I've had lots of issues
with NTP over the years,
and more and more I'm giving up on running the ntpd daemon: it seems too fussy. Poor
timekeeping hardware is a fact of life, and we've had issues with UNIX timekeeping
accuracy—without NTP—for as long as I can remember. Some months ago I stopped
using ntpd on my external machine, which is a virtual machine. Instead I
run ntpdate at regular intervals; it's not nearly as fussy.
Recently I have been seeing similar problems on cvr2.lemis.com, my computer video
recorder. I did the same thing there, except that, at least for the moment, I'm displaying
the output of ntpdate. It suggests that the clock gains about 2 ms per minute:
1 Aug 08:46:32 ntpdate[1815]: adjust time server 192.109.197.135 offset -0.002139 sec
1 Aug 08:47:32 ntpdate[1817]: adjust time server 192.109.197.135 offset -0.002128 sec
1 Aug 08:48:32 ntpdate[1819]: adjust time server 192.109.197.135 offset -0.002131 sec
1 Aug 08:49:32 ntpdate[1821]: adjust time server 192.109.197.135 offset -0.002129 sec
1 Aug 08:50:32 ntpdate[1823]: adjust time server 192.109.197.135 offset -0.002129 sec
1 Aug 08:51:32 ntpdate[1825]: adjust time server 192.109.197.135 offset -0.002130 sec
But sometimes that's not the case. This is getting the time from dereel.lemis.com,
on the same network, which is synchronized to four external servers. This example continued
this morning with:
1 Aug 08:52:32 ntpdate[1827]: no server suitable for synchronization found
1 Aug 08:53:33 ntpdate[1829]: no server suitable for synchronization found
1 Aug 08:54:33 ntpdate[1831]: no server suitable for synchronization found
1 Aug 08:55:33 ntpdate[1949]: no server suitable for synchronization found
=== root@cvr2 (/dev/pts/1) /recordings 46 -> ntpdate dereel 1 Aug 08:56:10 ntpdate[1951]: adjust time server 192.109.197.135 offset -0.150021 sec
=== root@cvr2 (/dev/pts/1) /recordings 47 -> 1 Aug 08:56:33 ntpdate[1952]: adjust time server 192.109.197.135 offset -0.140502 sec
1 Aug 08:57:33 ntpdate[1954]: adjust time server 192.109.197.135 offset -0.115489 sec
1 Aug 08:58:33 ntpdate[1957]: adjust time server 192.109.197.135 offset -0.090361 sec
1 Aug 08:59:33 ntpdate[1959]: adjust time server 192.109.197.135 offset -0.065123 sec
1 Aug 09:00:33 ntpdate[1961]: adjust time server 192.109.197.135 offset -0.039789 sec
1 Aug 09:01:33 ntpdate[1963]: adjust time server 192.109.197.135 offset -0.014346 sec
1 Aug 09:02:33 ntpdate[1965]: adjust time server 192.109.197.135 offset 0.002806 sec
1 Aug 09:03:34 ntpdate[1967]: adjust time server 192.109.197.135 offset -0.006571 sec
1 Aug 09:04:34 ntpdate[1969]: adjust time server 192.109.197.135 offset -0.001989 sec
1 Aug 09:05:34 ntpdate[2092]: adjust time server 192.109.197.135 offset -0.004155 sec
What was that all about? When I entered the same command manually, it worked. Looking
back, I see that the same thing happened in the middle of the night:
1 Aug 00:16:39 ntpdate[23671]: adjust time server 192.109.197.135 offset -0.002305 sec
1 Aug 00:17:39 ntpdate[23790]: no server suitable for synchronization found
1 Aug 00:18:39 ntpdate[23792]: no server suitable for synchronization found
1 Aug 00:19:39 ntpdate[23794]: no server suitable for synchronization found
1 Aug 00:20:39 ntpdate[23796]: no server suitable for synchronization found
1 Aug 00:21:39 ntpdate[23798]: no server suitable for synchronization found
1 Aug 00:22:39 ntpdate[23800]: no server suitable for synchronization found
1 Aug 00:23:40 ntpdate[23802]: no server suitable for synchronization found
1 Aug 00:24:40 ntpdate[23804]: adjust time server 192.109.197.135 offset -0.165361 sec
1 Aug 00:25:40 ntpdate[23922]: adjust time server 192.109.197.135 offset -0.139640 sec
1 Aug 00:26:40 ntpdate[23924]: adjust time server 192.109.197.135 offset -0.114256 sec
1 Aug 00:27:40 ntpdate[23926]: adjust time server 192.109.197.135 offset -0.088778 sec
1 Aug 00:28:40 ntpdate[23928]: adjust time server 192.109.197.135 offset -0.063217 sec
1 Aug 00:29:40 ntpdate[23930]: adjust time server 192.109.197.135 offset -0.037558 sec
1 Aug 00:30:40 ntpdate[23945]: adjust time server 192.109.197.135 offset -0.011177 sec
1 Aug 00:31:40 ntpdate[23947]: adjust time server 192.109.197.135 offset 0.003465 sec
1 Aug 00:32:40 ntpdate[23949]: adjust time server 192.109.197.135 offset -0.003604 sec
1 Aug 00:33:41 ntpdate[23951]: adjust time server 192.109.197.135 offset 0.000072 sec
1 Aug 00:34:41 ntpdate[23953]: adjust time server 192.109.197.135 offset -0.001847 sec
1 Aug 00:35:41 ntpdate[24071]: adjust time server 192.109.197.135 offset -0.001476 sec
1 Aug 00:36:41 ntpdate[24073]: adjust time server 192.109.197.135 offset -0.002038 sec
Initially I thought that it might be related to the activity on cvr2, but there's no
clear correlation. On Peter Jeremy's suggestion, started logging information
from dereel's ntpd. We'll take another look tomorrow.
Today was the first of the month, involving a complete backup of my systems.
For dereel, that involves running dump and tar, piping the result
through pbzip2 and writing to a disk on another system. pbzip2 is a parallel
process that uses all available processors, and it manages to max out my 4 processor system.
Today, not for the first time, the UPS started screaming “overload”. I don't
think that's the case: I suspect it's not prepared to deliver the same power that it once
did.
Not really an issue: over a month ago I bought new
UPSs, including one for dereel, but “if it ain't broke, don't fix it”,
since it involved taking down the system. Today it was clearly broke, so I swapped the
“850 VA” 500 W UPS for the new “1850 VA” unit, which I think means
1000 W. The VA ratings are just plain lies: they assume a cos φ of 0.6, which implies
big electrical motors. But on the new machine I read “Not to be used with fluorescent
tubes or non-computer peripherals”. It's not clear what a “non-computer
peripheral” is, but it's equally clear that the UPS is designed for computer power
supplies, which are required to have a cos φ of above 0.95. So at best the 1 kW UPS can
supply about 1050 VA.
On reboot, noted a console message:
Aug 1 11:07:37 dereel kernel: WARNING: /dump was not properly dismounted
That's a disk that I use for dumps from other machines. Some time ago, when I was having
problems with the system, I disabled automatic fsck for the disk, which doesn't
really do any harm: soft updates were enabled, so I could still use the drive. But clearly
it was a good idea to run fsck, so I did.
I wasn't quite prepared for the result: it found what appeared to be over a million lost
files, overflowed lost+found and ran for over an hour. The contents
of lost+found showed files that must have come from an older incarnation
of /dump:
/dump/lost+found/#0762030/Minimalist-wide/buttons/.svn/text-base:
total 1
-r--r--r-- 1 grog lemis 8438 Mar 10 2007 DVD_PLAY.png.svn-base
-r--r--r-- 1 grog lemis 6086 Mar 10 2007 DVD_PLAY_off.png.svn-base
-r--r--r-- 1 grog lemis 6409 Mar 10 2007 DVD_RIP.png.svn-base
Most of them
seemed svn-related. I wonder
where they came from. At the end of the hour, the file system looked pretty much the same:
the files were all relatively small, and the file system is 250 GB in size.
While waiting for that to happen, and in preparation for potential problems, connected up
the ALDI 1 TB disk that Yvonne bought on Thursday. It appears to be a Seagate drive:
Aug 1 15:22:56 teevee kernel: da0 at umass-sim0 bus 0 target 0 lun 0
Aug 1 15:22:56 teevee kernel: da0: <ST310005 28AS > Fixed Direct Access SCSI-2 device
Aug 1 15:22:56 teevee kernel: da0: 40.000MB/s transfers
Aug 1 15:22:56 teevee kernel: da0: 953869MB (1953525168 512 byte sectors: 255H 63S/T 121601C)
It also has an eSATA interface, which could
potentially be interesting, but currently I have nothing to drive it with. All in all a
good buy for $79—the drive I bought from MSYin June didn't have eSATA, and it cost $15 more
The days are getting longer, and there are indications that winter is coming to an end, but
not as quickly as I would like. Today the weather was noticeably cooler, and we got much
more rain than we're used to. So didn't do much outside apart from a bit of pruning,
planting some cuttings of Iceberg roses in the process. I really should finish the
greenhouse.
This camera has a clip-on light meter on top: the SV itself had no electronics whatsoever.
I've had a number of cameras since then, and I
still have a Pentax Spotmatic.
But one thing I don't have is a macro lens that works with bellows and other extensions. I
have a Zuiko Digital ED 50mm F2.0 Macro, but it only extends to 1:2 magnification, and the
only way to extend it further is with the Olympus EX-25 extension tube,
which costs an arm and a leg and only extends to 1:1. I can't use it with anything else
because it doesn't have an aperture ring, and the only way to set the aperture is via the
camera electronics. I've discovered that you can set the aperture like that, power off the
camera, remove the lens, and the aperture will stay at that setting. But that's of little
practical utility, and it mars a lens that I find otherwise superb.
I've also tried taking extreme close-ups with the 50 mm f/1.4 Super-Takumar that belongs to my Spotmatic. Unfortunately, the results aren't very good. The lens is
designed for normal focal lengths, and there are a couple of trade-offs to get the wide
aperture. But the Macro-Takumar 50 mm f/4 has a very good reputation, and the old ones extend to 1:1.
Interestingly, the newer ones don't: they had an “automatic” aperture (normally full open,
but stops down when the shutter is released and opens again afterwards). This seems to have
been difficult to achieve with a focusing system that extends the lens by fully 50 mm, so
they limited it to 35 mm, giving only 1:2 magnification. Since I can't use the automatic
diaphragm with my Olympus
E-30, it has no advantage for me.
But is this lens so good? The Olympus has 11 elements, including one with “extreme
dispersion” (dark green):
By contrast, the Macro-Takumar is basically
a Tessar design with only 4 elements:
So I didn't want to invest too much money in such a lens. But the few that come on the
market command surprisingly high prices, typically in the order of $150, about a third of
what I paid for my Olympus macro lens. I've been looking for some time, and finally I found
one on eBay—with a Pentax SV. Despite
that, I got it for $50, less than half the normal going price for the lens alone or for the
EX-25 extension tube. I suspect that was because the vendor got the details wrong: he wrote
55 mm f/4 instead of 50 mm f/4. Apart from the fact that there never was a 55 mm macro
lens, the photos make it very clear:
Amusingly, these photos (which included EXIF information) were taken on a day where I took photos of equipment for doing exactly the kind of macro photography I'm talking about. As I
wrote:
All I need now is a real macro lens.
So now I'll get the same camera again that I bought 45 years ago. I wonder if I should keep
it for sentimental value, or sell it again.
More investigation of
my NTP problems today. As
I suspected, NTP does not appear to be robust enough for my satellite IP environment, as the
following graph shows:
Both delay and offset increase until they go off the scale and ntpd performs a step
change. After the change, it answers any request with the error status
“unsynchronized”; it takes several minutes to recover. Thus the problem. I
don't suppose there's much I can do about it until I get a real Internet connection.
We're supposed to be in the middle of a decade-long drought, but I haven't seen that the
rainfall in recent years has been significantly lower than historical values. The
annual summary for 2009, the hottest year on record shows that Ballarat had 553 mm
rain instead of the normal 693 mm (elsewhere they claim it's 690 mm), but many other places in the area, notably south and west of
here, had much higher than usual
rainfall. Port Fairy had 868 mm
compared to the average of 664 mm, 30% more than usual. We don't have records
for Dereel, but until the end of July this yearBallarat has been closer the historical
average: 345 mm instead of 369. That was before the last weekend, where we had 42.2 mm in
Ballarat and 33.7 mm here. So no work in the garden. The pond even has water in it,
despite the porosity of the soil:
Somehow I'm not getting the big things in the garden finished. I need help from CJ to put
up the wind breaks, and I need to borrow his tin snips to make the glass clips for the
greenhouse. I could have worked on the pond, except there's still water in it. So did a
bit of weeding and pruning; the salvias
and osteospermums in the bed to the
south of the verandah have grown amazingly. In the meantime, Yvonne started preparing new pots for the spring.
We also have to prepare for the new vegetable garden: there were
two Lilly Pillys in the area, and I
transplanted one. I'm not sure it's going to survive: it had a long tap root which I
severed. Many Eucalyptus have a
similar root structure, and I know that they can't be transplanted. So I'll wait to see
what happens before I transplant the other.
The coming election is interesting because of the complete lack of interest that the
candidates for prime minster arouse.
As Laurie Oakes put it, they're
political pygmies.
And what about the Internet? Labor gives with
one hand—the National Broadband Network—and takes away with the
other—Stephen Conroy's
idiotic network filter. But we don't get the benefits of the NBN, which seems to think that
satellite communications are an appropriate technology for outlying regions. All we've
heard from the Liberals is that they would
scrap the NBN.
So, I idly thought, what does this say about the use of computers in the parties? I took a
look at the candidates and the technology behind their mail and web sites. The AEC has published a list of candidates for the electorate of
Corangamite, so I looked more carefully. Presumably they have an option of what
contact information they supply. Some provide addresses, phone numbers and email addresses,
others don't. It even differs between candidates for the same party. It's interesting that
the Australian Labor Party (alp.org.au)
have email addresses @australianlabor.com.au. Why? To make it more difficult to
write? This domain seems to exist only for email. There's
no www.australianlabor.com.au, and it doesn't even have an A record, only MX:
australianlabor.com.au mail exchanger = 5 alt2.aspmx.l.google.com.
australianlabor.com.au mail exchanger = 10 aspmx2.googlemail.com.
australianlabor.com.au mail exchanger = 10 aspmx3.googlemail.com.
australianlabor.com.au mail exchanger = 1 aspmx.l.google.com.
australianlabor.com.au mail exchanger = 5 alt1.aspmx.l.google.com.
So Labor uses Google Mail for their email communications. I'm amazed.
Looking at the web sites, the Democrats (who I
thought were dead), the Greens and the Australian Sex Party use
Linux (the Greens used to use FreeBSD; I
wonder why they changed. The Sex Party appears to be hosted in Houston, Texas; most of the
others are at least hosted in Australia. The Liberals and the
Citizens
Electoral Council (interesting only because the Greens put them below the Climate Sceptics on their
how-to-vote card) run Microsoft, as does the Australian Labor Party—now. Until a couple of months ago they were running Linux. From the change of IP
address, it looks like they have changed their hosting arrangements, and they're not
interested enough in the technology to want to determine the software. The Climate Sceptics
are interesting because they have recently changed their web site name from http://www.climatesceptics.com.au/ to http://landshape.org/news/. They, too, are running Linux, apparently in Seattle, USA.
So what does this all mean? It's interesting that, with the exception of the Climate
Sceptics, the smaller parties all use Linux. Maybe it's nothing more than a reflection of
what the web hosting companies are running nowadays. None of the candidates give me the
feeling that they use computers gainfully in their everyday life.
OK, I've had enough fun with Λκουλ. It might have been worth trying to get the wireless
card to work, though I fear it would have meant writing a driver, but basically a resolution
of 1024×600 is just not enough, especially when the web browser reduces the effective height
of the display to 376 pixels. So replaced the original software and packed it up. It'll go
back tomorrow.
The weather was cool and wet again today, so spent most of the day indoors. But we have
plants there too: this succulent I've been observing has disproved my theory that it's
following the light: it has now straightened out again, away from the light, and I suspect
it has grown a couple of centimetres since the last photos. Even the buds at the end are
getting further apart. Ended up propping up the stem on
our Hanukkiyah:
The next step was to attach the bamboo slat screening. I had intended to nail them to the
rails, but the things looked too flimsy, and the slats weren't even all the same length:
No help from the packaging: there's a photo of the screening standing up, but with no
visible means of support. Probably the manufacturer couldn't think of anything either, and
the photo was taken with people holding up the ends. CJ came up with the idea of nailing
them beneath the metal strips that we had used for the verandah, which looks like a good
idea. But we didn't have enough, so I'll have to go and buy some. Another half-finished
job. Then the weather got worse, so I didn't do anything else either.
One of the things that continually puzzles me is how to identify plants. Today it was the
mystery succulent with the long flower stem. Did some searching on the web and really came
up with an identification: it's a
Gasteria carinata var. verrucosa
“variegata”, to give it its full name. Found some reasonably plausible growing recommendations: looks like it shouldn't get any water at the moment, despite the fact that it's
flowering. Other
photos suggest that there's not much more to expect of the flowers: none of them show
the flowers opening. We'll see about that. At the moment the stem is growing over a
millimetre per hour. Put up a glass jar (all I could find) to mark the current position of
the tip:
Wendy McClelland has sent a letter to the editor of the Golden Plains Miner, the local newspaper, making
claims that she must know are untrue: we don't need a tower because the one
in Corindhap will cover us, and we
don't need it for networking anyway because we will soon get “nano-fiber optic
cabling” from the Government. Neither is correct: as the Optus project leader said,
and I mentioned last month, we can't expect much
coverage from that, and we've already confirmed that the National Broadband Network doesn't care
about Dereel.
So why does she make these claims? Is she lying, or just not checking even the most basic
facts? If she makes such claims about things that can easily be shown to be incorrect, why
should anybody believe her when she makes claims that are not so easy to disprove? Wrote my
own letter to the editor. I wonder what
will happen in the mid-term.
For some years now I've been toying with the idea of writing a cookbook for people with a
technical background, tentatively titled “Groggy's high-tech cookbook”. Beyond
thoughts about the material, it hasn't got very far. And now I've been scooped: Jeff Potter
has written “Cooking for
Geeks: Real Science, Great Hacks, and Good Food” Took a look at it on Safari. It's certainly not the
book that I would have written, but it's close. Probably one to buy.
At the time we thought it was a possum, but the
droppings make it clear that it's a rat. Piccola is delighted and spends much time in the shed, but she hasn't caught it yet. Yesterday
Yvonne bought a rat trap—out of plastic, which I
thought might be a little too weak. But I wasn't prepared for what I saw this morning:
The rat had somehow tripped the trap without being hit. Not only that: it had eaten
the part of the trap that holds the bait. Here's a comparison with an intact mouse trap of
similar construction:
Last month I wrote letters to Darren Cheeseman and Sarah Henderson, the two main candidates for the
seat of Corangamite,
outlining the concerns that I have with the local infrastructure: network connectivity,
power reliability, bushfire protection, mobile phones and TV reception. Today I got a reply
containing an express prohibition of publishing the contents, so I can't. She only
addressed one of my five points, mobile phones.
Well, that's something, I suppose. But the other four issues were all more important to me
than mobile phone coverage. This response implies that she doesn't care about the others.
So: should I put her ahead of Labor because she replied, or
behind because she doesn't intend to do anything about the real issues?
The sun's shining again, and I forgot to go into town to buy the metal strips for the wind
break, so did some weeding and transplanting. At least one plant,
a Euphorbia “Diamond
Frost”, which we bought some months ago, is already looking very unhappy:
I've already complained about the ABC's
idiotic or politically motivated decision to stop broadcasting high definition TV. But it
seems that's not enough; the standard programme is now 576i, the same resolution as we've
had for 54 years. That would be bad enough, but it seems that some of their own productions
deliberately make it even worse. Today I watched an ABC production “The Making of
Modern Australia”, reasonably interesting, but with appalling picture quality. It's
not immediately apparent from the images, but the details of the flagstaff at the right
change from one frame to another, far more even than the low resolution would suggest. The
details below are from a stationary image, so motion isn't involved. At a
“normal” size it's almost acceptable:
The result is a very visible flickering. There are more artefacts of this nature in the
image, notably to the right of the tree. The image is new material, as far as I can tell,
so there's no reason why it shouldn't be recorded in 1080p.
This doesn't happen with other programmes, only with ABC. I've seen it on multiple
programmes, but only since they got rid of their HDTV broadcast. The kindest thing I can
think of is that they're deliberately mutilating their broadcasts so that people can't steal
it. But is that acceptable for a public broadcaster?
The weather was nicer today, so did more work in the garden, in particular some pruning. In
previous years I wasn't sure how to prune
the Aloysia triphylla
(Lemon Verbena). Last time I erred on the side of caution with one bush, and cut the other
back hard a bit too late in the spring. Both survived (until I pulled the second one out to
build the wind break currently under construction), so today I decided to prune the other
one hard:
More work in the garden, and again not much. I'm gradually getting the shape of the garden
pond sorted out, but it's heavy work, and I'm beginning to feel my age. Still, if I
continue like this it should be finished in a few days.
Also some more weeding. In the area to the south of the verandah, and only there, I've
discovered a new plant. It's almost certainly a weed, but which?
Chris Yeardley is learning a new programming language this semester—C++. I suppose
it's “modern”, but I've given my
reasons why I stopped using C++ years ago. Still, one of her assignments interested
me: solve a Sudoku puzzle. Their textbook
is Introduction to Programming with C++, by Y. Daniel Liang, and it includes downloadable source code for
this program, but only for the standard 9×9 puzzle. Chris has to modify it for a 16×16
version. That sounded like fun, so I took a look at it.
As I discovered years ago, there are really some advantages to C++, and in some ways I wish
I had stuck with it. But for the fun of it I converted it back to C. With my previous
experience of this sort of thing, I thought it would be easy, especially since the program
uses almost no specific C++ features. There's stuff like this, of course:
cout << "Enter a Sudoku puzzle:" << endl;
I have always found this << construct incredibly
ugly, but of course it's trivial to convert it into a (much shorter)
puts ("Enter a Sudoku puzzle: ");
But the real issue I had was with the parameter passing. Liang allocates an automatic 9×9
array in the main function and pass it as a parameter to the other functions, so
you have things like:
int getFreeCellList(const int grid[][9], int freeCellList[][2]);
void printGrid(const int grid[][9]);
bool isValid(int i, int j, const int grid[][9]);
bool isValid(const int grid[][9]);
Is this good practice? I don't know. In a purely functional programming language it would
be. The alternative, which I would have used, would be to declare the grid globally and
refer to it directly. But in general you don't want programs with side effects, and this
would be one. On the other hand, language restrictions require you to specify the bounds of
all except the first array dimension, and that makes it not only extremely ugly, but also
impossible to choose different array dimensions at run time. I suppose you could set them
to a maximum value and only use as many elements as you want, but that looks tacky too.
The other issue was the use of const. Sounds like a good idea, but, not for the
first time, I couldn't find a way to get C to accept it. I didn't try very hard, since the
whole issue is moot if I change it to a global array. Interestingly, the copy of the code
that I downloaded is
not quite the same as the version that Chris gave me, and it doesn't include
the const statements.
The whole thing took me a couple of hours, mainly wondering what the “correct”
way is. It's also interesting to consider to what extent sample code in books perpetuates
itself through the people who learn from the books.
Into town today for a number of things, including picking up my “new” Asahi Pentax SV and buying some
hardware and a rat trap with metal components. Also to RACV to get an insurance quote. I'm currently with
Elders, and they have significantly hiked
their prices since last year. RACV proved to be over 20% cheaper, so started signing up for
that, and then discovered that they wanted a payment of $40 per year to perform direct
debit. What kind of stupidity is that? It makes life easier for all concerned. Why should
I have to pay for it? Disgusted, cancelled the policy.
Back home, played around with my new toy. Despite careful checking, it proves that it
didn't have the lens I had expected. As I wrote last
week, I thought it was a Macro Takumar 50 mm f/4. In
fact, it proved to be an SMC Macro Takumar 50 mm f/4, with an “automatic” diaphragm, which I also
mentioned as having no advantage for me. I should have noticed that from the photos, but I
was misled by the age of the camera; in fact the lens is considerably newer. The optics are
identical, but I note that the newer lens is supposed to have a better coating, so I now
have the disadvantage of a shorter focus range and the advantage of better coating and a
better resale price. I can live with that (the intention was to use it with bellows or
extension tubes anyway), and it wouldn't have stopped me buying the camera, but I'm annoyed
with myself that I didn't recognize the difference.
Both camera and lens are in really good condition for their age, far better than my
Spotmatic. It's also interesting to note how small the SV (the black one) is, even smaller
than the Spotmatic (same depth and width, but about 5 mm less high):
That has to do with the electronics, of course. The SV is completely mechanical and has no
light meter. I can't see myself ever using it, but at the price (possibly negative), it's a
nice collector's piece.
Set to to do some comparative photos with my three 50 mm lenses (the others being the
50 mm f/1.4 Super
Takumar and the Zuiko Digital ED 50mm F2.0
Macro), and chose the contacts of an old 2½" disk drive as the subject. That
proved to be quite good, particularly because of the sharp delineations and high contrast.
The results? Here corresponding ones with the Zuiko, Macro-Takumar and Super-Takumar at
approximately 1:2. At this level, they all look OK.
I still need to analyse them, but it seems that the Macro-Takumar is much
better than the Super-Takumar. None of the lenses really made me happy at very close range:
I'm wondering if this isn't a matter of diffraction effects. I'll analyse the photos
tomorrow and hopefully come to more understanding.
One of the problems I had were the EXIF data for the photos taken with the Takumars.
They're completely mechanical, of course, and the camera can't record any information, so I
had to modify my scripts to put it in manually. That shouldn't be a problem for exiftool, but how? The man
page exiftool gives some information, but not nearly enough, and refers to a man
page Image::ExifTool::TagNames, which I couldn't find.
I later discovered that this is because it's a perl module, and perl wants you to use
perldoc.
Found the documentation on the web, but I still wasn't out of the woods. The documentation
refers to only one lens parameter:
0xfdea Lens ExifIFD string/
But exiftool itself outputs information like:
=== grog@dereel (/dev/ttype) ~/Photos/20100809 482 -> exiftool orig/P8099432.ORF |grep -i lens Lens Type : None
Lens Serial Number :
Lens Firmware Version : 0
Lens Properties : 0x0
Lens ID : None
Which do I use? Images taken with the Olympus macro return:
=== grog@dereel (/dev/ttype) ~/Photos/20100809 485 -> exiftool orig/P8099394.ORF |grep -i lens Lens Type : Olympus Zuiko Digital ED 50mm F2.0 Macro
Lens Serial Number : 010110933
Lens Firmware Version : 1.008
Lens Properties : 0x4003
Lens ID : Olympus Zuiko Digital ED 50mm F2.0 Macro
By trial and error, I found that I can collapse these tag names by removing spaces and
downshifting, at least in the cases I was looking for. So I tried setting “Lens
Type” or “Lens ID”, but that didn't work:
=== grog@dereel (/dev/ttype) ~/Photos/20100809 487 -> exiftool -overwrite_original_in_place -lensid="Ashai Optical Co SMC Macro-Takumar 50 mm f/4" P8099422.JPG Warning: Expected one or more integer values in XMP-aux:LensID (ValueConvInv)
Nothing to do.
=== grog@dereel (/dev/ttype) ~/Photos/20100809 488 -> exiftool -overwrite_original_in_place -lenstype="Ashai Optical Co SMC Macro-Takumar 50 mm f/4" P8099422.JPG 0 image files updated
1 image files unchanged
Finally found a tag “Lens Model”, which Olympus doesn't set. But it works:
=== grog@dereel (/dev/ttype) ~/Photos/20100809 489 -> exiftool -overwrite_original_in_place -lensmodel="Ashai Optical Co SMC Macro-Takumar 50 mm f/4" P8099422.JPG 1 image files updated
=== grog@dereel (/dev/ttype) ~/Photos/20100809 490 -> exiftool P8099422.JPG | grep Lens Lens Type : None
Lens Serial Number :
Lens Firmware Version : 0
Lens Properties : 0x0
Lens Model : Ashai Optical Co SMC Macro-Takumar 50 mm f/4
Lens ID : None
So I had to modify the PHP scripts to look in two different places. That works, but why is
it such a pain? Why does the documentation use two different forms of name and expect you
to guess the correlation? I suppose there are other things that I could try, so this is
probably not the end of my playing around.
I've slowed down almost to a standstill with my brewing efforts until I can get my
infrastructure woes sorted out. In the meantime I've been trying various commercial
offerings. Today to Dan Murphy's and
bought two relatively cheap beers: Cascade Premium Light and “Original” Oettinger Pils. Tried them
both, and wasn't very impressed by either. The Cascade has taken an approach to lighter
beers that I think is designed to make people choose full-strength beer instead: it seems to
include a significant quantity of malt (or maybe barley or wheat) roasted in a way that
reminds me of breakfast cereals. My own approach has been to use the same malts as for
full-strength beer and increase the hopping level. I think I'll be able to finish the
Cascade, but I doubt I'll buy any more of it.
The Oettinger is another matter. It tastes just plain boring. It, too, could do with more
hops (“Pils” indeed!). But possibly I'll get used to it. Despite the name, the
beer comes from Gotha, and not
from Oettingen.
CJ was going to come over today and help me with the myriad remaining jobs in the garden,
but just as he was planning to leave, we got a few drops of rain. It was also dark and
dreary, so we decided to put it off until Friday. That proved to be the correct choice: we
got some pretty heavy rain in the course of the day, a total of 10.2 mm for the 24 hour
period, and the top temperature only hit 8.7°. Still, during a lull in the rain I managed
to spread some fertilizer on the north part of the garden (up to the smaller succulent bed)
and also do all the roses and citrus plants.
More investigation of the missing documentation for ExifTool today. Somebody told me
about perldoc, apparently what you need to read perl documentation. And yes, the
documentation was there, in /usr/local/lib/perl5/5.8.8/man/man3/. But why
can't man do this? Why do I need another program?
Ah, that's because perl isn't a UNIX-only language, and the documentation is in a
different format. Or at least, that's what I was told. Took a look
at /usr/local/lib/perl5/5.8.8/man/man3/Image::ExifTool::Olympus.3.gz and found that
it was groff source, so tried running it through the mandoc macros:
NAME
Image::ExifTool::Olympus - Olympus/Epson maker notes tags
So they are man pages. Why perldoc? Because they're in the wrong place?
Adding /usr/local/lib/perl5/5.8.8/man to the MANPATH environment variable
was all I needed to get man to display them.
Further investigation showed that the real issue was that my .bashrc predates
FreeBSD. FreeBSD has a
file /etc/manpath.config with details of which paths to set, and it includes:
# added by use.perl 2008-10-19 15:59:46
OPTIONAL_MANPATH /usr/local/lib/perl5/5.8.8/man
OPTIONAL_MANPATH /usr/local/lib/perl5/5.8.8/perl/man
But I set MANPATH in my .bashrc, so it didn't get set. Time to overhaul the
file.
As planned, started trying to analyse yesterday's macro photos of the pins on a 2½" disk
drive today. That was a non-starter. Despite being mounted on a sturdy tripod, most of
them had noticeable camera shake:
It's due to the image stabilizer in conjunction with a lens which doesn't report its focal
length. For some reason, my camera still had a manual setting for 800 mm, and the image
stabilizer compensated for movement which didn't exist. So I had to start all over again.
Even this image showed something, though: there's no detail. It looks like it's out of
focus, but it isn't. It doesn't seem to be diffraction either. So today I took fewer
photos and compared them.
The results? First the good news. The image quality of the SMC Macro-Takumar 50 mm f/4 was much better than that of the 50 mm f/1.4 Super-Takumar.
In particular, the Super-Takumar has considerable chromatic aberration and flare. Here
photos taken with bellows (about 3:1 magnification) with the Macro-Takumar (left) and the
Super-Takumar (right), first at the centre, then at the edge:
Image title: Disk Macro Takumar flash bellows NoIS 22 centre detail
I'm not sure about the difference in contrast. I suppose I should investigate that more.
The bad news: not a single photo was satisfactory. They all show the same kind of minimum
feature size that has irritated me before. The following three images were all taken with
the Macro-Takumar. The first two are taken at closest focusing distance (about 1:2 image
size), and the third is with extension tubes (about 1:1 image size). The first image is
taken at f/5.6, the other two at f/22:
As you'd expect, the depth of field of the image taken at f/5.6 is far less than that of the
other two; but the base of the pin is sharp. At f/22 the depth of field is considerably
better, but the overall sharpness isn't, maybe due to diffraction. And at 1:1, the
sharpness is better (since the image is bigger, which doesn't show here). But the outline
of the dust on the pin is still unsharp.
What about the Zuiko Digital ED 50mm F2.0 Macro? The closest I can come with it is 1:2, since a
single extension tube would cost considerably more than I paid for the Macro-Takumar, the SV
and the postage combined. I do have a 10 dioptre close-up lens, though. With those, I can
get:
That's still not sharp enough, but there's not much in it, and arguably the last
photo is the best.
But wait, there's more: mount the Macro-Takumar on bellows and get an additional extension
of 184 mm. In conjunction with the extension tubes (61 mm) and the extension of the lens
itself, (26 mm), I would have a total extension of 271 mm, giving me a magnification of
about 5.4:1. But what image quality would I get at that magnification?
Another fondue de fromage in the evening, this timea
with my white sourdough bread. Not a resounding success for a number of reasons. The
bread's not right for the dish, and we ended up with more religieuse at the bottom of
the caquelon than I have seen in some time:
The weather yesterday may have been unpleasant enough, but today was worse: not just showers
but heavy rain, a total of 32 mm (of which the weather station only recorded 25.8 mm). Stayed in the house nearly all day.
My test photos weren't overly spectacular, but more “normal” macros can be fun.
Spent some time taking some photos of
the Gasteria carinata. I'm still
wondering whether the flowers will open; you'd think it would be a necessity, but all I've
ever seen is what I got today:
One of the results of the bad weather was that I didn't get round to looking for the rat
until the evening. When I did, yes, we had caught a rat. But I still wasn't prepared for
what I saw. The rat was so mutilated that I've put the photos on a different page. Follow the link if you dare. Part of the
head and a forelimb appear to have been gnawed off. At first I wondered whether the rat
could have done that itself while dying, but the best I can think of is that another rat
came along, stole the bait from the dying rat's mouth and then continued gnawing. It seems
that rats can be cannibals.
Yet another day of rain! Not as heavy as the previous two days, but it's still getting on
my nerves. We've had a total of 49.2 mm over the last three days, about 8% of a typical
year's rainfall. So, once again, spent the whole day inside.
We're seeing lots of kangaroos lately, and we're continually chasing them away. But it's
been some time since I've seen this many from my office window:
The name of the Australian National Broadband
Network contains a buzzword that no longer has much meaning. What is
“broadband”? I suspect it's the up-and-coming word for
“Networking”. It's not alone amongst the meaningless, sliding-scale terms.
There's ”super-fast“, “ultra-fast” and “high-speed”.
What do they mean? Which is the fastest? When I first heard of “broadband”, it
was in conjunction with ISDN, and it implied
what still earlier had been called
a primary channel, 1500 or 2000
kb/s. I suppose that's still “broadband”, but most certainly not cutting-edge.
The NBN is gradually coming on line with speeds of up to 100 Mb/s—in areas which, it
seems, are all catered for by ADSL. It's not surprising that the interest is only marginal.
Most people don't seem to be concerned about the difference between 20 Mb/s and 100 Mb/s.
But there's one word that hasn't entered the buzzword jungle: latency. And it's the most
important one of all. It's been a while since they stopped talking about long fat pipes,
but the problems haven't gone away. I'm told that some people on fast urban (fibre)
networks have a round trip time of 4 ms. Most ADSL I have seen is in the 10-20 ms range.
My typical RTTs with satellite connections are in the order of 1000 ms (one second). And
that seems the most obvious reason I can see that it can take me up to 2 minutes to load a
web page which others can load in a second or two. It's not the link speed: that's reliably
1 MB/s or more. But the latency kills, and it seems that nobody who is running the NBN
scheme understands that. So, while the people with fast connections get faster connections,
people in smaller towns get no benefit.
As Callum Gibson puts it:
What the NBN gives us is freeways to a lot of people with a 3 lane
driveway, joined together by some highways and some people still with
dirt tracks. And I won't be happy until grO0gle stops complaining
about his internet connection, dammit!
Amen.
The NBN is—marginally—an election issue. At least the Labor have recognized the need for a better network
infrastructure, though the way they have gone about it seems really bizarre. There must be
some good reason why they can't force Telstra to split their network infrastructure into a separate publicly held company, but nobody in
the media has even mentioned the trade-offs involved. It's still better than what the
Liberals appear to be offering—only
12 Mb/s, considerably less than many people can get now. I suspect the latter is simply an
indication that the politicians involved have no idea of the concepts.
In a democracy the majority rules! ...
The majority doesn't want a tower. ... The majority has spoken.
Clearly the editor of the Golden Plains Miner has spoken. I never thought much of the
newspaper, but if they can't even present both sides of a dispute, they're completely
useless.
Finally the rain has stopped, and CJ came over to help me put up the bamboo wind breaks. As
we had suspected, that was more complicated than we would have liked, not helped by the poor
quality of the screens themselves. As a result we discovered—too late—that the
wiring holding the slats together was asymmetric, and we had put one up upside-down. CJ
thought that we might be able to keep it a secret, but I told him that it would be all over
Google within a couple of days.
It didn't take long, but it wasn't as pretty as I had hoped, and it didn't change the
appearance as much as I had expected:
The next step is to hang some old fencing mesh over the “outside” (away from the
verandah) and plant it with creepers. We would have done it today, but I need to buy some
staples. After that we'll see how well it works against the wind—if, indeed,
it can withstand it.
Yvonne borrowed a real, commercial DVD from Chris yesterday,
“Where the wild things
are”, and today we tried to watch it. It says “region 4” on the package,
without specifying what that means; but after checking, it seems to include Australia, so it
should be OK. What happened? The thing started playing and got itself locked in a loop
explaining why DVD piracy is such a bad thing. We couldn't watch the DVD. I'll have to
copy it to disk so we can watch it, presumably exactly what the DVD industry doesn't want me
to do. Protecting intellectual property is a valid thing to do, but why does the industry
(in this case, Warner Brothers) have to
annoy their customers? I know this has all been said before, but it still annoys me.
Yvonne and Chris left today for Olivaylle, this time without me. They'll stay
overnight and return tomorrow with a couple of horses. That leaves me here by myself. In
principle, it doesn't change much, but it's been such a long time since I've been here by
myself that it felt strange.
The weather didn't help. The rain has let up (it would be too much to say that it has
stopped), and instead we had high winds, the second windiest day since I started keeping
records.
mysql> select date, max(wind_gust) from observations where wind_gust > 39 group by date; +------------+----------------+
| date | max(wind_gust) |
+------------+----------------+
| 2010-07-10 | 48.1 |
| 2010-08-14 | 39.7 |
+------------+----------------+
One result actually happened yesterday, but I didn't find out until today: another part of
the Cathedral has broken off:
The rain has made its presence felt, though: the dams are as full as I have ever seen them,
even fuller than the first of the photos I took on 10 November 2007, a
week after the heaviest rainfall we've had here so far.
Even the lagoon has puddles of water in it:
Today was also photo day (so was 10 July 2010, the windiest day on my
records), and once again I had difficulty with the photos, which I do
in high dynamic range.
Basically, there's no solution: either take the photos and put up with the double images, or
take single low dynamic range photos. I chose the former, though I wonder if I shouldn't
reconsider: the LDR images are there as part of the HDR group, so I could always reprocess
them.
I did reprocess some of the photos I took two years ago, when we were just starting to build
the verandah. Some of the photos were pretty
terrible. At the time I was taking only raw images and converting them
to JPEG with ufraw, obviously not very well. In the
meantime I've been able to tune it better, so tried that and got a considerable improvement.
In general, though, I think my current procedure is better: take dual format photos, raw and
JPEG, use the JPEG when it's OK, and convert the raw image when it's not.
My comments yesterday about the difficulty of
using commercial DVDs aren't original, as I mentioned at the time. A couple of people have
pointed me to an image URL which sums
things up nicely:
Yvonne spent the night at Olivaylle, so I had to fend for myself today, including
preparing dinner for when she and Chris returned. That's not a big problem, since I do a
lot of cooking anyway, but in practice we divide responsibilities: I cook some dishes,
Yvonne cooks others. Today we had
“gyros” on the menu. That's
not the real döner kebab, which is
roast on a spit: instead we chop the meat up, marinate it and then fry it on the barbecue.
But how fine to chop it? What marinade? Yvonne didn't write down what she does, so I had
to experiment, and came up with this recipe. Also
had to make tzatziki, which caused less
head-scratching. Found a recipe on the web
and adapted it. It later proved to be slightly overloaded with garlic, and I think I'll
experiment a little more before writing anything up.
Apart from that, had to bake some bread, cooked some nasi lemak, made a Weißkrautsalat, tided up the kitchen and somehow managed to spend most of the day
there.
Between cooking, managed find time to look at the TV programme on cvr2.lemis.com.
But I couldn't: it had been down for 2½ hours, truncating a recording in the process. Went
to the cupboard to see what was going on. It was running, but no display, and the keyboard
didn't react to the NumLock key: dead in the water. Rebooted, which worked fine,
with no obvious reason for the crash. Shut the doors, went back into the office
and—it was down again! It took a couple of repeats to realise that it was related to
shutting the door; possibly the power lead was loose. The time it went down was probably
when I put the bread pan in the cupboard to keep it warm. Reseated the power cable, moved
the computer further from the doors, and had no further trouble. But it's strange that the
computer should wedge from that sort of problem.
Yvonne and Chris back in the evening, having brought three
horses with them: Jorge is giving up his horse breeding operations, and he ended up giving
them three horses: Zarzuela for Yvonne, and Capricho and Chewie for Chris. There's really
an “end of era” feeling about the whole thing, and it seems there was rather a
sorrowful parting.
The rain has let up a little today—so far this month we've had over 90 mm
rain—so off with Yvonne and Nemo to investigate how full the dam was. The two halves still
haven't quite joined:
As we got to the fence to the lagoon, we saw a number of kangaroos. Nemo was not on a
leash, and headed in their direction. I was just saying “Never mind, he won't get
through the fence” as he went through it. In a couple of seconds, he was gone, and we
followed with thoughts of kangaroos tearing him to pieces. We finally saw him almost on the
other side of the lagoon, and he came back little the worse for wear:
I had thought the marks on his legs were mud, but they proved to be burrs, which Yvonne had
difficulty to remove. Did some thinking about the dangers of kangaroos. They have two
basic ways of killing dogs: tear their guts out, or drown them. Before they do that,
though, they need to be cornered, and I suspect they just outran Nemo. And maybe it's a
blessing in disguise: last time a dog chased kangaroos round here was about a year ago, when
Diane Saunders' dog Roxy chased them off the property. After that, we didn't see any
kangaroos around here for six months or so.
By coincidence, got a call from Diane in the evening. Roxy had been chasing kangaroos
again, and one nearly drowned her. Di thinks that it must have been a young one, because an
adult would have torn her guts out. Anyway, Roxy is OK, but it's a timely reminder that
kangaroos can be dangaroos.
I'm still not happy with my macro photos of
the gasterias, and it'll probably be a
long time before I am. Planned to take some photos inside the narrow trumpet of the flower,
and got as far as a series taken with the SMC Macro Takumar 50 mm f/4 at various apertures. Here images at f/22, f/8 and f/4:
Of course the depth of field drops off as the aperture increases. But a careful comparison
shows something down-like on the petals, here the top left green patch on the photos above:
There's very little in it, but is it possible that the central photo (taken at f/8) shows it
most clearly? That would fit the theory of a trade-off between depth of field and
diffraction. This is an area about 0.3 mm across, but I'd still expect it to look sharper.
Apart from that, yes, there are several anthers visible inside the flower:
Last week I tried to get insurance at RACV, only to discover that they wanted to charge me $40
for direct debit—something that should be cheaper, if anything, since it lessens the
administration effort. On consideration, though, it appeared that the consultant had been
confusing direct debit with monthly payments. Today called up the RACV sales line and was
connected to Lucy, who asked a couple of security questions—date of birth, address and
telephone number, all easily findable on the web—and then told me that she had the
details of the insurance policy stored, and wanted to go through it. I asked her if it was
true that they would only do direct debit on a monthly basis, and she confirmed. I asked
her why, and she said that they would only do direct debit on a monthly basis, not exactly
an answer. After a bit more running round in circles I asked her to get me a statement from
the people who made the decisions, so she connected me to her superior, Janine.
Janine understood the question, and, it seems, picked the most likely explanation out of her
answer book: they can't do it annually because of the legally required 21 day cooling off
period before renewals. Otherwise they would be debiting money from my account without my
permission. That's nonsense, of course; they'd do the same if it were monthly. But I
couldn't get her to understand that the answer she was giving didn't relate to the question.
Finally she connected me up another level to Lee, who first took the time to find out the
reason: it seems that they did it up to about 7 years ago, but then stopped because of lack
of demand. That makes more sense, but it opens up all sorts of questions: why is there so
little demand, and what additional programming effort is required to do annual debits
instead of just monthly ones? I'm amazed.
Discussing it on IRC, though, showed another issue: almost nobody does direct debit.
To quote Andy S (he hides his surname): “direct debit is pretty evil”. But why?
Are people really so in love with cheques and required actions to pay bills that they don't
want it any easier? One possibility seems to be what happens when funds are debited without
authorization. In Germany, where direct debit is the standard way to pay recurring debts,
you can get an automatic annulment of any direct debit from your account. What happens
after that is another matter, of course, but at least you have control. In Australia, it
seems the law doesn't allow it, which also means that there is no mechanism in place for
such an annulment. I'm amazed that we're still living in such primitive times.
I was shown another cartoon about copy protection today:
Of course, there's a solution that's obvious, legal and effective: don't buy any of this
stuff. I never have done, and I won't until companies stop trying to tell me how to live my
life. And if I want to see a film, I wait until it comes in TV.
Thank God it's only 5 days until the federal election. This time round seems to have been
particularly stupid, and the most important topic seems to be how bad the rival party is,
followed at some distance by the issue of refusing human rights to asylum seekers (they're
both in favour, but differ in their approach to punishing them). If I believed all the
propaganda material I receive, I'd have to put both major parties at the bottom of the list,
and the only problem I would have would be to the order. Labor ahead of the Liberals or behind
them?
Today I got no less than three letters from the two
candidates. Sarah Henderson sent
both an “Action Contract” and an “Action Plan”, with contents that
only marginally matched. The “Action Contract” promises to tackle climate
change and reduce CO2 emissions; the “Action Plan” outlines what she
would do if she were elected, including $200 million for better roads (making it easier to
emit CO2) and only $500,000 for the environment, all invested in third-party
projects. She even promises bushfire protection measures—for some parts of the
Surf Coast, presumably
because it's closest to her office, or because that's where the tourists go, and a total of
$200,000. As Edwin Groothuis put it, it's a Climate Real Action Plan, with an appropriate
acronym.
Darren Cheeseman was more
frugal: a picture of a little girl asking “Mum, why can't we have pizza
tonight?”. I thought of possible answers before opening: imbalanced diet seemed the
most likely. But no, it seems that Mum has fallen prey to her employment contract and was
no longer doing overtime and is thus earning less money. Although she now has time, she
still buys pizza rather than makes it. I would have thought that working normal hours and
having time for your children would be an advantage, but I suppose there's a reason why the
party is called “Labor”.
I'm noticing a strong correlation between the weather and my inclination to do any work in
the garden. CJ was supposed to come along this morning and help put up the fence around the
vegetable garden, but he didn't show. Spent the time doing some weeding and attending to
the greenhouse, which needs it. Finally made
the clips for the roof glass, and in principle we could now complete it, but somehow I was
feeling a little tired, and decided to take the afternoon off.
That wasn't to be: CJ showed up with Sue and came in with his tractor to put the posts in,
involving removing some of the low-hanging branches of
the Eucalypts. I had to finally
transplant the other Lilly Pilly, after
establishing that the first one, which I transplanted two weeks
ago seems to be happy enough where I have put it.
Just before the last post, CJ ran into trouble: a hydraulic leak that, after some
investigation, proved to be coming from what looks like a fissure in a pipe under the
driver's seat. It was enough to mean that he can't drive the tractor home, so he left it
here until he can come over and repair it. He's pretty busy the coming couple of weeks
anyway, so it looks like we'll have to wait before we can finish the vegetable garden.
While I was working in the garden, Chris and Yvonne arrived
with Zarzuela, who had been at Chris' place for the last couple of days. Yvonne put her in
the west paddock, with the other horses in the north paddock, but that didn't seem to please
Zarzuela. We had turned the electric fence off to do the garden work, which was right up
against the fence, and so Zarzuela just went through the fence, breaking off an old post in
the process. And we can't even repair it at the moment, because CJ's tractor is out of
service.
How the weather changes from day to day! Yesterday was mild, sunny and windstill, and we
did lots of things in the garden. Today was like last
weekend, but cooler and with higher winds:
mysql> select date, max(wind_speed), max(wind_gust), min(outside_temp), max(outside_temp) from observations where wind_gust > 39 group by date order by max(wind_gust); +------------+-----------------+----------------+-------------------+-------------------+
| date | max(wind_speed) | max(wind_gust) | min(outside_temp) | max(outside_temp) |
+------------+-----------------+----------------+-------------------+-------------------+
| 2010-08-14 | 19.7 | 39.7 | 9.1 | 12.3 |
| 2010-08-18 | 25.6 | 41.7 | 7.7 | 9.8 |
| 2010-07-10 | 23.6 | 48.1 | 10.9 | 11.9 |
+------------+-----------------+----------------+-------------------+-------------------+
Spent most of the day researching panorama hardware, in the process writing up a web page on the subject. And that's about all
I did.
Off to Melbourne this morning for a number of things, including buying food. Over to Chris'
to pick up her fridge. Now that we have the navigator as well, we need an adapter to
connect both to the cigarette lighter socket. I bought one a while back, but today was the
first time I needed it. Plugged the fridge in and... nothing happened.
I had received a notification that my latest laptop was waiting for me at the Post Office
in Sebastopol, so went to
Melbourne via Ballarat to pick it up, and also dropped in at the shop where I had bought the
adapter. After much messing around discovered that you need to really push the plug in with
some force. It seems to click before, but there's another couple of millimetres travel
before it actually clicks and makes contact.
In Melbourne first to MSY in Brooklyn, where I
bought another HDMI cable. 2 m cables cost $5; 3 m cables cost $20. On the other hand,
since my laser printer currently has paper feed problems, and since it's relatively
expensive in toner to use a colour printer to print the occasional black and white printout,
checked the price of cheap laser printers, and got a Samsung ML-1660 printer for $77. Somehow the price relationships are all crazy.
Then to Fleischer's, the
German butcher in Boronia,
where Yvonne had a field day. Last time I was there, with Chris, I spent about $25. Today we spent nearly 10 times
that much. We won't need to come back for some months. And this time almost everybody was
speaking German, including the other customers.
From there to the Banksia Garden Centre, where we bought a few plants, but unfortunately the
greenhouse and water plant section was closed today, so we didn't get much done. Did manage
to identify one plant:
On to Dandenong South to pick up
a “crate” for Nemo. It's really a
collapsible cage, but it seems that it's a good thing for dogs. I bought it on eBay for $100, while normally the prices are closer to
$250. I had had some trouble getting pickup instructions, but the pickup itself went simply
enough.
Then to Hindustan Imports to try to
return some Tali plates that I had bought a few months ago, and which we decided we didn't like. They couldn't take them back because it had been
so long, and the owner of the shop had just gone back to India for a month. They'll contact
us, though; nice of them, given that we have used them once. After that to Springvale for a late lunch and
to buy some vegetables, then headed off home.
We left Springvale at about 16:20, and had to head through the middle of Melbourne in the
rush hour. It wasn't as bad as I had feared, and we had crossed the West Gate Bridge by
16:50, and arrived in Werribee at 17:10.
But then things went to hell: long queues, and we had to find a petrol station (to which the
navigator conveniently misguided us). Looked for a different way out of Werribee, avoiding
the town centre, and finally got out at about 17:50. To do so I had to outsmart the
navigator, which wanted to take me down the most congested route; it looks as if the best
way is to select a “via” in the direction I want to go, rather than the
destination, and then remove the fake via point when I've got past the congestion.
Ultimately across country
via Anakie
and Meredith, arriving home at
the surprisingly good time 18:50, roughly as fast as you can go without breaking the speed
limit.
Also unpacked the “new” laptop (a Dell Inspiron 5100, because I have two similar
models already) that I had bought on eBay last
week. It was advertised as being extremely slow, which it was, so went into the BIOS to see
what was going on there. But I couldn't change the page: it looks as if the keyboard is
defective. That's not what the vendor told me. More investigations tomorrow.
Apart from the extremely undignified behaviour, it's clear that this is a coordinated action
by the Federal Liberal Party. Just insert
the names of the two cogscandidates for each electorate. Apart from that, the
accusations are just plain misplaced. None of the accusations relate to actions for
which Darren Cheeseman was
responsible. After all, to the best of my knowledge no illegal immigrants arrived on the
shores of Corangamite
during his tenure—using the Liberals' stupid logic, that should be good. About all
that this has achieved is to greatly annoy me and decide my voting order for tomorrow: the
Liberals come bottom. Wrote another letter to Sarah Henderson telling her so. A couple of weeks ago I was undecided, but now I really
hope that the Liberals lose.
The photo on the left is the “new” Inspiron 5100, and the one on the right my
old Inspiron 1100. Both have only 1 DIMM, but in the factory they insert the memory in slot
A first. That suggests that somebody has removed one of the DIMMs. In addition, they one
in the 5100 was only 128 MB. No wonder the machine was so slow.
Called up Dell's “Premium Service phone
support” on 1-800-054-429. It took them 54 minutes to answer, and when they did, I
was connected to sales. I wonder what a “regular” support line would be like.
Fortunately, they were able to connect me to support with only minimal waiting, and
they were able to tell me the configuration of the machine as delivered: 2 128 MB
DIMMs. So one of them has been removed; it's difficult for the vendor to claim he didn't
know why it's so slow. Put in the 256 MB DIMM from the Inspiron 1100, and it worked a lot
better.
The keyboard was another matter. Only about half of the keys worked. Connected a USB
keyboard to confirm that the problem was local to the keyboard, and then swapped keyboards
with the 1100, helped by the online service
manual which I couldn't find on the Dell web site, but which Google found with no
difficulty. About the only issue was removing the cover above the keyboard; the manual
didn't make it clear that it also covers the hinges:
After replacement, the keyboard worked—better. Other keys now didn't work. So what's
the problem? I had half suspected the cable connection, but it looks OK:
So I'm no further. What do I do next? The Inspiron 1100 was functional except for a
defective lid, so maybe I should put the display of the 5100 on it. But when I tried to
power it on this morning, it just powered off again. More investigation needed. Somehow
I'm getting the feeling that God doesn't want me to have a kitchen computer.
After that, turned my attention to the new laser printer. That was fairly straightforward,
but it didn't print anything. With the help of people on IRC, established that it doesn't
follow standards: instead it uses something called SPL (“Samsung Printer
Language”), a proprietary protocol so obscure that I can't even find a useful web link
for it. And of course I need special drivers,
and they're only available for CUPS,
software that drove me mad years ago. So I
connected it to pain.lemis.com, my Microsoft box, and installed the supplied drivers.
It works, it's even relatively fast, and one day I may take the plunge and install CUPs on a
FreeBSD box.
So I currently have the following partially defective Inspiron laptops:
My old Inspiron 5100, bought seven years ago,
currently running Microsoft “Windows”. Functional except for the battery
charge circuits, which make it less than ideal for carrying around.
An Inspiron 1150, bought nearly 6 years ago. For
some reason, I recently thought that this was an 1100. Functional except for the
damaged panel frame. The problems I had with this machine yesterday proved to be
because I had removed all memory.
The “new” Inspiron 5100, which has at least a defective keyboard and
keyboard controller. It's possible that the DVD drive or controller are also defective:
I had some problems with it yesterday. It also doesn't have a disk drive.
So, as I noted yesterday, the most obvious course of action would be to swap the displays
between the “new” 5100 and the 1150. Got started on that and ran into problems:
This is the area under the keyboards of the 5100 and the 1150 respectively. The display has
already been removed from the 5100, which is straightforward enough, but on the 1150 there's
no socket there, and the cable goes under the metal cover:
That's not what the manual shows, but it covers only the 5100, 5150 and 1100, and not the
1150. Looks like I will need to dismantle the displays and swap the panels. What a pain!
While taking the photos of the laptops, ran into strange problems with flash exposure. The
initial ones (above) worked fine, but then I took a photo of the two laptops together, and
got extremely underexposed images. Here one with normal exposure (left) and 2.3 EV
overexposure (right):
It seems to be related to the position of the flash head. The ones taken above were taken
with the head pointing down 15°, which is necessary for close-ups. After resetting it to
its default position, things looked very different (again normal exposure (left) and 2.3 EV
overexposure (right)):
A bit more work in the garden, and planted the two
surviving Abutilons from the ones we
propagated last November, and also replanted the unhappy
looking Fuchsia from the ex-Cathedral to
the corner of the new wind break area, where it will be out of all except early morning sun.
Also planted a Banksia
“Honeypot”—our first Banksia—in the dry eastern part of the garden.
Also a bit of work on the greenhouse, removing the roof gussets in preparation for putting
on the flap mechanism at the top.
Off to Sebastopol this
morning to take Nemo to the local chapter of the
German Shepherd Dog Club of Victoria to
get him started with tracking. That was a non-starter. Since finishing his puppy training
only a month ago, he has forgotten a lot of the material, and we had to start all over
again. Back home in no too happy a frame of mind, having made fools of ourselves and
achieved nothing.
I'm still trying not to have to rebuild the display panels on my Inspiron laptops. Further
investigation reveals that the “old” 5100 (the one I bought in July 2003) can charge the battery that came in the new 5100, just not
the new battery I bought a couple of months ago. Is there some issue with the battery? It
charges without problems in the 1150. Anyway, that means I could use the 1150 as a charger
(the laptops all only take one battery), use the old 5100 as a kitchen computer and one of
the others as a replacement for pain.lemis.com, which is always on mains power and
which I usually access via rdesktop, so the keyboard issue isn't serious. In
preparation, did a backup of the disk image of pain, which took all day.
The winter has been relatively cool and decidedly wet, but today it was neither. We had a
very mild frost on the paddocks, but it didn't reach the garden, and later in the day the
temperature was highest we've had all month, it was sunny and there was little wind. Took
advantage of the latter to spray weeds and also the grass in the ex-cathedral, where we're
planning to plant some birches. Also more work on the greenhouse, and finally got a section
of the roof flaps in place, after discovering that all the work we did yesterday had been
unneccessary.
That sag in the middle of the roof is not very reassuring; I suppose that's the purpose of
the arch gussets, of which I have far too few. More head-scratching.
Also planted the Alyogyne
huegelii “West Coast Gem” that we bought last Thursday, in front of the
garage, and planted tomato, basil and petunia seeds, which I'll keep in the bathroom
(coincidentally the brightest room of the house) until I finally get the greenhouse
finished. Also more weeding, which will probably keep me going for ever.
It's been a while since I set up photo albums for Piccola and Nemo, so spent some time bringing
them up to date. locate and grep are my friends: I was able to sift out the
photos with:
=== grog@dereel (/dev/ttyp6) ~/Photos 241 -> locate Piccola | grep big > /tmp/Piccola
That, along with a few Emacs commands, enabled me to build the page almost without
typing any data. Also created a similar page for Lilac; considering she's coming up for 14 years old, it's about time.
The next step in my Inspiron saga was to put the disk from pain (Inspiron 5100) into
the Inspiron 1150. It has two operating systems on it, and FreeBSD ran fine. Microsoft “Windows”
didn't. It spontaneously reset, and the second time around I was given an option to boot
into safe mode, which produced a list of what looked like driver names and then reset.
Tried the disk from the other 5100 (only Microsoft), and got similar results, though this
time it was a Blue Screen of Death instead of spontaneous reset. I wonder what's wrong
there. The Inspirons are strange in that the disk is on the secondary IDE controller, but
that's the same for all of them; otherwise FreeBSD would have had difficulty mounting the
root file system.
Gave up on that one. The good news was that the data on the disk wasn't damaged, and I
could still boot Microsoft on the original pain. Now it's back to replacing the
frame for the display on the 1150. Started on that, but didn't have enough time.
Spring is only a week away. The rose bush in the north garden bed has had a bud on it since
autumn, and I was hoping that it would hold out until the first day of spring, but warm
weather of the last couple of days has it ready to bloom:
So it seems that putting the arch gussets in the middle of the greenhouse should help cure
the sag of the roof. It did, though it still looks a little uneven. Here before and after:
Now I just need to put in some of the last screws, in the processing changing some for
flatter-headed ones, tighten them up, glaze the greenhouse and we're done! I was spared the
work of looking for an excuse not to do it immediately by the arrival of yet more rain.
This month is in the running to be the wettest August on record (since 1908) in Ballarat.
So far they've had 150 mm there (compared to “only” 113 here), and the
wettest
August (in 1909) had 167 mm.
I had just got into the office when Yvonne called from Chris'
place. Chris' brother Jonas and his wife Patrizia are going back to Germany for a couple of
years, and they've had a shipping container with the belongings they're not taking with them
delivered to Chris' place. And because of the extreme wetness, the truck had got bogged
down in the paddock. They had been trying to get him out for a couple of hours, and they
asked me to come over to help tow him out, Chris using a tractor and I their Landcruiser.
That didn't work, and Yvonne and I went looking for somebody with a bigger tractor. We
found that, but not the owner, so came back empty-handed.
Got back just in time to see Chris moving the Landcruiser, dragging along Nemo, who was tied to the front of it. Fortunately nothing went
wrong, but it brought home to us that if you must tie a dog to a vehicle, it should
be clearly visible from the driver's door.
After that, they tried again, and finally got the truck out; it must have taken 3 hours.
I've taken to updating my spam blacklists again, but in one case it didn't work. I've been
meaning to revise my Postfixmain.cf
file for some time, and finally got around to doing it—to find that a whole lot was
missing, including the entire smtpd_client_restrictions and
smtpd_recipient_restrictions. Fixed that up, and am now refusing invalid reverse
lookups as well (reject_unknown_client_hostname). It's amazing how much that
rejects. For the 24 hours to midnight UTC I get:
Normally the starter looks like the one in the first photo, but this one has something that
looks like worms on it. To the naked eye it looks surprisingly like powdered yeast, but the
details don't seem to match. It doesn't smell obviously bad, and maybe it is yeast, but I
need to investigate first. I haven't thrown it away, but I suspect I won't bake anything
with it.
The weather's cool and wet again, but we found time to do some work in the garden, putting
up most of the wire mesh for climbing plants on the new wind breaks, using up the fencing
mesh around the old vegetable garden in the process. Stopped for lunch after doing 3 of the
4 sections, and then it started to rain again...
The kangaroos have left us alone for a couple of days, but they're back again, and another
found its way through the garden, apparently without doing any harm:
Took another look at replacing the display panels in the Inspirons. I had every reason to
be cautious. The instructions for dismantling the assembly seemed to be out of order: the image was in
the section before the instructions, and it didn't quite match what I found:
In particular, the instructions show 4 screws on each side. Yes, there are 4 screws, but
only two of them are needed. The other two are mounted at right angles to the image in the
instructions, and they don't need to be removed to remove the panel. Here the right side of
the panel in detail:
The bracket at the left of centre does not attach to the panel, and the screws don't need to
be removed. The two screws at the end have been removed, so the panel can be
removed. This image also shows the broken hinge at the right.
More to the point, though: how do you separate the bezel from the top cover? The
instructions say:
Starting at the bottom of the display panel, use your fingers to separate the bezel from
the top cover and lift the inside edge of the bezel away from the top cover.
The alternative is to insert a small screwdriver, but that has the danger that it will
damage the surface. But there's nowhere to put your fingers. After a bit of
experimentation (after all, I have a damaged display cover to experiment with), discovered
where the clips are located:
They're the little E-shaped things on the top edge. After releasing one clip at the bottom,
I was able to prise up the bezel enough to insert a screwdriver in the position of the next
clip and not do any damage. After all that, finally got a machine together that had all the
components I need. In passing, it's worth noting that the display panels are more different
than I expected. In particular, the cables are different. Here the reverse of the panels
for the Inspiron 5100 (left) and the 1150 (right):
Clearly they're not interchangeable. But there's no reason to believe that the panel of the
1150 will die before the panel of my other 5100, so it could still be of use.
I've already commented in a number of places on
the fact that the Golden Plains Miner newspaper didn't publish my letter about the
Dereel mobile phone tower, most recently (3 days
ago) on the dereel.com.au web site. The Miner arrived today, and my letter was in it,
in its entirety apart from some changes in format and typesetting breakage that partially
obscured Wendy McClelland's name. So: are they just slow, or is this a reaction to the post
on dereel.com.au? I'll give them the benefit of the doubt, but it's surprising
that it's taken them 3 weeks to publish a letter.
Yvonne off shopping this morning, but she called soon after
to say that the car had broken down: “The lights all went on on the dashboard, and the
battery has failed.” Well, I've heard this before,
and so I asked if she had enough petrol. No, it can't be petrol, she still had 30 km to
empty. So off to Enfield with
a canister of petrol and got her on her way again. Also got a promise that she would fill
up the tank when the computer showed less than 100 km to empty.
Yvonne came back with a number of climbing plants: a
Solanumjasminoides, a Jasminummesnyi and
a Clematiscirrhosa.
The latter appears to be the “Freckles” cultivar, though the label doesn't say
so. In the afternoon put up the remaining length of wire mesh on the wind break and planted
the creepers, also two of the three cuttings of an
unspecified Lonicera that we got from
Nancy Brewer last spring, and some
cuttings that we had been growing from
the Jasminum polyanthum and
Lonicera that we had planted on the north side of the verandah. I had left some of the shoots to grow along the ground, where they readily
set root. The jasmine in particular forms very strong roots about 20 cm apart and grows
very long:
This section must be over 2 metres long, long enough for one side of the wind break, and
the roots were so long that we had to break them off to get them out of the ground. I
wouldn't have expected it would be that easy to clone 10 jasmine plants.
The disk that came with the “new” Inspiron 5100 had a bad sector, and on
examination I discovered a 60 GB Fujitsu drive that I had used in an earlier attempt to get
the kitchen computer running, so used that. It has plenty of space for two operating
systems, so started with Ubuntu in the hope
that it might handle ACPI better than FreeBSD.
How I hate Ubuntu networking! It insists on giving me a DHCP lease, and
unfortunately I still have my satellite modem set to deliver them, so I had to change
the /etc/network/interfaces file. I know enough Linux to do that, so did it, and
rebooted. And for some reason the system could no longer find the firmware file. Even when
I reinstated the original file, it didn't work properly. My best guess is that this
horrible GUI-based configuration, which doesn't even display text properly in the default
configuration, has kept its own records which it won't reveal to me. It might be better to
start the installation again with no DHCP server running; but that's a real cop-out.
So I moved on to install FreeBSD. In the course of time, it became clear that I haven't
been keeping records well enough. After installing FreeBSD, I had to modify this horrible
GRUB boot manager to recognize
FreeBSD. I wrote of the pain last month, and the
information there greatly helped today. But then I needed to update the /boot/grub
hierarchy. I found out how to do that last month, but I forgot to write it down. Today I
went through all the documentation I remembered, but I couldn't find out the canonical way
of doing it. Looked through the man pages, but that didn't help much. They're both very
short and just plain wrong:
NAME
grub-install - manual page for grub-install (GNU GRUB )
SYNOPSIS
grub-reboot [OPTION] entry
In the end used gub-mkconfig, but that requires an explicit output file, and I'm sure
I didn't do it that way last time. After that I was able to boot FreeBSD, but didn't get
very far: inserting the WLAN card caused a panic. I've seen that before, too. Put off further work for tomorrow.
Yvonne woke up this morning with a very painful right thumb
joint, possibly as a result of yesterday's exertion in the garden, and we decided that she
should go and see a doctor. But where? The Eureka Medical Centre is such a pain. For some
time I've been planning to visit MedicAid (what a silly name, given
the US Medicaid programme). David
Yeardley told me that they would give appointments from people out of town, but when Yvonne
called, they didn't want to know. So off she went to Eureka, which took her a total of 4½
hours. Why can't these clinics make appointments?
And the result? Arthritis, it seems, though I have my doubts about how thorough the doctor
was. He prescribed anti-inflammatories, not the most harmless of medicine, and didn't give
Yvonne any advice or warnings at all. Fortunately the product
(Celebrex) seems to be one of the less
aggressive ones.
The weather's still wet, and I didn't get much done in the garden. Finally spread some
fertilizer in the main eastern bed, getting through about 5 kg in the process, and also
tidied up the creepers round the new wind breaks. Also tried to save
a Salvia that I had planted in that area
and then trodden on, breaking the stem close to the ground. It's not beyond hope that it
might recover, but it won't be needing what branches it had, so planted them in potting mix.
That's how we got this plant, so hopefully it will strike again.
More playing around with the kitchen computer. Upgraded the FreeBSD system to the latest 8.1-STABLE, which took most
of the day, and then looked at the issues of swap and kernel dumps. Since I have a Linux
swap partition on the disk, it makes sense to use it for swap for FreeBSD as well, and that
worked fine. But it didn't work as well for the dump device:
# dumpon /dev/ad2s2 GEOM_PART: Partition'ad2s2' not suitable for kernel dumps (wrong type?)
Apart from this horrible message framed as a question, that makes sense in many cases: you
don't want a UFS file system as a dump device. But in this case it was annoying. Maybe I
should fix dumpon to accept Linux swap partitions. As it was, decided to change the
partition type to a FreeBSD slice (165/0xa5) and create a swap partition in that
slice. And that failed with some unspecific error, perhaps because it was already in use as
swap space.
Rebooted the system from the installation disk and tried it like that. That worked, but
when I tried to reboot, GRUB was no
longer able to boot it. Why? The partitioning and the contents of the FreeBSD root file
system hadn't changed. It looks as if I might have to reinstall Ubuntu anyway, so decided just to install the
FreeBSD boot0 boot manager in the partition table. That worked, and I was able to
reboot FreeBSD normally again. And, to my immense surprise, I could also boot Linux: GRUB
must leave something in the partition after all. Maybe it was just as a result of my
attempt to do so earlier, but that claimed to have failed.
So, finally, I was in a position to insert my WLAN card and see what happened. A
panic, of course, but I was able to get a dump and look at it. For the first time ever, I
had a dump that was smaller than the kernel: this machine has 384 MB of memory, and the
compressed dump came out at only a little over 40 MB, and a couple of MB smaller than the
kernel with debug symbols. The crash was in wi_info_intr at line 1511, which
doesn't look very plausible:
That meant that one of the previous series is pretty much obsolete: the first photo
incorporates not only the view, but also the standpoint from which I used to take it (the
rightmost post of the verandah). Here's last week's photo:
Things didn't go as smoothly as planned, and I ended up having to repeat a couple of
panoramas. All in all I ended up with something of a record: 343 photos, taking up nearly 6
GB of original images and another gigabyte of generated images—7070 MB in total. A
good thing I bought a new CF card recently; until then, I had only ever used more than 4 GB.
Before I got any panoramas, though, I was surprised by a completely unexpected problem. My
normal processing takes two or three images with different exposures and merges them
with align_image_stack, then uses hugin to stitch them together. Today I ran the script, selected the files to
load, and pressed “Open”. And hugin crashed with a SIGSEGV. Why? I
spent hours investigating, but I still don't know.
The last time I used hugin was to look at some settings. I didn't actually load any
files. Did I change anything in the settings that might cause the problem? Restored
the ~/.hugin file from a couple of weeks ago, but it didn't make any difference. So
it looked like I needed a debugger. And for that, I needed to build it with debug symbols,
something I've done before with some pain. But
today it took a long time to get that far. First the port wanted to rebuild various
packages, and with libglut-7.4.4 I ran into dependency problems:
checking for DRIGL... configure: error: Package requirements (x11 xext xxf86vm xdamage xfixes x11-xcb xcb-glx) were not met:
No package 'x11-xcb' found
No package 'xcb-glx' found
That looked like missing dependencies in libglut-7.4.4, and so the first step is to
install it manually. But where are x11-xcb and xcb-glx? Nowhere. The
only files I found with x11-xcb in their name was in a backup of cvr2, a
Linux machine, and I found no file at all with the name xcb-glx. Searching
for xcb was more fruitful, and it seems that the port x11/libxcb satisfies
x11-xcb. But try as I might, I couldn't find what xcb-glx was supposed to
mean. Neither of these texts shows up in any of the pkg-plist files in the Ports
Collection. Why have the names been chosen? How does anybody expect people to find them?
In desperation, ran the build through ktrace, which showed the convolution of the
build process. In /usr/ports/graphics/libGL/bsd.mesalib.mk found the following
lines:
Success! I was finally able to build libglut-7.4.4. Then back to
building hugin, which also worked. Fired it up, worried that it, too, might crash
when I opened a file. But it didn't. It crashed much earlier, before the initial screen:
Fatal Error: Fatal installation error
The file data/splash.png was not found
at:/usr/local/share/hugin/xrc/
Strange. Went looking, and sure enough, there was a
file /usr/local/share/hugin/xrc/data/splash.png. Why did it say it wasn't
there? ktrace to the rescue again, and found:
45615 hugin CALL open(0x8531070c,O_RDONLY,<unused>0x1b6)
45615 hugin NAMI "/usr/local/share/hugin/xrc/data/splash.png"
45615 hugin RET open 8
...
45615 hugin CALL write(0x2,0xbfbfd790,0x6c)
45615 hugin GIO fd 2 wrote 108 bytes
"Fatal Error: Fatal installation error
The file data/splash.png was not found
at:/usr/local/share/hugin/xrc/"
Considering my options, decided to install hugin on cvr2, the video recorder
running Ubuntu Linux. To my surprise, it
installed a much older version of hugin, 0.7.0, nearly two years old. There have
been three releases since then, and this one doesn't have the panomatic control point
detector. Started it anyway, and it ran for a while, then crashed without a word. On
reflection, some hugin components need lots of memory—gigabytes of
it—and cvr2 doesn't have much memory. Off to the web site to look for binaries, but they don't do
that for systems which provide their own. But there's a newer version of hugin
there. The FreeBSD port is version 2009.4.0, and the current version is 2010.0.0.
Downloaded that and frobbed the port to build it. First problem was in the
file src/foreign/vigra/vigra_impex/png.cxx, which used setjmp
and longjmp with jump buffers in a png_structp, part of libpn. The
version I have defines, in png.h:
struct png_struct_def
{
#ifdef PNG_SETJMP_SUPPORTED
jmp_buf jmpbuf PNG_DEPSTRUCT; /* used in png_error */
png_longjmp_ptr longjmp_fn PNG_DEPSTRUCT;/* setjmp non-local goto
function. */
#endif
png_error_ptr error_fn PNG_DEPSTRUCT; /* function for printing
errors and aborting */
...
And png.cxx doesn't define PNG_SETJMP_SUPPORTED. Fixed that, and also
discovered I still need to apply the other patch that was already in the port, and finally
got the port built and installed. What happened?
Fatal Error: Fatal installation error
The file data/splash.png was not found
at:/usr/local/share/hugin/xrc/
More debugging. This time I had debug symbols by default, and was able to find the place
where it happened pretty quickly, in the function MainFrame:
if (bitmap.LoadFile(huginApp::Get()->GetXRCPath() + wxT("data/splash.png"), wxBITMAP_TYPE_PNG))
{
...
} else {
wxLogFatalError(_("Fatal installation error\nThe file data/splash.png was not found at:") + huginApp::Get()->GetXRCPath());
abort();
}
Where's bitmap.LoadFile()? No idea. The variable bitmap had been
optimized out, and I had to single step to find out where I landed, in some library that
opened the file and then for some reason returned 0. This is all PNG, which I don't use, so
just commented it out:
/* if (bitmap.LoadFile(huginApp::Get()->GetXRCPath() + wxT("data/splash.png"), wxBITMAP_TYPE_PNG)) */
bitmap.LoadFile(huginApp::Get()->GetXRCPath() + wxT("data/splash.png"), wxBITMAP_TYPE_PNG);
if (1)
{
There seems to really be some issue with PNG images. On starting, I got an error message,
but I was able to ignore it and finally build my panoramas. But what caused the
problem? I still don't know, but my installation is nearly a year old, and nowadays that
seems to be too old. I'll have to start upgrading, but I don't want to do it under
pressure.
CJ popped over with a bloke he knows, Phil Higgins. Phil's son-in-law is building a house
round the corner, and Phil has a bobcat, which we used to pull out the stumps in the
cathedral, and also a couple of peach trees:
Today was the last weekend of the month, when I've taken upon myself to take photos of the flowers in the garden. Once again, to my surprise,
there's a lot of stuff in the garden, though the survivors from autumn are looking less than
happy, and most of the spring stuff has yet to take off:
Now that the stumps are out of the ex-cathedral, our plan is to turn it into a grove of birches with a couple of
Australian native plants on the edges. Today made the first step and planted about half of
them. They're all pretty tiny, but the position is recognizable (at least in the larger
versions of the images) from the kangaroo protection mesh we've put around them. Here
before and after:
In the middle at the back is
a Podocarpus elatus
(Illawarra plum), and on the extreme
right is an Araucaria
bidwillii (Bunya Bunya pine). Both can grow to 40 m, but they won't do it in our
lifetime. The birches are self-sown from the two in the east garden, which are clearly male
and female. They're only about 30 cm high, but they're looking quite happy with themselves
for the time of year:
These photos were taken 5 seconds apart with identical exposure parameters, but the first
one is unrecoverably overexposed. It must be the flash unit (Mecablitz 58 AF-1 O digital), but why? Is the unit defective?
That wasn't the only strangeness. Juha Kupiainen has a new (well, to him anyway) 400mm
f/5.6 Sigma telephoto lens for his Pentax K100, and today he went out taking photos, some of
which looked less than perfect:
We spent some time discussing the matter on IRC. The funny thing is that the aberration is
over the entire field of view, while typically it occurs in the corners. Andy S voiced the
opinion that this was due to problems using lenses (like this one) which were designed for
film cameras on digital cameras, as light coming at extreme angles relative to the lens axis
can accidentally trigger the wrong receptors on the sensor. While possible, that seems
unlikely to me, at least with a long telephoto lens with a correspondingly small angle of
view. Looking at the Wikipedia article, this could be axial chromatic aberration. The real
issue is: what's causing it? Is this as good as the lens is at full aperture? I took some
test photos with various lenses in one of my
comparisons. Here's a pretty ordinary Hanimex 300mm f/5.5 lens at full aperture:
At the time I came to the conclusion that it wasn't a very good lens, but it's an order of
magnitude better than the results that Juha got with his Sigma. Clearly it'll be
interesting to see what happens when he stops it down, but I suspect there may be something
mechanically wrong with the lens. I'll follow developments with interest.
The weather's gradually getting warmer, and it brings home to me how cool the entire winter
has been. Today was the first day where the outside temperature hit 19° since the beginning
of winter, and we felt more motivated to work in the garden. Planted another 5 birches in
the cathedral area, and then on to work on
the greenhouse replacing hex head bolts with
screws and adding the glass clips for the roof until I ran out of screws.
I hate working on this greenhouse, and today it became apparent why: it's all a kludge, and
I'm continually repeating work and wondering whether what I'm doing is right. I've had four
attempts at putting up the roof trusses: the first
time the wrong way round (the top mounted at the bottom), the second time with hex head nuts that were too thick to
put the glass on top, the third time to replace the nuts with thinner screws, and now to put
in the mounting clips for the roof glass. These screws are horrible! Turn the (electric)
screwdriver too fast and it tears the heads off. I had to drill one out and replace it.
More thinking about Juha's chromatic aberration yesterday, and dragged out a couple of 300
mm lenses, conveniently the same angle of view on my E-30 as the 400 mm lens had on
Juha's K100D, and took photos of one corner of the verandah, which gives higher contrast
than the others I took.
The results were interesting. The sharpness is a little complicated by issues of focus,
camera shake and position, but there are some general things
The corners
These photos are the top left corner of the image, taken with the Hanimex 300 mm f/5.5
(left) and the Zuiko Digital ED
70-300mm F4.0-5.6 (right), at f/5.6, f/8, f/11, f/16 and f/22:
The tendril of the vine is blue in all the Zuiko images, but is it chromatic aberration?
The beam above doesn't show any colour fringes. There's a much more prominent blue fringe
along the stem of the vine in the Hanimex photos. On both lenses, the effect doesn't change
much with aperture.