by the Australian Broadcasting Corporation. And in Australia the standard is PAL. What earthly reason do they have to master the DVD in NTSC, which has a lower resolution and is only available to some people? It's not the origin, which is British (also PAL). I'm both baffled and marginally offended.
One way or another, the definitions are too difficult to interpret. They're not designed to annoy honest people like myself, but they do a very good job. Too many laws address secondary actions: “prohibit what we can control” rather than “prohibit what causes us harm”. That's particularly stupid here, because they can't control it either. And efforts to do so cost more than the potential damage.
This is probably not the only restriction on DVD usage. But it's the only one on this DVD package, so it has to be complete.
| Saturday, 22 October 2011 | Dereel | Images for 22 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Nickel-Zinc batteries: first real experience
|
Topic: photography, technology, opinion | Link here |
Today was garden photo day, and the first time that I did the verandah panorama using
Nickel-Zinc batteries in the
flash unit. Everything went well for the first 20 flashes or so. The recycle time was
under 3 seconds the whole time (with NiMH
batteries it's between 5 and 6 seconds). Then the unit didn't recharge. Looking at the
display, it showed a flashing 
symbol. Low battery already? I waited a few seconds, but there was no change, and since I
didn't really need the flash for the last few images, I just took them anyway.
After I had finished, took the batteries out. After that many photos, NiMH batteries get very warm, but these were not noticeably warmer than the surroundings. After charging last week they were at about 1.9 V. Before starting this morning, they showed voltages of 1.791, 1.802, 1.813 and 1.817 V. After the photos, they showed voltages starting at 1.730 V and going to 1.742 V—not, it seems, a difference in the batteries, but that they were recovering. An hour later they showed voltages round 1.765 V. I put them back in the flash unit, and it worked normally, again with recycle times a little over two seconds.
What does that mean? Certainly the batteries aren't discharged. The nominal voltage is 1.6 V or 1.65 V, depending on whom you ask. And the 1.73 V correspond exactly to the “Electrochemical open circuit voltage potential” mentioned on the Wikipedia page, whatever that might mean.
So why did the symbol light
up? It's reasonable to assume that it means “low battery”, but it could also mean “waiting
for unit to cool down”. What does the manual say? Yes, according to the manual it really
does mean “low battery”, only. That's not a foregone conclusion: it would be reasonable to
assume that the 
symbol
means “batteries exhausted”, but in fact it means “You have connected the external battery
pack, and there are still batteries in the unit itself. Please remove them”. On the other
hand, there's no indication for the cool-down phase described in the manual, so it's quite
possible that this really did mean “cooling down”, and nobody has bothered to document it.
So: what happened? Based on my experience with NiMH batteries, I had half expected the cycle time to increase round this point to let the unit cool down, and this could be exactly what happened. The only thing that confused me was the battery display, which could have occurred on other occasions. Next week I'll wait and see what happens.
|
Installing FreeBSD, revisited
|
Topic: technology | Link here |
When it comes to installing FreeBSD, I wrote the book. I've been doing it on a regular basis for over 15 years. So today when Chris Yeardley wanted to install FreeBSD, it should have been a breeze, right?
Well, no. She downloaded a DVD from the web and burnt it herself, and her laptop told her it wasn't bootable. So she sent it over along with a USB disk drive to install to. That requires a dedicated machine (or a VM, which I didn't even want to try). And only three machines have DVD drives: dereel, teevee and Yvonne's computer, lagoon. Did a bit of investigation and found an old Hyundai machine that CJ had left here after another case of Powercor damage which left the motherboard dead. It has 2 DVD drives in it. But how do you get them out? Hyundai seem to have gone to some trouble to make it difficult to work on:
|
|||||||||||||||||||||||
There's a plate riveted to the frame in the front of the side plate. Why? I later discovered that you can remove the front panel relatively easy and access the drives from there, but by then I had given up and tried a different tack.
There are other issues, of course: after installation, you have a basic system. I already have a build disk with all the ports I need, which would be a much better starting point for Chris. Fired up my test machine, and was quickly reminded that Powercor had put paid to the USB bus three months ago. In addition, the Ethernet NIC is now not working any more. OK, not a worry; just put the USB drive on teevee and copy across the net. Started updating to the latest version, which took the rest of the day.
| Sunday, 23 October 2011 | Dereel | Images for 23 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Apple “Pages”: pain itself
|
Topic: technology, opinion, gardening | Link here |
Today Yvonne picked up a USB stick from Helen Vincent with a draft of the Spring edition of Wellingtonia. Helen uses Apple “Pages”, and in the past we've had significant problems understanding it. In this case, she was unable to send me a copy of the document by email, because it proved to be 26 MB in size, and her ISP has a hard limit of 20 MB. How do normal computer users move files around? She could upload them to the server, of course, but she doesn't know how to use scp.
Thus the USB stick. And stupidly, I asked for the file only in “Pages” format, and not in anything that would show me the layout (such as PDF). I was completely blown away by what I found, after breaking up the single line of the document before every < character:
Clearly that's not what Helen wrote, but it goes on like this for half the document. And it's not as if I haven't seen lorem ipsum before and commented on the meaningless and incorrect syntax. But I didn't know what it meant. Now I know: “pain itself”, in accusative, and missing the first syllable (should be “dolorem ipsum”). Somehow that's appropriate. It reminds me of the story of the Microsoft TV commercial with music from Mozart's Requiem, which in the original had the words “Confutatis maledictis flammis acribus addictis” from the Dies irae.
Looking further, I found the real text, along with a couple of punctuation issues (primarily excessive spaces, something that WYSIWYG software almost encourages). But I couldn't make any sense of the document. And though there were some normal-sized images, there were still had lots of tiny images:
But I recognized at least one of these photos. It's the one I investigated a couple of months ago. This is the original size, so there's no point trying to enlarge it:
But what about these?
It seems that “Pages” carries around so much excess baggage with it that it's impossible to work out what belongs and what doesn't. “Pages” doesn't solve problems, at least not for me. It would be easier to write the thing from scratch in groff than help Helen with this software.
But why was the document so big? By the time I got it, it had grown to 66 MB. Not all the images were tiny. Looking at the index, I found:
That's nearly 50 MB of TIFF images, and the name suggests that they're not even being used—75% of an already bloated document!
|
Looking for a new keyboard
|
Topic: technology, opinion | Link here |
While writing my diary for yesterday, discovered I could no longer reformat paragraphs. Stopped Emacs, restarted it, no change. Then I discovered that the q key wasn't working, and the command is bound to M-Q.
You can get keyboards almost anywhere, and they're not expensive. They're also not what I want. For me at least, the keyboard is the most important interface to the machine, and it must be exactly the way I want. In particular, that means not having to look at the keyboard when typing (thus the complete unsuitability of touch pad keyboards). But you need to look at a keyboard to find the function keys in their current position, somewhere off the top of the main keyboard.
The only ones I've found that I can find without looking for them are to the left of the main keyboard, the way IBM made them until the late 1980s. And that's what I have been using for the last 20 years. The best keyboard I ever had was a Northgate OmniKey, and I still have four or five of them, but they're all 20 years old and in rough condition. The one I was using now was a flaky copy, the Avant Stellar. I have two of those, so I put the other one in. It worked for a while, but then the Ctrl key stopped working, which is even worse.
Found an OmniKey which looked as if it was OK, along with a cable, and that seems to work better, but a number of keys bounce. It's interesting to note the failure mode: the OmniKeys develop bounce, and the Avant Stellar, ostensibly a direct copy, suffers from key failures.
But this is stupid! Why should I be using any 20 year old component? Why can't I buy a modern keyboard? Did a bit of searching, but I couldn't find any. I can see some soldering coming my way.
|
Photo backup pain
|
Topic: technology, photography | Link here |
Ran my photo backup today. I write to an external disk, and until I get eSATA working properly, I've been using a USB connection. And from time to time the system (Yvonne's lagoon) freezes up. On one occasion it has reported I/O errors before doing so, but on others there has been no indication of why. Today I finally finished fsck and found 4 files in lost+found, so deleted them (the backup will put them back in the correct place), and—the system froze again!
Why do I have so much trouble with USB? Why do I even try? Put the eSATA card in defake and ran the backup from there. It works fine if you boot with the disk connected, but hot plug still doesn't seem to work. High time that I completed some of the upgrade projects I've been planning.
| Monday, 24 October 2011 | Dereel | Images for 24 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
USB installation, final part
|
Topic: technology | Link here |
I've been taking my time with the installation of FreeBSD on Chris Yeardley's USB disk. The last status was that I had copied all the data across—I think. I had turned off the machine before checking. Today I fired it up again and discovered some serious issues with the disk partitioning: c partition went beyond the bounds of the slice, probably a result of trying to copy the partition exactly from the other system, where the slice was larger.
So, blow it away and start again. Not a problem. Or was it? The partition table included 3 partitions: a and d, both 20 GB, and e, the rest of 1 TB. But the slice was only 256 GB, so I first removed partitions d and e. Then I tried to create a new partition d with the rest of the slice. “Operation not permitted”. ktrace shows that it happens on an open call:
I tried both with bsdlabel and sysinstall, with interesting results. bsdlabel showed these partitions:
But sysinstall showed:
Tried removing partition d and rewriting the label, but the rewrite failed. So I disconnected the drive, and the system (teevee, no monitor) froze.
Put the drive in lagoon while teevee was running fsck, and all went well. The rest was plain sailing, apart from the time it took to copy the ports tree, with multiple work directories, a total of about 25 GB. Finally I can give the drive to Chris. But there's something seriously wrong with USB disk support—maybe it's with disk support in general. This one suggests that some kernel copies of disk data don't get flushed at the right time.
|
More TV recording problems
|
Topic: multimedia, technology, opinion | Link here |
I've been running ceeveear with three tuners for over a week now. Time for the USB tuner to give up on me, and indeed it did. One recording hung up completely. I caught it in the middle, shut down cvr2 (which runs Linux, where I haven't been able to work out how to hot-plug USB devices), unplugged the tuner and continued (recording on the first of the PCI tuners). The remainder of the recording wasn't particularly good, but it was readable. Why do I have so much trouble with USB? It's not the operating system, and it's not a specific device. I don't like blaming the technology, but there's a good reason to do so here.
|
Friends' newsletter, impasse
|
Topic: technology, opinion, gardening | Link here |
Helen Vincent sent me a copy of the newsletter draft in PDF format today. What a disaster! I really, really don't understand what's going on here, and I'm not sure that I want to. So far we have been blaming the low resolutions on the low resolution images stored in the document, but that's clearly not the only issue. Here's an image as extracted from the document with unzip, and the way it looks in the PDF:
That's original size in each case. But the quality! Looking at the bird on centre left, the already low resolution image has been made much worse:
Why does this happen? I don't know, but somehow it's the last straw.
| Tuesday, 25 October 2011 | Dereel | Images for 25 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Installing FreeBSD on Samsung laptops
|
Topic: technology | Link here |
Over to Chris Yeardley's today with the USB disk for her Samsung RF511 laptop. In principle it should have been “plug in, frob a couple of config files, and we're done”. But it didn't work like that. First, the laptop has big problems with—wait for it—USB. I'm not sure why, but in many cases it didn't recognize the drive. We thought it might be a problem with the cable or one of the slots, and moving things around seemed to help, but even then it doesn't honour the boot sequence (first USB, then DVD, then internal disk) unless I first go into the BIOS and then out again.
All went relatively well until I tried to start X. Then I saw a message I've never seen before:
Investigation of the configuration file showed that it apparently has two screens, strangely with an nVidia "GF106 [GeForce GT 555M SDDR3]" and an Intel "2nd Generation Core Processor Family Integrated Graphics Controller"; presumably the latter is just a consequence of the choice of processor. But the real issue seems to be that this chip is too new:
Tried it with the proprietary nVidia driver, but that froze the system. Isn't proprietary software nice? Started loading the latest version of the driver, but it was via satellite at 8 kB/s, and it would have taken over an hour, so we put it off until tomorrow.
|
“Windows”—no longer broken, just scratched
|
Topic: technology, opinion | Link here |
As a result of the problems with rebooting Chris' laptop, I was presented with the “Windows” 7 boot screen for the first time. They've done away with the broken window emblem and the “my display panel is broken” bloom that earlier versions used, and instead introduced a number of scratches on the surface. I wonder if I'm the only person who sees it that way, or whether even somebody in Microsoft is doing it as some kind of joke.
| Wednesday, 26 October 2011 | Dereel | Images for 26 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Laptop installation, postponed
|
Topic: technology | Link here |
Over to Chris Yeardley's place this morning to complete the installation of the nVidia driver. It had failed: no kernel sources. Brought the thing back home, installed sources and driver, and it still froze. What a good advertisement for FreeBSD!
I suppose the next step to getting it running on this laptop is to reinstall X. That will take time, and Chris doesn't have time—she has an assignment which depends on it due in this week. So put a disk together in my test machine (the one with the broken USB bus) and took that over there for her. I can see more effort necessary to get her up to speed as quickly as she needs.
|
Another firefox crash
|
Topic: technology | Link here |
Since I gave up using windows with firefox, it crashes much less frequently. The last one I recorded was on 21 July 2011. Today was another case:
Another JavaScript problem.
| Friday, 28 October 2011 | Dereel | Images for 28 October 2011 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Weather station problems
|
Topic: general, technology | Link here |
The weather station reported an outside temperature of -3276.7° again today, for a considerable period of time. This time I looked at the internal unit: it was displaying --, suggesting that it couldn't communicate with the external unit. So maybe -3276.7° is its way of saying “no signal received”.
That's got to be batteries, right? Changed the batteries, and it worked again—for a while, about 3 minutes. Then I lost communication again. Tried to reset the internal unit by removing and replacing the batteries, but it didn't respond. Further investigation showed that it, too, had flat batteries. After replacing them, things worked normally again.
There are a number of things to note here. I had already thought that I had to reset the internal unit after replacing the batteries in the external unit. But I didn't realise that communication only dropped some time after replacing the batteries, and I still don't understand why. And it seems that the batteries aren't important for the internal unit once it's connected to the USB bus.
|
How to steal credit card details
|
Topic: general, opinion, technology | Link here |
The business with American Express got me thinking: to steal enough details about a credit card to access the account by phone, that's all you need. As a checkout person in almost any retail outlet, you do:
Find a credit card number, for convenience with an unusual name, say, “GREG LENEY” (a name American Express invented for me in the past) or FAUSTO D'AMBRO. It doesn't really matter which brand of card.
Look up name in phone book, get phone number and address.
Look up name on Google, get birth date.
Call up the issuing bank. Answer “security” questions. Get access to account.
The really, really silly thing is that the calling phone number (displayed on the “consultant”'s terminal) is the only thing even close to some kind of security. To get that to display, you actually have to use that phone, which usually requires some kind of physical access. Also a good reason to insist that the phone number is a land line.
|
Strange screen refresh issues
|
Topic: technology | Link here |
While editing my photo files today with Emacs, discovered a strange patchiness on the display:
|
|
||||||||
It happened several times with different kinds of smudge. I suppose this is an Emacs issue, but I haven't seen it before. Running the text cursor over the area recreated the text, so it's clearly a refresh problem, but what made it happen right now?
| Top of page | Previous month | Greg's home page | Today's diary entry | Next month | Greg's photos | Copyright information |
