Yvonne's birthday today, and she wanted a fondue de fromage, so decided to try baking a sourdough
white (wheat) bread, the first time I've done that. It's amazing how different wheat dough
is from rye dough. Put it in the oven to rise, and covered it with a plate. Not a good
idea:
It had also become very moist; I had the feeling that I would be able to pour the dough out
of the form. Scraped the stuff back and was able to bake a reasonably good loaf; next time
I'll use less water.
Fondue itself was good—very
old Gruyère and
some Appenzell. We found a
surprising number of opened bottles of wine with only a little in the bottom, which we got
through. I think that's the end of the Rosemount.
More investigation and discussion of the HSS flash mode today. Somebody pointed out that
the instruction manual for the Mecablitz 58 AF-1 O digital (section 12) states:
If a camera with a between-the-lens shutter (see the camera's operating instructions) or
HSS synchronisation (see 15.4) is used, flash sync speed is not controlled
automatically. As a result, the flash can be used at all shutter speeds. If you need the
full light output of the flash unit, you should not select a shutter speed that is any
faster than 1/125 sec.
Why 1/125s? The camera can synchronize at up to 1/250s. Tried it out, and sure enough, the
guide number rose from 28 to 52—more than doubling. Found also that it didn't
decrease in the manner I expected:
Speed
Max distance
Guide number
(1/)
(f/4)
(200/24° ISO)
125
13
52
160
11
44
200
9.2
36.6
250
6.9
27.6
320
5.8
23.4
400
4.6
19.6
500
3.4
13.6
640
2.9
11.6
800
2.3
9.2
1000
1.7
6.8
1250
1.4
5.6
1600
1.2
4.8
2000
1.2
4.8
2500
1.2
4.8
3200
1.2
4.8
4000
1.2
4.8
5000
1.2
4.8
6400
1.2
4.8
8000
1.2
4.8
That probably has some relationship to the duration of the individual flashes and the manner
in which they're performed. And in any case, the manual is wrong, not just here. In normal
mode, you can get the full power at 1/250s—or at least, the unit promises it.
So why the slow shutter speed? It means that half the power of the flash occurs after the
shutter has closed.
Downloaded the manuals for the Olympus flashes and found a similar thing. Again, the
maximum power is only available at 1/125s, though the guide numbers decrease in a different
manner.
Later Reinhard Wagner came up with an explanation: the full power flash duration is 1/125s.
The Olympus FL-50R has a full power duration of 1/500s. But that just poses more questions:
why the 1/125 with Olympus?
I had intended to brew today, but I suffered from a massive attack of disinterest, and
finally found an excuse to put off until tomorrow: I needed to reinforce my starter. I
wonder whether I should give up brewing; I really need a lot of effort to persuade myself to
brew.
The kangaroos are back! They haven't done much harm—yet—but I found tell-tale
signs: a broken branch on one of the Salvias, and chewed-off branches
(right) on the Salix melanostachys:
Spent more time looking at flash today. If the Mecablitz 58 AF-1 O digital requires 1/125s for a full-power flash, how can it
deliver its full guide number at 1/250s? Took some test photos of a forgettable subject and
came up with the obvious answer: it can't. Here are histograms taken of four photos: the
first two were taken at 1/250s and 1/500s at f/22 and a distance of just shy of 4 metres.
The guide number at this speed should be 81, so this should have required exactly full
flash:
It's clear that the first one is noticeably darker than the second: the unit can't deliver.
In fact, it's also a little darker than I'd expect, and at f/8 (well within the range of the
unit), I got different results. Here there's effectively no difference between 1/250s and
1/125s:
So: the Mecablitz doesn't deliver what it promises, at least for cameras with flash sync
above 1/125s. It doesn't warn you, either: it shows the same maximum distance for 1/250s as
for 1/125s. It also doesn't seem to quite deliver at 1/125s. I now have the manuals for
this flash (obviously, but there are at least two different versions with different page
numbers, and many of the page references in my printed copy are wrong), the Olympus
FL-50R and the Olympus
FL-36R. All give detailed information on guide numbers (I had thought the Mecablitz
didn't, but that was just because the page reference was off by 10 pages). I need to look
at them more carefully so that I can understand what's going on. It's interesting to note
that the FL-36R, the weakest of the units, can exceed the power of the Mecablitz, the
strongest of the units, in some cases.
Understanding flash parameters
To understand why that is, you first need to understand just how high-speed sync works. And
I'm having trouble with that. How do you determine the highest shutter speed for electronic
flash and focal plane shutters? I've demonstrated one orientation point: the shutter speed
must be at least as long as the duration of the flash. Another is the highest speed at
which the entire sensor is visible at one time (in other words, the speed of traversal of
the blinds). But that's conservative: with that speed, you can illuminate the entire sensor
with a single infinitesimally short flash. Real world flashes light for a real period of
time—the Mecablitz between 1/125 and 1/33000s, according to the manual. I suppose
1/33000s is close enough to infinitesimal that we can discount the difference. But it would
be interesting to measure the real speed.
That's not as difficult as it seems: run, say, the Mecablitz at minimum power and flash
duration, set a shutter speed higher than 1/250s, and see how wide the slit is. Simple,
isn't it?
Well, it would be, if Olympus and Metz hadn't both put obstacles in my way. When the
Mecablitz is mounted on the camera, there's no way to select a speed faster than 1/250s. My
camera has a cable connection, but the Mecablitz doesn't. I have an extension shoe with
cable connection, but no flash cable. That's a thing that you used to find on every street
corner. Went to eBay and found almost only
cables with a flash connector on one end and a 2.5 mm jack on the other, like I already
have. What happened to flash cables? Checked my wireless remote unit, but the receiver
seems to be partially damaged: it has a shoe connector and also a flash cable connector.
The former doesn't work, but the latter does, as I checked with my studio flashes. But then
we're back to square one: no way to connect to the Mecablitz with a cable. I'd be happy to
buy a cable, but the only one I could find was only 30 cm long, and if I buy something like
this, I'd like to have it much longer so I could mount the unit remotely.
Yes, I know about the wireless remote flash. Everybody loves it—except me. It
somehow seems to be a deliberately limited kludge.
In the end, gave up and used the studio flashes. To my surprise, even on the weakest power,
the flash duration of the studio flashes appears to be relatively long, as shown by these
images taken at 1/500s and 1/1000s:
There are two things to note here: firstly, the light strip isn't sharply defined. At a
guess I'd say the transition is about 1/5000s, which is probably in the order of the flash
duration. The other thing is that the strip at 1/1000s is much less than half the width of
the strip at 1/500s, suggesting either that the shutter speeds aren't accurate, or that the
flash starts some time before the first blind opens. I also took one at 1/2000s, which came
out completely black, so I assume that it's the time the flash starts. This also suggests
that the shutter traversal time is considerably less than 4 ms (1/250s). Clearly I'll need
to find a way to repeat the experiment with the Mecablitz. And that means finding a flash
cable.
Light meters: then and now
One of my first photographic acquisitions was a selenium cell exposure meter, in about 1964.
Prior to that I had used guesswork, helped by rules of thumb supplied with each film:
something like “Set the shutter speed to the ASA number, and then for bright sunlight
use f/16, for very bright sunlight use f/22, overcast between f/5.6 and f/8” and so
on. I didn't use the exposure meter much: a year later I got a Pentax SV with clip-on
meter, and the year after that
a Pentax Spotmatic. No
guesswork, no extra light meter. And it's been that way ever since.
But there are still reasons for light meters, especially with close-up and studio work, or
to try to out-guess flash units. And surely they can't be very expensive. So while looking
for cables on eBay, also looked for exposure meters.
I didn't find any cheap exposure meters. It seems that only Sekonic makes them any more, and they cost hundreds of
dollars—it would be cheaper to buy a cheap digital camera and take its word for it.
But there are plenty of (incident) light meters. The rather inconsequential difference is
that they only display the light level: they don't calculate shutter speeds and apertures
for you. I don't understand why not: that's always been trivial, and nowadays it's
especially so. I suppose the issue is that nobody but an absolute professional needs an
exposure meter any more.
Did some investigation and found formulae for
converting lux into exposure values. The
Light meter page on Wikipedia
gives the formula:
where N is the aperture, t is the shutter speed (in seconds),
E is the illuminance in lux, S is the ISO linear sensitivity,
and C is the incident-light meter calibration constant, typically 320 or 330.
Getting back to my rule of thumb of the early 1960s, where the shutter is equal to the ASA
(ISO linear) sensitivity, I can factor out t and S and have the
formula N = √ (E / 330) to give me my apertures.
But Sekonic have done it for me, sort of. The instruction manual for the
Sekonic L-308S includes on page 19 a table converting lux into EV, giving equal value
to the constants 320 and 330: EV 7 is 320 lux, EV 17 is 330,000 lux. At least it makes it
clear that I needed a meter that would measure to 200,000 lux (the maximum on offer);
cheaper models only measure to 50,000 or 100,000 lux.
And flash measurements? Sekonic is very clear about what their meters do, but the others
are quite vague; they mention the term “flash” in the item description, but it's
not clear how it should work. The one I finally chose also has a “maximum”
value. We'll see how much use that is.
40 years ago today I moved into my first house, Comp Cottage in Bow, Devon. I was at the
University of Exeter at the time, and as a
foreign student with no family in the UK, I had nowhere to stay during the vacations, so I
had to find more permanent accommodation. How times have changed! In view of my interest
in computers, the name of the cottage was amusing, but even more so in conjunction with the
post code EX17 6HQ. When I joined DECUS, they
entered my address as “Comp. Cottage ... EX17 GHQ”. I wonder what mental image
they had of the place.
Fire!
Just when we thought the bushfire danger was over, large quantities of smoke appeared from
the north in mid-morning. Called up the Bushfire Information line (1-800-240-667) and
received a recorded message telling me to call 000 to report a fire. Clearly there was a
fire somewhere, so did that.
There I was told fairly quickly that there was burn-off activity in Enfield State
Forest—during the fire danger season, and 415 ha of it. Asked why this was, and was
told “That's done by the DSE. They
don't confide in us”. Somehow I get the feeling that, though the fires may be under
control, the people who control them aren't.
More animal activity in the garden overnight. I thought I had heard something, and a couple
of times shone a torch out into the garden, but didn't see anything. This morning, though,
found the cage around the Salix
melanostachys had been knocked over, though no obvious damage had been done, but
something had left a number of scrape marks on the ground, and it had attacked the tomatoes:
Finally I ran out of excuses not to brew, so today was the day. To make up for the dislike
of brewing, I brewed more than ever before—over 15 kg malt, yielding over 80 litres of
wort. I'll ferment half of it at a time. Things went relatively smoothly; the sparge took
longer than usual, presumably because of the bed depth, but it didn't stick. Then, just as
I was about to pitch, the temperature sensor circuitry failed—completely! No idea
what's wrong. It was merrily reporting temperatures via cu, which I stopped. When I
started tempcontrol, it timed
out. It wasn't tempcontrol itself; cu showed exactly the same behaviour. How
can that happen? I'm still baffled, but now I'm flying without instruments. Quickly put
together a program to turn the heater and cooler on and off manually. I could just have
plugged them into power points, but it only took about 5 minutes, possibly less than it
would have done to mess around with the wiring.
Question on a forum today: which software should you choose
for HDR image processing? I answered with my
experience with enfuse, but
there were others, notably a package with the unpronounceable name qtpfsgui. Took a look at that, and discovered
it had to load a behemoth tarball qt-everywhere-opensource-src-4.6.1.tar.gz, 160 MB
of it. Also had the usual problems with wrong libtool versions, but finally got it
installed. Ran the program—has to be a GUI, of course—and got as far as
selecting the images from a menu before it SIGSEGVd. Installed on zaphod,
currently a Ubuntu test box, and got further:
it didn't SIGSEGV until it tried to save the image. In that brief time (made much
longer by network speeds), confirmed that it did seem to do something useful, though of
course it's a lot more complicated than enfuse. Still, photography is one area where
a GUI makes a lot of sense. Pity it's so unreliable. Should I look inside? 200-odd MB of
download suggest “No”.
Into town shopping today, without finding anything much of use. I had hoped to find some
clips for the gas lines for my kegs, but they didn't have anything small enough. Also to
Midland Irrigation to look for a surge suppressor for the bore pump, but though they were
nearly empty, they also didn't seem to have any staff on hand, so left there too. About the
only thing of use that I bought was some wire and a couple of spray units for the garden.
Back home, tried out the spray units. We already have a big one, but just setting it up is
a hurdle, and it's too heavy for Yvonne. We had a 2 litre
one, which proved to be identical to one we already had, so we can take that back. The
other was a 5 litre unit from ALDI, which on
the face of it looked quite good. But the wand is too short, and the valve doesn't cut off
properly, so that'll go back too. Instead did some spraying with the back pack unit, and
did most of the garden.
Limited perspective: breeding ground for bad language?
Had a discussion on IRC the other day, talking about baking bread, as one does. Talked
about wheat (carrying on from my experiments of the other
day), and one of the participants mentioned baking with “white” instead of
wheat. What's “white”? I asked and was told to ask the supermarket. I've
already commented on the nonsense the
Australian supermarkets produce—reinforced by another package of spring onions I saw
today, marked “shallots
(eschalots)”. And people think I should rely on the supermarkets for
terminology.
It proved that “white” is “white flour”, which, in Australia at any
rate, is wheat. So what's wheat? Maybe the new supermarket name for maize? But that's
“universally” (in other words, in all places where the nay-sayers have heard it)
called “corn” nowadays, a name
that once meant “grain”.
Of course, there are plenty of English-speaking people who call maize
“sweet corn” or
“Indian corn”. But if
you avoid them, you can say “corn” and be understood. You just need an
adequately small sample of people from the same limited environment, preferably with limited
intellect, and there's no problem at all.
To support my argument, I quoted the Oxford English
Dictionary, which probably has the most balanced and even description that I know of
English words. But that means it includes terms that people don't know (indeed, it's full
of them, but with indications of their status), and so many people reject it because it
doesn't match their (limited) experience: “If grO0gle was a Christian he'd be claiming
the KJV is the One True Version? And OED is the One True English?”. And this after I
had pointed out that the OED is descriptive, not proscriptive—possibly words that
people don't understand.
And that, I suspect, is the crux of the problem. This nonsense renaming directories to
“folders” comes from this
same lack of understanding, a local view of things. Yes, language evolution is a natural
thing. It's unlikely that the term “folder” will die out, inappropriate as it
is, and probably in a few years “corn” will really mean “maize”.
But intelligibility is also important, and it's unlikely that English usage will diverge
significantly, so I doubt that the name “wombok” will become readily understood in
the English-speaking world, and it's as good as impossible that “white” will
come to mean “wheat”.
It's interesting to consider that this is just the tip of the iceberg. This discussion was
among intelligent people, and still they didn't recognize the narrowness of their viewpoint.
What do the unwashed masses do? I suppose TV advertising and sitcoms will increasingly
shape the language of the future—the latter particularly, since they tend to be
broadcast round the world in most English-speaking countries. It's probably already
happening, but I don't watch TV advertisements or sitcoms.
More work investigating the problems with my temperature control hardware today, not helped by the hot, rainy weather. The
machine I have there is no longer the old Intel 486, but it's also a little suspect, so
dragged another machine in there to see if it could communicate with the temperature
sensors. No. Looks like it's the temperature logger board. I should have another kit
somewhere, but where?
Chris over in the afternoon to pick up some MPEGs I've recorded for her, and ended up
staying to dinner. While preparing dinner, Yvonne somehow
managed to break a container of hop pellets in the deep freeze, spreading pellets
everywhere.
Chris helped me salvage what I could—in fact, most of them—in the process
exhausting the battery of one of our rechargeable torches.
That wasn't the only thing that failed today. When going to bed, Yvonne tried to turn on
the light at the head of the bed—and the switch failed. It's a cheap unit, and we can
replace it, but we have to go to Melbourne for that. And then she heard the water pump
running continuously. Out to investigate—of course the other torch went flat in the
process—but it really looks like we have a leak in the system somewhere. Damn. That
could be expensive.
Ten years ago I started at Linuxcare. What a time it's been! Somehow the dot com bubble, of which Linuxcare
was part, marked a turning point in the computer industry: it moved from being
technologically to financially motivated. I don't think it has improved things. As I
commented a while back, I haven't found any real
innovation in the last 8 years. Yes, there has been significant progress, but how much of
it is new?
Woke up in the middle of the night with the recognition that the water leak might not be as
sinister as it seemed. The pump cut in and out very frequently, suggesting that the
pressure cell might be lacking pressure. Checked, and sure enough, it was.
Pumped it up again and tried it out. Same problem. Further investigation showed that it
was most definitely a leak: a few minutes after turning off the pump, we had no pressure at
all in the water system. Damn! And of course, it had to happen at the beginning of a long
weekend. Did some walking around to look at where water might be escaping, but it's made no
easier by the fact that we had over 14 mm rain overnight. Looks like we'll be effectively
without water for a few days.
Looked at the wort that I had put into sterilized containers the other day. They're
fermenting! I had had large quantities
of sodium hypochlorite in
there for some time, but they've started fermenting almost immediately. It looks very much
like yeast, not some other organism, but to be on the safe side I should put in some of the
wort from the fermenter.
Playing around with the house photos today; HDR imaging can possibly solve some of my
problems with the shadows in sunlight. Ended up taking a ridiculous number of photos,
nearly 100 in total, bracketed at +2EV and -2EV, and merging them.
The first thing that's very clear: you must do this on a tripod, at least if you're
using align_image_stack. It can't handle the slightest differences. The following
image (right) looks like it has a really bad case of camera shake, though it was made from
the other two images. Maybe the parallax has something to do with it.
Otherwise, the results certainly show more even lighting. Here a partial image of my north
panorama. Clearly the trees look better, and there's more detail in the shadows. But the
detail in the sky looks quite unnatural:
Of course, there are two ways to make the panoramas themselves: make three, one with +2EV,
one normal, and one with -2EV, and merge them. Or merge the pre-merged images; the problem
there is that the stitcher doesn't like component images with different exposure. I'll try
both alternatives, but it takes time.
These multiple exposures make one other thing clear: I took over 100 photos today, most of
them hardly worth looking at. I need to find a way to classify the images and only show
those of sufficient interest. More programming ahead.
The Pelargonium “rhodo”
which we got as a cutting three months ago is flowering.
Took some photos, and had to compensate by fully 2 EV to get the correct exposure:
As planned, spent some time looking at what I need to do to rate my photos and display only
those with at least a certain rating. Did a bit of playing around and was able to do
selective display, but the real pain is in passing the parameters to the next page. I need
a better solution for that one.
Also did some work on HDR panoramas. Tried both methods that I had considered yesterday,
but only one of them worked. I had considered two alternatives: make three, one with +2EV,
one normal, and one with -2EV, and merge them. Alternatively, merge the pre-merged images.
I had anticipated problems because the stitcher doesn't like component images with different
exposure, but that proved not to be a problem. The real problem was that enfuse
doesn't like images with different sizes, so the first alternative didn't work.
Was it worth the trouble? Here the comparison, first the normal panorama, then the HDR
version:
Certainly the HDR image is a lot more even, but the dark skies still worry me. There's also
the issue of motion: in that photo above, one of the HDR images of the horses shows the
motion between the individual photos:
Still no water, of course—if anything, the leak has got worse. Yesterday we were able
to have a shower, though I don't know how much water we lost in the process. Today the
pressure was too low. Looks like we'll have to shower at Chris' place for the next few
days.
Today was also the day we invited Ray and Lee Nottle to dinner. That didn't make sense
without water, so we moved it to Chris' place as well:
It proves that Ray works for Powercor, and
he had actually seen my last complaint. He's quite defensive of Powercor's strategy, but he
also gave me some interesting information: our power subgrid (if that's the term) is BAS 22
(Ballarat South), the line we have across the property is a SWER (Single Wire Earth Return)
line, a kind they're phasing out. As the name suggests, the earth provides the return
circuit, about 8 A for the kind of cable we have here (at “12,702”V, a
resolution completely out of keeping with the actual voltage stability, but which is derived
from 22,000 × √ 3 / 3).
Another abbreviation is the term ACR, for “Automatic Circuit Recloser”,
something I've always just called “recloser”. And the do log all
failures! Eddie Barkla and others have told me that they don't. Why? Ray said he'd have a
chat with him. It also seems that there's compensation for the number of power
failures, not just the duration. That, too, is the first time I've heard that.
Spent much of today trying to locate the leak in the water system. Yvonne called CJ in the morning, and he suggested he should come along in the
afternoon. He did that, bringing Sue with him, and while we looked for the leak, Sue and
Yvonne removed the remaining fruit from the peach tree.
Everybody seems to think the problem is with the pump. Yes, it is a Davey, but everything
was pointing to a leak: when the pump runs, it produces reasonable pressure. But when it
stops, the pressure quickly dissipates. Clearly a leak.
CJ still wasn't convinced, so I thought I could kill two birds with one stone, and dragged
out the new pressure pump we bought from ALDI a week or two ago, with the intention of putting it in the circuit instead of the Davey
pump. But all the fittings were different, and it would have taken a lot of work. So
decided to show CJ how the pressure dropped as soon as the pump was turned off, and that
convinced him. First we tried to guess where the water pipe went under the ground. I had
already established that the pipe points away from the house, so CJ dug a bit around it to
find where it went once it was under the ground. Answer: further away from the house. The
following two photos are taken from opposite sides:
Gave up on that and set off looking under the house to see where it came out again. CJ
disappeared under the back of the house and came back quite some time later having found
nothing. So off at the other side of the house, where we had to remove the base boards to
take a look under the guest toilet (closest to the water tanks):
Yes, we found some water pipes. 10mm copper, clearly coming from the other end of the
house, which is so close to the ground that we can't get under it. Gave up and decided to
call a plumber tomorrow. As expected, this looks like being slow and expensive.
While CJ was there, did some thinking about a new vegetable garden, which we'll put around
the greenhouse. The old garden is so tight that we can barely move in it. But we'll have
to find a way to keep the kangaroos and wind out. Maybe a fence would do both, though
possibly we'll have to booby trap it to stop the kangaroos jumping over.
I couldn't get my mind off the pump business. Where's the leak? It's pretty clear that
it's not in the immediate vicinity of the pump, but that still leaves the part under the
ground and the part above the ground. Our hot water service is a solar-assisted tank on the
highest part of the roof. To do things properly, I should have checked whether there was
anything wrong up there, but that would mean getting up on a ladder, not made any easier by
the gutters which make it difficult to rest it against the house. And the sudden drop-off
of pressure pointed strongly towards the leak being lower down.
And what about the verandah? When we built it, we
found a surprising number of pipes underground. Had something subsided and broken a pipe?
But the verandah doesn't seem to have subsided, and the pipes we saw were 10 cm pipes, for
sewage and drain water. The supply pipe was probably only 2.5 cm internal diameter. Did a
lot of studying the photos I took at the time—and being happy that I took
them—but I really can't see that there's a problem there. I hope I'm right: it would
be particularly expensive to fix a problem down there.
Of course, all would be a lot easier if the weather had been normal and dry. We could have
looked round for wet patches in the ground, like we did successfully in Wantadilla years ago. But since Friday we had over 50mm rain, so
the soil is moist everywhere.
OK, back to first principles: block off the exit of the pump and confirm that the pump works
well when there's clearly no output. I would have done that a lot earlier, but the symptoms
were wrong, and I couldn't find a screw-in plug. After about 30 minutes searching, finally
did find one and screwed it in. The behaviour was unchanged. After a bit of thinking about
it, it was clear that the built-in non-return valve must have failed. A typical good news,
bad news story. The bad news is that the pump has failed, the good news that we (probably)
don't have a leak elsewhere, and that we haven't lost any water.
Why did it take me so long to find the problem? Everybody said it was the pump, and I've
had similar experience myself. But I was making the implicit assumption that the leak
couldn't be back into the tank. If you're wondering if the pump has failed, you shouldn't
exclude specific modes of failure. Black mark for Groggy.
Would the ALDI pump have pointed in the right direction? Maybe. That depends if it has a
non-return valve. But more importantly, today's Monday (and a public holiday, Labour Day).
If I had thought this through properly, I could have replaced the pump two days ago.
Planning to go to town today, so started copying MP3s to my SD card. I do that with the
Apple for reasons that I forget. The player in the car is pretty simplistic: it plays the
files in the order it finds them in the directory, which in general is not what I want. So
I use special scripts to copy them, and I need an empty file system to start with.
Just started copying when I realised I had done something wrong, so hit ^C
to restart. The system hung, and I had to reboot. After rebooting, tried to
“erase” the card (rebuild the file system), but all the options on “Disk
Utility” were greyed out. No explanation why. I don't know the command to build an
MS-DOS file system on Apple, so tried just copying again, this time with a shell script,
which told me: “Read-only file system”. Why doesn't “Disk Utility”
tell me that? Once it was UNIX with the reputation of ambiguous error messages, as one of
the fortunes in the FreeBSD fortune program
recalls:
Ken Thompson has an automobile which he helped design. Unlike most automobiles, it has
neither speedometer, nor gas gauge, nor any of the numerous idiot lights which plague the
modern driver. Rather, if the driver makes any mistake, a giant “?” lights up
in the center of the dashboard. “The experienced driver”, he says,
“will usually know what's wrong.”
Now, it seems, modern software gets error indications, but it doesn't want to scare the poor
end user, so it suppresses even the “?”. You'd think that it's better to
frustrate the end user than potentially scare him.
But the card wasn't set to read-only. Did a bit more playing around, and in the end the
Apple refused to see the card at all. I got the feeling that the card might be defective,
but put it in swamp, my FreeBSD test box, and it worked fine—and the MP3 player
in the car played the tracks in random order. No idea what's wrong with the Apple, but it's
not telling either.
Into town this morning with the pump to get a replacement from the Ballarat Pump Shop.
They're the other end of town, and they have a remarkably run-down appearance, but I've been
happy with their expertise in the past. Spoke to Amy, described the problem, and she said
that they could repair the pump in a couple of hours. Explained that we were from out of
town, and she put it on a test bench, were it worked perfectly: there was a non-return valve
in the supply line. So bought a non-return valve, which she fitted for me, and away after
paying a total of $30 in repairs—much less than I expected.
While we were there, spoke to another mechanic about surge limiters for the bore pump. They
won't work: without that surge, the pump just won't start. He recommended me to get an
inverter to start it off—it'll need to deliver a surge of about 35A, as he discovered
from his documentation. That's enormous! Our whole house power supply is only rated at
63A. Down to an electronic shop, where they had inverters from Jaycar, but the biggest
one only delivered a surge power of 3.5 kW, or about 15A, and it cost $529. It's
clear that I can't get the current I'm looking for at a price I can pay. We'll have to
investigate the diesel generator again.
Then down to Big W looking for coffee
machines and millipede repellent. The only coffee machine we found at a price we were
prepared to pay cost about $16, but it had a mesh filter, and the shape didn't really accept
filter bags, so gave up on that and looked at what Warehouse Sales had. They only had a single
coffee machine, a big espresso maker that cost roughly 50 times the price of the one we saw
at Big W. So we'll wait for a special somewhere.
Didn't find anything for killing millipedes either. The person managing the garden
department told me to go to Bunnings and
look for a specific product: she had ordered it before Christmas and was still waiting for
it to trickle down the Woolworths' hierarchy. She was clearly very angry about the matter.
The batteries in my cordless drill are gradually giving up the ghost. It must be years
since I bought it, and I've been looking for a replacement for some time. Replacement
batteries aren't the answer: firstly, they're not available, and if they were, they'd
probably cost more than the $30 I paid for it. I even bought another really cheap model
without speed regulation for $10, but it seems that the days are gone when the market was
flooded with them. The cheapest I've seen so far cost round $100, and I've seen high-end
models costing as much as $800. Not the money I want to spend replacing batteries.
At Big W, found a drill that looked pretty much like what I was looking for. Only $50 with
“82 accessories”. How many batteries? Does it even have a charger? No
mention—just a few buzzwords. Here's the entire description:
Clearly you'd expect one battery and a charger, but is there a second? No mention, just
that there are 32 screwdriver bits and 22 drills. You have to look at the photo of the kit
to guess what the other 28 accessories are: but it did show two batteries, so bought it in
the hope that this was part of the description.
Then to Formosa Gardens looking for millipede killer. Again, nothing much—just something for lawns, and
another with the warning “not to be used for edible plants”. I was trying to
keep them out of the tomatoes, so that was no good. Why is there almost nothing to combat
millipedes? The assistant at Formosa confirmed that she, too, had had a lot of problems
with millipedes lately. Is it just a matter of documentation? There are lots of sprays for
caterpillars and grubs. Is their metabolism so different from that of millipedes? In the
end bought a spray against caterpillars and things. We'll see how that works.
Back home and connected up the pump. Another case of good news, bad news. Yes, the pump
cut out and didn't cut back in, so we have final confirmation that we don't have any
significant leak on the high pressure side. But the (brass) non-return valve leaked!
There's a seam where it appears to be screwed together, and water was spurting out there for
about 50 cm in each direction. Tried to unscrew it to tape it up, but it seems that wasn't
intended: instead I broke off the thread, which appears to have been (badly) cemented.
Nothing for it: after some cursing and swearing, back to the Ballarat Pump Shop with the
remains. At least Amy didn't accuse me of breaking the thing, but I asked her to pressure
test the replacement, and that leaked too, though not as badly. She didn't try another;
they're going to send the whole lot back. Instead she found a plastic unit and tested that;
no problem, but it cost me $10 more.
Back home to replace it, this time with replaced high-pressure connections too. The
additional length of the non-return valve meant that I had to change the hose on that side
as well. Amy tells me that the Philmac connectors I had been using were no good, and that
the O-ring would leak after being taken apart. I said that I had been using them for years,
and they had always worked for me, to which she said “You should report it to them.
It must be the only one that does”. So to be on the safe side I bought their
favourite kind. Also asked her about her experience of Davey pumps: they're the worst she
knows, apparently. Most of the pumps they repair are Daveys. At least this time it
confirms my own opinion.
The connectors have four parts: the body of the connector itself (here an elbow), a plug
which fits into it (here the Philmac version with an O ring), a compression sleeve and the
cover. The plug fits into the hose on the right, with ribs that stop it coming loose, and
into the body on the other side. The cover presses the sleeve onto the end of the hose,
ensuring a tight fit.
Fairly straightforward stuff, and it works well. The main difference in the new connectors
was a tube about 4 cm long instead of the O-ring, which fitted closely inside the body. So
I put two connectors on each end of a short piece of pipe and took the fitting to the pump.
Pulled back the cover, and the sleeve came with it: the sleeves on these new fittings have
something like barbs on them, and they catch inside the cover.
How do you get the thing back out again? The hose was only about 20 cm long, and I had
cleverly pulled the cover back so that it was touching the other cover. No way to get at
the sleeve any more. I couldn't pull the cover down with the sleeve inside: that's what
they're designed to stop. In the end, I had to cut the hose and remove the covers and
sleeves, then cut the hose off from the ribbed part of the inside connector, and start
again. 20 minutes' time wasted because of this bloody sleeve.
Carefully back to the pump and tried again—and despite my caution, it happened
again! GROWL! Back to the garage, cut the thing apart again, and managed to
jab the knife into my thumb:
Bloody indeed! Spent some time tending to the thumb, and came across the idea of not
putting the sleeves on until I was ready to tighten up: they have a gap at one point (very
bottom on the photo above), and they're flexible enough to fit around the hose. After about
an hour, finally finished what should have been 10 minutes' work and got the thing running
again. 4 days without water! I never want to see a pump again.
My light meter arrived today, and with it some of the worst documentation I have ever seen.
They thoughtfully included a 9V battery, so clearly the first thing was to put it in. But
where? There doesn't seem to be anywhere to insert it. Even the flap on the back only
reveals an (undocumented) adjusting screw:
As they say, if all else fails, read the instructions. So I did. They're 7 pages long, and
don't even include the obligatory safety warnings, though they do include specifications.
Only 4 pages are relevant:
The parts and positions diagram doesn't show where the battery compartment is, and the
instructions for changing the battery don't say where it is. There's nothing obvious on the
body of the unit. Gave up and decided to contact the vendor, but first looked at the
description on the auction:
A removable cover! Not mentioned anywhere in the “documentation”. Nor, of
course, how to remove it. It proved to be flexible enough to just bend and pull off, after
which I finally found the battery compartment:
But what an effort! Why can't they write documentation any more? Played around with the
unit and tried to measure flash intensities with it. Complete failure; it doesn't recognize
the flash at all. Looking at the description (about which I complained last week), it seems that it only offers a maximum
reading, which does work. I had thought it also included the term “flash”, but
those were the others. All of them seem to make about 2 measurements a second, and it seems
that they measure the instantaneous light, so it would be a complete coincidence if the
flash went off just then.
Flash exposure: still a problem
Taking the photos of the pipe fittings for yesterday's diary proved more of a problem than I
had expected. The first photo came out fine, but then I wanted one with the elbow as well,
and that was greatly underexposed. Changed the aperture from f/8 to f/4 and saw absolutely
no difference:
Discovered that I had the studio flash units turned down to a fraction of their (not very
high) maximum intensity, so turned the up, and still saw absolutely no difference (first
photo). For these photos I was triggering the studio flash units as slaves via the camera
internal flash—and they did, indeed flash. Closed the flash unit and connected them
via the cable, as I had done for the very first, correctly exposed photo (not shown here),
and got a completely overexposed image:
Looking at the photos afterwards show that it's clear the studio units didn't contribute at
all to the first three images: the shadows are missing. So one conclusion is that they must
not have fired quickly enough. When I triggered with the cable, they fired correctly.
But that's not all. The camera's flash has a guide number of about 11. The photos were
taken at a distance of about 1 metre, so the minimum aperture would have been f/11. The
first three photos were taken in manual mode at f/4 and f/8, so the flash was powerful
enough to expose the image correctly, but it was greatly underexposed. They were also
exposed almost identically, so it seems that the camera thinks that this was a correct
exposure. There's something very wrong with this situation (including the automatic
exposure, which would have given me an aperture of about f/2.2, far too wide for this kind
of photo).
Harvested some more potatoes today, not too early: some of them are sprouting again. Like
the tomatoes, some of them have also been eaten by millipedes. Sprayed the tomatoes with
the new spray (with the particularly specific name “Success”), but a direct hit with the spray didn't do much to them.
Hopefully they'll crawl away and die, but possibly we'll have to accept that
“Success” is a failure as far as millipedes are concerned.
The yellow one is the oldest; it's been effectively useless for years, since its batteries
died. I bought the red one in the hope that the batteries would be compatible, which they
are from the point of view of electrical characteristics and shape. But the position and
shape of the clips are different. The blue one (which I bought yesterday) is really almost
identical to the red one; I'd guess it comes from the same factory. It certainly uses the
same moulds for the casing. It would be handy to have two drills, but once again they have
changed the clips. The battery from the new drill doesn't fit either of the other two
drills. Three different connectors, all for the same battery type. Why do people do this?
Took some photos of Piccola this afternoon, in a
bit of a hurry. Discovered I had left the camera in manual exposure mode, with the result
that the first three photos were underexposed by 4 EV:
Not a serious problem, since I saw it in time, and I was able to take some more, but I must
remember to reset the camera to default settings after doing macro photography. Still, for
the fun of it, tried to fix them. My standard method is to use the JPEG images from the
camera unless they need correction, and to use the raw images if they need correction. Did
that, but for the fun of it, also used GIMP to
recover the JPEG image. The results were surprisingly good. Here the image recovered from
JPEG, the image recovered from the raw image, and a correctly exposed image:
Clearly the correctly exposed one is best, as is particularly visible with enlarged images,
but the recovered raw image is nearly as good. The JPEG is noticeably worse, but still
surprisingly good for 4 stops underexposure—it's as if I had taken the photo at
3200/36° ISO instead of the default 200/24°. Still, it's also a justification for keeping
the raw images as well as the JPEGs.
But I had just seen the page! Checked: I can access this page until I'm logged in. Then,
suddenly, it disappears. Did some more investigation and found that the number of books
available dropped markedly when I was logged in (right):
Contacted Safari customer support, who completely missed the point and told me that she
could access my account, and that it must be a browser-related problem. I replied with the
specific URL (again) and asked if she had tried it. For completeness' sake I tried
different browsers and operating systems: in addition to my normal combination of FreeBSD and firefox, tried Linux/firefox, Mac OS X/Safari and Microsoft XP/“Internet
Explorer”, all with the same results, which I reported. I was completely unprepared
for the response:
We do have some known issues with Linux and I will be sending you the known issues and
fixes for Linux tomorrow.
No mention of the URL I asked her to try. Even when I pointed out that Linux didn't behave
any differently from the other operating systems, she just responded with:
Does this "Page Not Found" happen with all books or one specific book? I am sorry I am
not used to reading messages in the format you reply back in and I am finding it a bit
difficult to follow.
It seems to me that she's not used to reading messages in any format; she certainly missed
important data right from the start. But are interleaved replies difficult for people to
understand? They may have become a little unusual for the majority of people used to
separating text and reply, but they're certainly not the cause of her complete ignorance of
the text.
In the meantime, went looking for other books. The 126 photography books I can access
include several that weren't in the 197 that I previously had access to. My best bet is
that they have decided to present different books to different people. In that connection
it's interesting that when I log in, the page shows “O'Reilly” prominently at
the top of the page. But presumably they want to sell these books even to people who have a
limited account. Why “Page not found”? Why not “Sorry, this book is not
in your subscription. Pay $Bigbucks to access it”? Mentioned this in one of my mail
messages to customer support; maybe I'll get an answer, but it would be out of character.
Would I pay a nominal sum to access these books? Maybe, but it would help immensely if the
site were easier to navigate. It requires flash (one of the suggestions was
“You must upgrade to Flash 10”, though that appears to be incorrect), and seems
to be designed to be navigated only with the mouse (scroll bars through the entire
book in some cases). If ever there was a case for using the PgUp
and PgDown keys, it's here. But they only work some of the time, and I can't work
out why or when. The “Help”, such as it is, doesn't even tell you how to use
the site.
That's not the only thing that makes Safari a pain. How do you find books in the first
place? They've divided the site up into categories, such as the Digital Photography
category I use for this example—but there's no easy way to get an overview. You can
search, but then you search the entire site. And you can display the books in some useless
order (by default it's “Insert date”, in other words when the book was imported,
which has nothing to do with the date of publication), but only hits 20 per page, so you end
up having to search 10 pages of “How to use your new Canon PowerShot G10” or
“How to use Photoshop the same way all other books tell you”. How simple it
would be to have a search which applies only to a category, and which returns one line per
title, so you can hope to get all results on a single page?
It's things like this that make me wonder what people are thinking. They're clearly not
putting themselves in the viewpoint of the user—and yet you'd think that Safari is a
general enough product that they would want to use it themselves. Are they just
unable to see how a few minor things would improve matters so much?
Safari's not the only one. While discussing the matter on IRC, Jürgen Lock pointed me at
PaperC, a German site with free access to texts (and
relatively high prices, it would seem, for copying data or bookmarking it, something that I
find too tedious on the web). But it doesn't even have a category listing; you need to
guess some search terms and search the entire site to find anything at all. And the search
terms aren't easy; there are plenty of books on photography, but when I search
for Fotografie, I find exactly one book. Maybe things are better if you pay, but
the only things they offer for the paid subscription are things that I don't need, so I
can't be bothered checking.
More moving horses around today. Chris brought 3 horses here, including Darah, and picked
up three others. Yvonne was expecting some fireworks, as
often happens when new horses are put together, but nothing much happened, and I didn't get
much in the way of photos:
A little later, though, Yvonne saw a horse running down the road towards the forest. She
thought it might have been Carlotta, so off to count them, but all ours were still there.
Off in the car to look for it, and found it not too far away: Jesse, one of Lee Nottle's
horses. Looks like they're going to have to look more carefully at their fences.
You get used to monitors suddenly turning off when the (real) screen saver cuts in. Today
was different: my :0.3 monitor, a Sλmsung SyncMaster 2233SW, turned off by
itself. It proved to be dead, after only 11 months. At least it's still in the guarantee
period. Confirming that was more difficult than I expected: it had taken the display card
output with it (it's a dual output card, so now it's a single output card).
So I'll be off to Melbourne soon to get a replacement. I currently have four displays, one
of which is an ancient and fuzzy LG Studioworks 900B, so it makes sense to replace it, but
to keep it running and use the replacement as a stand-in for the defective Sλmsung
until it's repaired.
But what do I buy? Spent some time comparing prices on the one hand and specifications on
the other. You get the former from the vendors, of course, but where do you get the latter?
You'd think the manufacturers would have them on their web site, and some of them do. But
others, notably ASUS (or is that
ΛSUS?), seem to think that web sites are there to show how clever their web designers
are. They have a link to “Product Comparison”, but then you have to enter
product numbers. And what are these products? Who knows? The “submit” button
was conveniently outside the non-resizeable box, so gave up on that and looked at the
“specifications” of a random monitor, the VK266H. Lots of flash, maxing out my poor
Apple's CPU (Firefox on FreeBSD didn't want to display it at all). To see
anything except a fake Marilyn Monroe, you have to select “Skip”. Then it tells
me everything I need to know, conveniently written in grey on black:
So it's a 25.5" wide screen display. Who cares about the resolution? It's SPLENDID!
Buy! Buy! Buy!
Bye-bye, ASUS. You'd think they don't want people to buy their products.
I'd like to say that that eliminates ASUS from my choice, and certainly their web site is
the worst I've seen, but the others aren't much better. I suppose it's modern: it's exactly
the same problem I had with Safari Books
Online yesterday: no overview.
Safari got back to me today, prepared to escalate the problem if more than one book was
involved (why more than one book? But clearly they hadn't read my last messages with the
counts of the numbers of books that were missing). Also confirmation of my suspicions from
O'Reilly: I only have access to some of the
books. That's a commercial decision, of course—they had no obligation to give me free
access to any of the books—but it seems a little short-sighted, and has definitely
caused more work than it was worth in this case.
House photo day again today, and more playing around
with HDR. Learnt from last week and decided
to use the 5 shot bracketing that the camera offers, even though I only really needed 3. It
certainly added to the number of photos I took: a total of 208, including a couple of missed
shots. It also means that I'm accumulating lots of images: today's totalled nearly 4 GB.
It doesn't make much sense to put the component images of HDR photos on the web, so I've
moved them elsewhere. And gradually things are looking better, as the comparison of one
panorama shows (the first on 26 December 2009, the second today):
She played around with it for at least half an hour before it finally escaped. Still, I
suppose that means that she has that much more practice in catching mice.
The change in the garden isn't always for the better. Nearly all summer long the petunias
in the bed around the Ginkgo bloomed
furiously. Then, about a month ago, they started to wilt. Here photos from 6 February 2010 and today:
What's causing that? It started, paradoxically, after we got some significant rain last
month. I turned off the sprinklers for a couple of days, and it seems that the flowers
didn't like it. I put them back on again, but they haven't recovered. Was it the lack of
watering (only a couple of days), or something else? Not enough fertilizer? I wish I
understood these things.
Another power failure this afternoon.
Atypically, it lasted 7 minutes. Normally it's either very short (recloser), or it takes
hours. It seems that Ray Nottle was planning to call the incident room directly (he works
for Powercor), but the timing seems too
short to attribute it to his involvement.
The outage may have been short, but it was enough to shut the machine down, and when things
came back up, we were off the net. Fortunately power cycling the modem solved that issue.
Since we had lost power anyway, tried putting a CRT monitor on the :0.3 X display
output. It worked! No idea why it didn't before; maybe it was just a frequency mismatch.
Anyway, now it's just the monitor I need to replace. Once in a blue moon (there is one this
month), a power failure can be useful.
Read claims of chromatic aberration in the Zuiko Digital ED 50mm F2.0
Macro lens on the OLE-E.DE forum today, so out
to investigate. Took a number of photos,
including some very forgettable, out-of-focus ones with my old Asahi 50 mm f/1.4
Super-Takumar. Results: yes, there's something at full aperture with the Zuiko. I can't
make up my mind whether it's chromatic aberration or flare, but it's pretty minimal. Here
an overall view, then small excerpts (right, just below centre) of photos taken at f/2 and
f/4. To see the flare, you need to further enlarge the images.
It's difficult to say too much about the Super-Takumar, since almost everything was out of
focus. But the overall quality of the out-of-focus areas shows distortion and more
chromatic aberration, here at f/1.4, f/2 and f/4:
These photos are at the same scale, so it's clear that we have to do with a very different
quality of lens nowadays. In many ways, it reminds me of the tests I did with telephoto lenses last August.
But to be really sure, I should think out better images for comparison (and pay more
attention to focus!).
Mail from Safari books online today:
“How did we do with support?”. As I've documented, pretty terribly. But at
least this way I had the satisfaction of telling them my opinion. I fear, though, that the evaluation of such questionnaires
is no better than the rest of their service.
Spent some time looking at what to do with my Samsung monitor today. Called up MSY, who pointed me at the manufacturer, so off to the web
to look for a local representative. More clever web programmers—I think that Java has
more breakage to answer for than any other aspect of web programming. Filled out the
service locator form, pressed “Go”—and nothing happened. Discussed
it with others on IRC, and some could get it to work, some couldn't. Tried with all the
usual web browsers and operating systems, and finally tried the error console on firefox. Thousands of error messages,
without any clear understanding how serious they are.
Gave up and called Samsung's customer service number. After 20 minutes waiting, finally got
connected to somebody who quickly decided that I didn't need a service centre: they would
send me a new (really refurbished) monitor. Under the circumstances, it's certainly good
that I called. The closest service centre is
in Nunawading, on the other side of
Melbourne, and 176 km away.
As if that weren't enough, got a call back half an hour later confirming that the new
monitor will be delivered on Thursday. Under those circumstances, I don't even need to go
to Melbourne tomorrow.
Still, why did the web page work for some people and not for others? It proved to be the
“Go” button: it's broke. Microsoft “Internet Explorer” even reports
an error when you click on it. But if you enter the post code and then press Enter,
it works. I suppose that's typical of modern web software.
I've already commented on how Canon and Nikon have brought out new cameras with maximum sensitivity “102,400” ISO,
and how that shows a complete lack of understanding—or maybe the triumph of marketing
over engineering. Doubtless they'll slip back in line with the existing decibel-like scale
when they have a full range of intermediate sensitivities (the two between "51,200" and
"1,024,000", the equivalents of the ISO logarithmic sensitivities 48° and 49°, for example).
But these values become unwieldy—6 digits where we once used 2, and still generally do
use 3. The ISO logarithmic values are still 2 digits, and they will continue to be so for
the foreseeable future (until 1,667,216,000/102° (really 102.25°) ISO, by which time people
will probably also long have dropped these silly power-of-2 sensitivities).
So I've decided to use logarithmic ISO values where practical, and they're now on my web
photos. I wonder if anybody will follow me. In the process, it's interesting to note that
ISO logarithmic sensitivities are
really log10 (ISO_linear) * 10 + 1. How did they end up
so close to each other? The scales were adapted unchanged from the ASA
(now ANSI)
and DIN, who almost certainly didn't consult
each other when working out their scales.
In the evening with Chris to a BLUG meeting;
Josh Stewart was talking about software licenses, though apparently this had nothing to do
with the fact that today was rms' birthday. It
was Chris' first visit; she's only just starting to learn about non-Microsoft software.
There were more people here than last time—about 10, I'd guess. Considering the
population of Ballarat, that's quite a
lot. I can't recall any meeting of LinuxSA which had more than 30 attendees,
but Adelaide is 15 times larger than
Ballarat. Had a lively discussion, and ended up promising to give a talk about my
weather station software. A surprising number of
people—4—were interested enough to ask various detailed questions on the spot.
The weather has been cool and moist for a while—certainly a welcome change—but
not today: again we had a maximum of 35°, and
the horse flies are everywhere, so we
spent most of the day in the house.
What's a horse fly? According to Wikipedia, the same thing as a March fly. But we had them
last month, and these look quite different. I must get some and take photos.
Another power failure! This one lasted
over 3 hours, and kept me busy getting the machine back up and running. And again found a
new problem with the Apple software. My old machine loses the date when it loses power, and
I have to set it manually. The canonical way to set the date on Apple is the tag
“Open Date & Time”, which require much clicking. Why bother when you have a
simple program to do just that, date?
Admittedly, date has a relatively painful interface. You specify a single string of
digits with minimal punctuation, representing, as the FreeBSD man page says, [[[[[cc]yy]mm]dd]HH]MM[.ss (centuries, years,
months, days, hours, minutes and seconds. The [] represent optional groups. I've
already established that Apple changed the format to mmddhhmm[[cc]yy]: they put the
year and century at the end, not the beginning. And each time I forget it. Today I entered
it almost correctly, but accidentally hit an additional key at the end, and got an error
message:
=== root@boskoop (/dev/ttyp3) /Users/grog 10 -> date 031709042010f date: illegal time format
usage: date [-nu] [-r seconds] [+format]
date [[[[[cc]yy]mm]dd]hh]mm[.ss]
But that's the original form, as from the FreeBSD man page. Tried it anyway. It doesn't
work:
=== root@boskoop (/dev/ttyp3) /Users/grog 11 -> date 201003170904 date: illegal time
So this is just another untidy area where they've gone and half changed something, and
forgotten to finish the change.
Later got a call from Toll Priority:
they wanted to deliver the replacement monitor. No complaints from me, of course, but it
makes me wonder why they called to ascertain that I would be there tomorrow.
The monitor arrived and worked. I'm considerably less irritated with Samsung than I was, though their web site is yet
another indication of how web sites seem to be out of tune with the rest of a company's
operations. In addition, I've now also got rid of the scratches that the old one had. Took
the opportunity to try to tune my X installation. What a pain! To start with, the Samsung
monitor came up with a virtual resolution of 1920x1200, though it's a “modern”
1920x1080 panel. That proved to be due to the specs of the replacement monitor I had put
in: there was a mode line
I had added the 1600x1200 for the CRT monitor, but that increased the virtual
height to 1200. Isn't it sad that old monitors have higher vertical resolution than modern
ones? 10 years ago I had a monitor with 2048x1536.
The other issue I have isn't really an X problem: the left-most monitor is still a CRT, and
for some reason it frequently comes up out of sync on a cold boot, such as I had today. It
made more sense to put the system console on the second panel, which is as simple as
swapping the plugs. But that meant I had to reconfigure X to change the sequence. That's
easy enough: the configuration file defines Screens (what you see in front of
you), Monitors (the physical hardware) and Devices (the display cards). Individual sections
describe each instance of each of these and how they relate to each other. So for my first
two screens I have:
So really all you need to do is to swap the Device entries, and it should work.
But it didn't. It seems that something in the drivers attaches the monitors in the
sequence it wants, and not in the sequence I specify.
There's also a section at the top which says how they relate to each other:
Tried that, and it worked, but it's not invisible: the display numbers derive from the
screen number, so this meant that I had, in sequence from
left, :0.1, :0.0, :0.2 and :0.3. More head-scratching
needed. Maybe I'll swap the two halves and put the console on :0.2 (screen 3),
which would be more convenient anyway.
My /home file system is getting ever fuller, and after the catastrophe I had
setting up the replacement, I had postponed things. But
it's finally time to do something. I could just run newfs on the disk and start
again from scratch, but maybe there's something on the new disk that got lost from the old
disk, so started coalescing the files in preparation for comparing them. That took all day:
there was the best part of a terabyte on the disk, about 400 GB of which disappeared by
evening.
The delivery driver brought me another tidbit:
a horse fly landed on his shoulder, so
I smacked it and collected it. It wasn't dead—a German horse fly would have
been—so I was able to take some photos of it, as planned yesterday. The strange
position of the right wing appears to be the only harm the insect had experienced when I
caught it.
Continued moving data across to the new /home file system today. What a time it
takes. I now have nearly 7 million files on the file system, many of which are copies and
old versions.
Filesystem 1048576-blocks Used Avail Capacity iused ifree %iused Mounted on
/dev/ad8s1f 888151 798178 18920 98% 6690558 110857472 6% /HOME
It'll be a while before I work my way through that.
A year ago I lent Aligi Voltan my 1826 Savary jeune bassoon, the last remaining
Savary that I had. I was dubious about whether he would be able to use it, and with good
reason: it seems that it had been shortened to raise the pitch, and repair was quite
involved. Strangely, though, he liked it and offered to swap it with a old Heckel instrument, serial number 3620, built in 1892. I'm a little sad about losing my last
Savary—I was sad about losing the first—but an old Heckel is certainly interesting too, and considering that I don't
perform on them, it seems a shame to keep them from people who do.
So we agreed to the swap, and Aligi sent the instrument on 2 March 2010. I followed its progress via the tracking site (tracking numer ZA001603147IT) as it worked its way
from Padova
to Milano (2 days), then via Germany to
Australia (another 2 days), and it was cleared by customs on 7 March 2010:
The next day was a public holiday, so I expected it on 9 March 2010.
But nothing arrived. Two days later, Yvonne checked at the
post office—still nothing. I called Australia Post, but they couldn't do anything with the tracking number, and suggested
I get it traced from the other end. Finally, on 12 March 2010, I
asked Aligi to trace it, and today I finally got the results: it wasn't Australia Post at
all who had to deliver it; it was UPS. They had
given it the tracking number 1Z11WR410473735162. Went to the tracking page and was
amazed:
They had been sitting on it for 11 days with the claim that I had requested them to do so.
Clearly the next step was to ring them up; but we don't have a phone book at the moment, and
getting contact details from web sites is usually painful, so went to the White Pages web site. What
they presented me was amazing:
That was what it looked like at the time. When I looked again a year later, things
were correct.
That's the phone number of a competitor of UPS. Yes, the page is long, but there's not a
single reference to UPS (not even the string) in the entire response. I'm left wondering if
this is deliberate; certainly UPS must be able to claim damages.
Finally found the number (131 877, or, as they obfuscate it, 131 UPS) on the web site after
all. Rang up and spoke to Kristina and asked her to get it delivered today. She asked
about my address, which was correct. She told me that they didn't have my post code (3352)
in their computer. What kind of system is that? I offered to pick it up
at Ballarat railway station, like I have
had to do before (two years ago today, in
fact). And like last time, she
talked about deliveries to the post office with Australian
Air Express. I told her that I had always had to pick up the parcels at the railway
station, but she didn't listen. When I insisted, she said she had listened, but clearly she
didn't understand, or maybe the concepts were just too difficult for her.
She did collect all the delivery details, though, although they were all correct, and
finally (on the second attempt) connected me to her supervisor Jennifer, who sounded a lot
more human. She said she had never seen anything like that before, that they weren't
allowed to hold on to parcels for more than 5 days, that she would follow up with the
delivery people, and that I should hear from them within half an hour.
That was round 10:00, and sure enough, round 10:20 Nina called me and said she would arrange
delivery and call back. That she did at 10:35, promising delivery between 14:00 and 15:00.
The courier arrived at about 15:20. He had been sent out just for this delivery, and was
armed with nothing better than a Melway,
which doesn't cover this area. So finally I have my new bassoon:
But what a lot of stress! And every time I think that UPS may be getting better, they come
along and do something like this. I suppose it's par for the course that they had lost my
customer login altogether. I was able to re-register with the same user name, mail ID and
password. But what should I think of a parcel service that loses things?
We've been pretty lucky with smaller pests in the garden. Possums and kangaroos yes, but
few sucking insects. But today I found
some aphids on one of the roses on the
verandah:
Yvonne wanted some of the images on Jill's camcorder, a
Sony DCR-SR45. I've had bad experience with many cameras, which come with
proprietary interfaces where a standard USB SCSI emulation will do, but took a look anyway.
Kudos to Sony: they did it right and presented a FAT32 file system with MPEG-2 files.
Copied them without any problem.
Well, not immediately. Jill came back again a little later: “Houston, we have a
problem. Those files you copied aren't there any more”. It eventuated that some (not
all) of the files showed up on the camera's screen, but without an image, and when I tried
to play them, it displayed the message “No applicable files available”. Took a
look at the file system: yes, they were still there, and I could access the copies on the
camera with mplayer, so they weren't
damaged either. And the timestamps were all unchanged, as you'd expect, of course, if you
opened them only for reading.
What happened? My guess is that the camera lost some internal index, but I couldn't find
anything obvious in the file system. Downloaded the handbook and looked in vain for
any information—didn't even fine the meaning of the message.
Finally called Sony support (1300 13 7669, or, as they like to say, 1300 13 SONY) and spoke
to Alex, who hadn't heard of the problem either, and who went enquiring. He came back
suspecting that it was file system corruption due to disconnecting the cable without first
umounting the computer. It seems that you need to tell the camera before
disconnecting the cable. That's the only consumer device I know of that needs this kind of
care. I didn't change anything on the camera, so there should be no potential for file
system damage. But Alex confirmed that it's possible to get severe damage as a result. On
his recommendation, reset the camera, but the problem remained.
That brings up questions of merchantability and fitness for purpose. I certainly wouldn't
have expected it of Sony. Further research on Google found other instruction manuals which do describe the error message. They
state literally: “No applicable files available: Image files have been deleted”.
So why doesn't the message say that in the first place?
Jill later told me that the camera has a function (conveniently not described in the manual
I downloaded; there must be a second one) to reformat the disk drive, so she'll do that when
she gets home, after downloading the files to her computer. Alex recommended taking it to a
service centre, but there aren't too many in
the Bundaberg area, and I don't think
they can do much anyway: it's clearly a serious and long-standing software issue.
Improving satellite IP reliability
Call form Daniel at Aussie broadband in the afternoon. They're putting me on a dedicated link for the weekend to see if that
improves things. It's probably a bad time to start: the last few days have been relatively
good, only 4 outages in 5 days. And given the (apparently) sporadic nature of the outages,
I don't think they'll find anything of interest in a single weekend.
Finally got round to picking some hops today, some Tettnang that looked like they were going
brown. The horse flies convinced me not to spend too much time on it.
So finally I have my Heckel bassoon. What to do
with it? Play it, of course, but also take photos. The latter proved to be more
complicated than I thought, at least for images of the whole instrument. My 50mm macro lens
is just too long-focus: I would have to be about 4 m from the instrument, and then I need a
way to hold it so the background doesn't show through. Finally ended up putting it on a
carpeted floor and taking the photos with the Zuiko Digital ED
12-60mm F2.8-4.0 SWD at about 25mm focal length, in the process getting better photos
of all my better bassoons (I have three that are currently packed away), here the Moosmann:
The camera is just off the top of the image; I couldn't get far enough away to get it all
in. Finally a use for the full length of my enormous tripod. Here the first photos:
More will follow, but it's going to be fun to get a legible copy of the maker's label: it's
been filled in with generations of linseed oil.
Playing the Heckel
Aligi obligingly included a couple of reeds with the instrument, but as he said, they're no
different from modern reeds. It doesn't have a crook key, nor a hole in the crook, which
makes things interesting. And intonation is clearly an issue. I suspect that the crook is
not the correct one. There's only one, and it doesn't appear to match either of the crooks
in the images on http://www.heckelbassoons.info/ (also shown in
my page). f (all fingers off) is very sharp
compared to g (second octave), so maybe I need something longer. But then d' is very sharp
compared to c', and I can't see a reason for that. The d hole (third finger left hand) has
some wax in it, so clearly this is something that has been known before. I'll have to
discuss with Aligi, who, I am sure, definitely didn't want to mislead me: he did warn me of
intonation issues. Still, it sounds nice, and once I get used to the different feel of the
keywork, I think it'll be a nice instrument to play.
Another day spent almost entirely with photography. I had hoped to take fewer photos than
last week, when I made a number of mistakes, but in the end I took more—a total of 201
photos, and at the end of processing I had 4185 MB of photos in my source photo directories
and 105 MB of web pages.
Was it worth it? I don't know. Here are two views. In each case the first is a slightly
overexposed single image (the middle of my set of 5), and the second is the blended HDR
image.
There's no doubt that the gradation of the HDR images is more pleasing, but the slightest
wind makes them unsharp, as is particularly evident in the second image. I can't make up my
mine what to do.
More looking around the garden. Sue Giddins and her mother had looked in briefly yesterday,
and the mother looked at our Hibiscus and pronounced them magnesium deficient because of a
couple of yellow leaves:
Yes, there's something wrong there, but it doesn't look like magnesium deficiency: it's
mainly on one side of one leaf, and the others look happy enough. Turned it over and
found White fly, something that we
thought we had seen the last of years ago in Wantadilla.
Took a look at our surprising dying petunias,
which are now looking even worse than last week. Noted black spots on them, and they're
sticky, so I assume that this too is the result of some insects. It's probably too late for
the petunias, but we can try. Yvonne had to leave for an
unplanned shopping trip, so got her to pick up some magnesium sulphate
(or Epsom salts, as Yates prefer to call it, presumably so that gardeners
aren't scared by using chemicals) and another sprayer—they only cost $10
nowadays, so it's practical to have one full of each kind of spray.
We also seem to have problems with another plant nearby, a
volunteer Wattle (presumably
an Acacia pycnantha). We've
seen this before in Wantadilla too: our super gardener
Ben Henderson piled mulch up to the base of the trunk and allowed fungal infection. And
yes, once again the mulch was a bit close to the trunk. Pulled it back; hopefully the tree
will survive.
One of the few things that seem to be relatively worth while growing in the garden is
garlic. We didn't plant them. They were there when we arrived, and after digging them out
from under the Osteospermums, they
grew well. Now we have some of the biggest cloves of garlic I have ever seen, over 40 times
the weight of a small clove of garlic:
Since 00:00 on 19 March 2010 I've been on a clear channel on
IPStar. It's not difficult to determine when
it started. The “TCP speed” value has doubled:
I've also not had any disconnects, though that's not a very reliable statistic: I've had up
to a week without dropouts before. But my ISP is talking about keeping me on the clear
channel, which might have some advantages. Tried speedtest.net again, which wasn't very conclusive. It showed speeds round what I'm supposed to be
getting (about 1 Mb/s), while previous tests gave ridiculous values of up to 11 Mb/s
downlink speeds. But it still claims that I'm in Dortmund, Germany, and gives me
advertisements in German.
Rebuilding file systems, continued
More playing around with my new file system, and managed to coalesce things quite a bit; I
should be in a position to switch over soon. Yes, I'm being very cautious, but with good
reason. Even as it was I managed to delete some files accidentally. I think I have a
bug in mklinks where it might delete files
from the destination directory even if they're exactly the same file (same directory entry)
as in the source directory. Fortunately I found out in time and was able to restore them
from backup. But it's a worry anyway.
The flash unit's firmware can be updated through the USB port and adjusted to the
technical requirements of future cameras (Firmware Update).
For more information, visit the Metz homepage at www.metz.de.
Of course, there's nothing on the home page about firmware update. And the firmware update page says nothing either. All I find is the ability to download a
file called MB_58_AF_1_Olympus_V2.1_GB_Win.exe, and they don't even tell you what to do with it
when you have loaded. As you might guess (but not reasonably assume) by the name, it's a
Microsoft executable. So to update the firmware you need a Microsoft system, it seems (and
a USB cable, which Metz doesn't supply with the unit). Score 1 against Metz for supplying
only Microsoft executables, and 2 against them for worse than useless documentation.
Ray Nottle over in the afternoon, didn't find us, and left a wheelbarrow containing about 10
kg of courgettes, a couple
of agaves, one variegated, and a bucket
with some evil-looking liquid in it, which Yvonne threw over
the garden bed. Ray came over again later: the liquid was worm castings, and should have
been diluted 20:1. I've seen what happens when you don't, so Yvonne out again and hosed the
whole thing down. Hopefully it won't do any harm now.
Yvonne has some new DVDs from Alex Kurland (“Hip-shoulder-shoulder”, which she (Alex) repeats every 20 seconds). I
can't stand hearing it any more. Asked Yvonne if she'd like a DVD drive on her computer,
and indeed she did, so solved two problems there. Funny that I didn't even think of putting
one in before.
Finally a dry day with little wind, perfect for spraying poison. Got through about 12
litres of glyphosate; hopefully I
won't have to do that too often. Also planted
the agaves, the variegated one in the
Japanese garden, and the other in the rear part of the east garden, the one we're planning
to work on in the near future.
The white flies are by no means gone, so sprayed
more pyrethrum, but I don't think we'll
get rid of them until the tomatoes are all harvested, which could be another couple of
months yet. Also found some black insects on one of the petunias on the verandah; they're
probably the same kind that killed the petunias round
the ginkgo.
In general, we're ending up with lots of things to harvest. A couple of weeks ago we had a
glut of peaches, and now it's cherry tomatoes, which we're harvesting faster than we can
eat. I'm reminded of the stupid advertisements of the Diggers' club. In their garden annual 2008-2009 they
claim that you can feed a whole family for $6.50 per week. Clearly you'd have to be
vegetarian. But what nonsense! They have only 9 kinds of vegetable, and they are fresh
only at certain times of the year. In the following, I'm quoting their own book “The Australian Fruit and Vegetable Garden” for the harvest times.
Lettuce, which can really grow all year round in this climate.
Beans. They don't say much, so I'll pick broad beans because they're the most reliable.
With staggered planting, you can harvest from May to October, I suppose.
Broccoli. In our climate, you can grow this any time too. Further north you can only
harvest from July to December. No broccoli in the first half of the year unless you
freeze it.
Peas. Harvest from March to December, further north from April to October.
And that's all. What a diet! It's extremely unbalanced. It's also amazing that they don't
have any potatoes. Even more amazing is that, since they published this, they've changed the composition of the
seed pack: instead of carrots, they've added capsicums (harvest December to March). They've
also toned down their claims (and raised the price to $8, and dropped the harvest from 108
kg to 96 kg). This gives you nothing to eat in January except lettuce, broccoli, onions and
capsicum. Yes, you can dry things, notably the beans, and you can keep onions for an
extended period of time, but can any thinking human being want to live on that kind of diet?
And what are Diggers thinking of choosing that limited combination of vegetables?
Another anniversary today—Yvonne's and my silver
wedding anniversary. We didn't make much of a fuss about the wedding, so we didn't make much fuss about the anniversary either—just some
better than usual food. We had planned our standard recipe for Coquilles
Saint-Jacques, but forgotten to buy the necessary mushrooms. But then Yvonne found a
recipe for Coquilles Saint-Jacques aux
courgettes et au thym in “Les recettes de l'auberge de l'Ill”, and we had plenty of
courgettes, so tried that instead, with
good results. I made a maigret de canard. And that (apart from some cheese a little past its use-by date) was our wedding
anniversary celebration. I wouldn't want it any other way.
Why am I going to so much trouble with the new disk? All the data I was trying to
“recover” was just a copy of valid data on /home, and out of date at
that, so I could just run newfs and start again. Originally I had intended it to
replace the current system disk, but the way things go, I'll probably need both. So I'll
revive the file system organization that I used from about 1992 until 2007, where I had my
own stuff on /home and other files on /src. The new disk will be /src,
and will also include multimedia files (photos, music). Started copying things
across—even at modern speeds, it takes forever, and only really finished in the
evening. Then started moving the real contents of /src (currently split
between /home/Src and /dump), in the process discovering a number of symlink
loops to make my life a misery.
Came into the office in the morning to find that bsdtar had died copying a recursive
symlink, apparently hundreds of gigabytes of it. The core file was 3 GB in size, so clearly
it had run out of memory. Removed the source loop and set off to remove the hundreds of
files I had copied multiple times. I had deliberately not set soft updates for exactly this
eventuality, since I've had problems with crashes when deleting large numbers of files with
soft updates enabled.
Then I went off for breakfast. When I came back, the system had crashed anyway! So the
problem isn't related to soft updates. Maybe it's related to the sheer volume of data
involved: when I remounted the disk (read-only), it had deleted about 300 GB of data. I
suspect an out-of-memory issue there.
So: give up and start All Over Again. Spent most of the day copying the files, finally
finishing round 17:46. I'll hold on to the old copies for a while until I'm sure that
nothing has got lost. And then I'm done! But how do I delete the old files?
We still have plants popping up that we didn't know about. This one is on the west side of
the house; that's the disused gas heater behind. Time to transplant it, I suppose. We also
have a number of tulips, and the first daffodils are already poking their heads out.
Did a bit more work in the garden; we really need to tidy up the area to the right and in
front of the verandah. Did some work, but not too much. Also finally put up wires
for the raspberries that we planted over a year
ago. Some of them are bearing fruit already, not a foregone conclusion in that bed:
we have various brassicas there that are
looking very unhappy. And the tomatoes in the next bed seem to be suffering from a variety
of pests, including snails, white fly and possums. I don't think we'll get too many off
them.
The petunias in the baskets on the verandah are no longer looking as happy. They, too, have
sticky stalks, though I can't find anything obvious. Trimmed them and sprayed them
with pyrethrum, and discovered a large
number of little black spots:
Are they some kind of parasite, or maybe just seeds? They're hard, and they don't seem to
have much in the way of extremities, so maybe it's the latter.
While piling up some irises for later use,
came across a large number of strange looking insects, which unfortunately ran away before I
could get them in large numbers:
The Zuiko
Digital ED 50mm F2.0 Macro is an excellent lens, but for a macro it has a very
wide aperture, and the camera insists on using it, even when it has the choice. The photo
above was taken with flash at 1/100 s, f/2, and the extreme shallow depth of field is very
obvious. Why? It would have been just as easy to select f/8 or so and just flash more
strongly. I can do that with aperture priority exposure, but I have to think of doing so.
One of the many strangenesses about the E-30
A question on the
Ole-E.de forum caught my eye this morning:
somebody claimed that you could only use manual lenses on the Olympus E series in M (manual)
or S (shutter priority) mode. The latter is obviously wrong (and proved to be a typo on the
part of the person who wrote it), but it got me checking. Yes, you can use the lenses in
any position except S (i.e. also P (program) and AUTO (default values)). But the
test photos I took (with an Asahi Super Takumar 50 mm f/1.4) showed a serious problem:
camera shake. Tried three separate series of photos, and couldn't get rid of it. This
detail was taken at 1/125s mounted on a sturdy tripod, and it still shows clear evidence of
camera shake:
Finally it dawned on me: I had “image stabilization” enabled, and the manual
focal length setting was where I had left it last time I had taken a photo with a manual
lens, at 800 mm. Turning it off solved the problem. But why should there be any reason for
image stabilization to cut in when the camera is mounted on a tripod and thus can't move?
A little more work in the garden, in the process finding more of these red beetles in
exactly the same place I found them yesterday. This time I remembered to use a smaller
aperture:
They're clearly not ladybirds: they
don't have any wings. They dispersed very quickly when I found them: there were only 22
seconds between the first and the last photo. Most of the beetles seem to be immature; only
the three in the first photo appear to be adult, and they were the first to disappear. So
what are they? Are they harmful?
How do you enter non-standard UTF-8 characters via the keyboard? There are a number of
ways, but the most common is the so-called “Compose key”, where you press a
button to signal that the following keystroke(s) should be interpreted as a special
character. I've done this under Emacs literally for decades with a set of functions written
by Howard Gayle in 1987. They work well enough for the keys they cover (ISO 8859-1), but
only under Emacs. And X offers something similar, if only I could find out how.
As ever, the real issue is the documentation. There's plenty of documentation out
there—they're like Andy Tanenbaum's standards, a maze of twisty little documents, all
different. The first step was: how do you specify a compose key? A friend gave me the
incantation:
setxkbmap -model pc105 -layout us -option compose:rwin
That worked (well, I chose the Menu key instead), but it unmapped the other codes I
had set. Clearly it's dicking in the same place as xmodmap, and with the help
of xev established that it had remapped Menu to Multi_key, so put that
in the .xmodmap:
-keycode 117 = Menu
+keycode 117 = Multi_key
That worked. And now? How do I use it? In principle, the article amusingly titled
Linux Compose Key Sequences appears to be what I'm looking for. But not all combinations work. It seems that things
differ here between XFree86 and X.org. All the documentation I found related to XFree86, and I
use X.org.
After some searching, found a file /usr/local/lib/X11/locale/en_US.UTF-8/Compose,
which appears to define multiple ways of composing a number of characters, including Greek,
Hebrew, Devanagari
and Hanggul. But they need special keys
to do many of them. The Greek characters, for example, appear to be only the exceptions,
and they require keys to generate the base characters:
<Multi_key> <quotedbl> <Greek_UPSILON> : "Ϋ" U03AB # GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA
<dead_acute> <Greek_alpha> : "ά" U03AC # GREEK SMALL LETTER ALPHA WITH TONOS
Still, it's a good start, but not all of those codes work either. For example, it says:
<Multi_key> <C> <equal> : "€" EuroSign # EURO SIGN
<Multi_key> <equal> <C> : "€" EuroSign # EURO SIGN
<Multi_key> <c> <equal> : "€" EuroSign # EURO SIGN
<Multi_key> <equal> <c> : "€" EuroSign # EURO SIGN
<Multi_key> <E> <equal> : "€" EuroSign # EURO SIGN
<Multi_key> <equal> <E> : "€" EuroSign # EURO SIGN
Only the first two combinations work for me, but e= and =e also work. But
not all the time. They don't work in Emacs, for example, though others do (the German
characters äöüß for example). There's still a lot to work out.
One of the irritating things about Hugin is that it doesn't understand the focal length information supplied by Olympus cameras. Today, I
decided, was the day to do something about it. That's simple enough: it pops up a window
like this:
So off looking for the text Focal length multiplier, which showed up
in src/hugin1/hugin/xrc/assistant_panel.xrc—an XML file. sigh. So
clearly I need to find where the program is when it produces the window. And that requires
a debug build. Hugin uses cmake, something that I don't know, and which puts me off
with its coloured messages, but after a bit of searching discovered that I could get debug
symbols with the invocation
How come? That command line clearly
specifies /usr/ports/graphics/hugin/work/hugin-2009.4.0/src/hugin1/base_wx/ImageCache.cpp.
And the file exists. What's it trying to tell me? And why does it happen? It seems to be
directly related to running with a debug build. Yet more head-scratching.
I'm quite happy with the sourdough rye bread I've
been baking for nearly a year now, but it occurs to me that it would do any harm for it to
rise a little more. And that should be easy enough to do: just make the third-level starter
a bit bigger, so that the yeast will cause more carbon dioxide. So today I made it a total
of 700 g instead of 500 g, adding 600 g of rye flour instead of 800 g for the final bread.
Result: no difference. Why?
Photo day again today. What do I do with my HDR panoramas? Without HDR, the shadows are
too dark. With HDR, the photos are blurred, and I use up gigabytes of disk for photos I
don't really need. Wouldn't it be nice if you could tell the camera how many photos to take
and how far apart the exposure should be bracketed? Yes, I can select 3 or 5 photos and
separations of 1/3, 2/3 or 1 EV, but I really only need two photos with a separation of 2
EV—I think. Tried that out today, doing it manually, and thus leaving more time than
desirable between the shots. The results were acceptable. Here the same panorama as HDR
and normal photo:
Yes, the leaves and flowers have clearly moved between the two exposures of the lower image,
but the overall impression is more uniform. I'm still undecided, but I'll probably do it
like this for a while.
Another attempt at debugging Hugin today. Jashank Jeremy had pointed out to me that the failing compiler invocation the other
day had a ; in the middle:
OK, but where did it come from? Spent lots of time looking through how cmake does its thing, involving much ktrace,
and found that the invocation gets put in multiple files called flags.make. But
where does this information come from? After an hour I still didn't know, so in the end
just faked it:
=== grog@dereel (/dev/ttyp9) /usr/ports/graphics/hugin/work/hugin-2009.4.0 75 -> for i in `find . -name flags.make`; do sed < $i >$i.foo 's:pthread;:pthread :'; mv $i.foo $i; done
After that I was able to build hugin, and caught the query for the focal length
relatively easily—a stack backtrace of fully 54 frames! I suppose that's modern now:
most of this stuff seems to be in the Gnome functions that handle the GUI. As usual, inside
any large program is a small program struggling to get out, and the code was relatively
clear. The following is in function SrcPanoImage::readEXIF () in
src/hugin_base/panodata/SrcPanoImage.cpp. CCDHeight and CCDWidth
are clearly the dimensions of the sensor, and normally they're used to calculate the
crop factor, the ratio of the
area of the sensor to the area of a 35 mm frame.
if (CCDHeight > 0 && CCDWidth > 0) {
// read sensor size directly.
...
} else {
// alternative way to calculate the crop factor for Olympus cameras
So it seems that Olympus EXIF data don't include CCDHeight and CCDWidth,
and they give the diagonal instead. But they've already taken that into account. I was
able to confirm that getExiv2Value () didn't return a valid olyFPD,
and with a little more searching discovered exiv2(1), Yet Another EXIF utility, which actually prints out these
StudlyCapsKeys:
=== grog@dereel (/dev/ttyp4) ~/Photos/20100327 19 -> exiv2 -v -p a house-e.jpeg | grep Focal Exif.Photo.FocalLength Rational 1 9.0 mm
Exif.OlympusEq.FocalPlaneDiagonal Rational 1 2160/100
Exif.OlympusEq.MaxApertureAtMinFocal Short 1 1024
Exif.OlympusEq.MaxApertureAtMaxFocal Short 1 1273
Exif.OlympusEq.MinFocalLength Short 1 9
Exif.OlympusEq.MaxFocalLength Short 1 18
Exif.OlympusEq.MaxApertureAtCurrentFocal Short 1 1050
So it looks as if the correct key was Exif.OlympusEq.FocalPlaneDiagonal,
not Exif.Olympus.FocalPlaneDiagonal. Presumably other Olympus cameras (compacts?)
use the latter key. With that, it was straightforward enough to work out a patch:
+ /* For some reason, Olympus E series cameras have a different key. */
+ if (olyFPD == 0.0)
+ getExiv2Value(exifData, "Exif.OlympusEq.FocalPlaneDiagonal", olyFPD);
+
if (olyFPD > 0.0) {
// Windows debug stuff
// fprintf(stdout,"Oly_FPD:");
But is this the correct calculation? It's all based on the (diagonal) crop factor, but they
seem to use it to calculate the horizontal field of view. The height of the image doesn't
enter into that calculation, and while they can calculate the angle correctly for specific
aspect ratios, the calculation breaks down when the aspect ratio changes. More
investigation needed.
Chris and David Yeardley over for dinner. At the last minute decided that we could drink
some red wine after all, but I didn't have any glasses, and I suggested that we
should—exceptionally—drink it out of the white wine glasses. I'll never do that
again, and in the end had to get up and get some red wine glasses—and they didn't use
them! There's no pleasing some people.
There's no doubt that the quality of service has improved on my satellite connection.
Before they gave me the current channel, my TCP test was giving a speed rating of 0.5
(meaning that setting up a minimal TCP connection took about 2 seconds), and the average
ping time was also about 1.5 seconds (something I didn't bother to record). Now the speed
is roughly 1 (1 second for the TCP connection), and the ping time is down to about 1 second
(here to freefall.freebsd.org):
round-trip min/avg/max/stddev = 793.850/851.411/905.639/31.968 ms
That's still not good: a satellite connection adds a latency of 480 ms, and a transpacific
ping takes about 180 ms, so the ping time should be about 660 ms. But it seemed good
enough to try to use VoIP again. The last
time I tried, with Wideband's VoIP,
things were so bad that I had to give up again. But I still have an account with MynEtfOne, so decided to try that.
Lots of issues with DNS and connectivity: I had to reset the IP address to be inside the
RFC 1918 /24 network that the modem
uses. But then it Just Worked. Called up Yvonne, who didn't
first twig to the fact that I appeared to be talking to her on the same phone she was
talking on, but with a 1 second delay. Apart from the delay, the quality appeared OK.
Hopefully it'll stay that way.
More pain burning DVDs
Yvonne asked me to burn a DVD for her today. Simple, right?
Every man and his dog can burn DVDs, and I have the additional experience of programming
streaming DVD recorders. There are so many programs that can do it.
But it wasn't that simple. The burner on Yvonne's computer didn't want to recognize the
drive. Is this an atapicam problem? I don't know, but the software on Yvonne's
machine isn't the newest. Still, not a problem. I have another drive in my test box. Put
it in there and got a message from Jörg Schilling telling me that this version of cdrecord would no longer burn DVDs, and that I
should buy Cdrecord-ProDVD. That was running as swamp, my FreeBSD, and I recalled having burnt DVDs before on
NetBSD, so tried that. Surprise! It had
Cdrecord-ProDVD on it, and it burnt a test DVD+RW with no problems, apart from what appear
to be spurious error messages:
Mar 30 12:28:41 kimchi /netbsd: cd0(viaide0:1:0): Check Condition on CDB: 0x43 00 00 00 00 00 00 00 0c 40
Mar 30 12:28:41 kimchi /netbsd: SENSE KEY: Illegal Request
Mar 30 12:28:41 kimchi /netbsd: COMMAND INFO: 1124073486 (0x4300000e)
Mar 30 12:28:41 kimchi /netbsd: ASC/ASCQ: Illegal Field in CDB
Mar 30 12:28:41 kimchi /netbsd: SKSV: Error in CDB, Offset 6
Put a DVD+R in for the final disk, and it took over a minute before it started writing.
When it was done, it took another minute for the drive to eject the disk, and Yvonne's
computer wouldn't recognize it. It worked in my DVD player, though. What's going on here?
Why is this such a pain?
More thinking about Hugin's field of
view calculations. Established that it reported a horizontal field of view of 87.75° for my
Zuiko
Digital ED 9-18mm F4.0-5.6 lens, so off to check what others said. The obvious place
is the manufacturer's site (the link above), which tells me that it's “100 to 62
Degree”, presumably at 9 mm and 18 mm. Checked other places, but it's difficult to
find specifics. I've already established that there are some very stupid claims out there.
Time to write my own program, which to my surprise gave
me an angle of 87.73° horizontal field of view, a very close correlation with Hugin.
So presumably it does take sensor aspect ratio into account somewhere. Added a few other
things to the program to print out information about other angles and crop factors. They're
quite interesting. Here the results for the lens:
Sensor details:
Width: 17.3 mm
Height: 13.0 mm
Diagonal: 21.6 mm
Area: 224.9 mm²
Width ratio to full frame: 2.081
Height ratio to full frame: 1.846
Diagonal ratio to full frame: 1.999
Area ratio to full frame: 1.960
So Olympus is reporting the diagonal field of view, something that's really useful.
It's also interesting to note that none of the crop factors are exactly 2.0, and the
all-important horizontal field of view of the 9 mm lens is less than that of an 18 mm
full-frame lens:
We don't seem to have done nearly as much work in the garden this year as I had intended,
but it has still changed beyond all recognition.
The Watsonias have grown greatly. Here
a detail from one of my weekend house photos,
as of 19 January 2008, 24 January 2009 and
23 January 2010, where they're in bloom:
There must be 100 bulbs in the meantime, and they're completely overflowing the bed. Time
for them all to go: the thyme is enough for the space. Pulled out most of them, but it was
tiring work, though Yvonne immediately planted some of them
in the new area to the east of the house. We're of divided mind about whether to cut off
the stems or not. I suspect that, like Irises, they should be cut off, but they still look
pretty viable, so we planted most of them with the leaves still on. We'll see how they
compare.
Another thing to go was the rest of
the Osteospermums in that garden.
They have a tendency to look straggly by the end of the summer, and we have plenty of other
things to put there, inlcuding some tulips and daffodils that I still haven't planted, and
of course some watsonias to come after they have finished flowering.
More playing around with burning DVDs today, and this time things worked fine. The strange
error messages still appear, and the copious commentary contains things like
Device seems to be: Generic mmc2 DVD-R/DVD-RW/DVD-RAM.
This was a DVD+R, but it said the same for the DVD+RW I tried before. And next time I had
the same problems as the first time. I wonder if the burner is dying on me. But wouldn't
it be nicer to have clearer error messages? I'll try a different burner; somehow I have the
feeling I have lost one.
Consensus is that it was a Ringtail
Possum. But surely the cats couldn't have killed it by themselves. In any case, it's
probably been doing some of the harm in the garden, so hopefully we'll have less of that
now.
More work on the Watsonias in
the north garden today. Yesterday I got 80% out; today I spent the other 80% of the time
getting the remaining ones out. They were entangled in the thyme and needed to be pulled
out one by one.
The roots of the thyme plants are on the extreme left of the plant; the Watsonias stopped
them from growing further to the left. There are really a remarkable number of them. Apart
from the ones in the basket, there are also a couple of clumps which I haven't disentangled,
and also the ones that Yvonne planted yesterday. I've
decided that there's no point leaving the shoots on: many are showing new shoots parallel to
the old, and I suspect the old one will wither away.
I don't know what they are, but they remind me
of Galanggal, so for the moment I'm
calling them the Galanggal mystery. I can't recall any other kind of flower amongst the
Watsonias, so maybe they lost the competition to flower. I'll plant them in a pot to see
what they do.
Millipedes everywhere
Everywhere I look in the garden at the moment, I
find millipedes. What do they eat?
I've seen them eating tomatoes, but the galanggal bulbs above had small holes in the bulb
which looks like some kind of parasite has been getting at them. Are they behind some of
the die-back I've been experiencing? Why are there so few products on the market to get rid
of them?
I've been really dragging my feet on picking my hops, but today I got some of the
Pride of Ringwood sorted out. It looks like I'll have more than enough for the next year,
though possibly I'll be short on aroma hops. It'll take a couple of days to harvest them
all.
What is it doing there? There was nothing in this part of the garden when we moved in.
About the closest would have been the pig sty,
but it's not pointing in an appropriate direction. Looking back to the house, it looks like
it goes back to the “bunker”, where
there is also a metal pipe still sticking out of the ground. It's made of PVC and is thus
relatively new, and it's also damaged, so clearly nothing is in it. Yet another of the
puzzles we have in this place.
Do you have a comment about something I have written? This is a diary, not a
“blog”, and there is
deliberately no provision for directly adding comments. It's also not a vehicle
for third-party content. But I welcome feedback and try to
reply to all messages I receive. See the diary overview for more details. If you do
send me a message relating to something I have written, please indicate
whether you'd prefer me not to mention your name. Otherwise
I'll assume that it's OK to do so.
