I've been keeping a satellite link statistics page for
some time now, and the results have been less than encouraging. Satellite links suffer from
high latency, of course—a round trip ping is limited by the speed of light to
480 ms. But that doesn't mean that the packets should get lost, and the latency I really
get out of the link is much higher. Here's what I've seen in the last 24 hours:
This is a good day, but lots of pings are failing, and the average time to download a small
web page is 2 seconds (1 / TCP speed). I've also had a ping -a running for the last
few days. The -a option means “beep if a packet is dropped”. The
continual beeping got on my nerves, and I stopped it:
--- www.lemis.com ping statistics ---
26330 packets transmitted, 25430 packets received, 3.4% packet loss
round-trip min/avg/max/stddev = 586.205/1180.124/5092.674/254.790 ms
Look at those times! An average of 1.18 seconds round trip time, more than double
the already slow 480 ms that the laws of physics impose, and well over 10 times what it
would be with ADSL. How any packet can take over 5 seconds to get from here to Canberra and
back is beyond me. Where has it been hiding? And 3.4% packet loss is Just Plain
Unacceptable.
We had already planned to plant some in the back of the garden, up against the paddock, in
the hope that maybe the kangaroos would develop a taste for them and leave other things
behind. Today I uprooted 25 seedlings from the patch (see where they came from? No, I
don't either; there must be a couple of hundred there), and planted them along a dripper
line:
They're not looking happy yet, but they're tough as old nails, and I suspect they'll make
it. If not, there are plenty more there, and I can be a little more gentle with them.
Yvonne also got into the act: we have
a Grevillea
rosmarinifolia up against the verandah, and the whole time we had been building the
verandah we had been considering removing it. It's prickly, and unlike
most Grevilleas it doesn't flower very
long, and the flowers are hidden by the foliage. So next spring we'll put something else
there, probably creepers that really grow. Yvonne thinks we might be able to transplant the
Grevillea, and despite my doubts I suppose it's worth a try, so today she had a go at
pruning it in the hope that a pruned tree would be more likely to survive transplantation:
Somehow felt restless and didn't do very much all day. Finally got round to looking at the
tutorials on the DVD that came with the digital photography special; they
don't look too bad, but boskoop, my old 400 MHz G4 Apple, is clearly too slow to
display them. Managed to get them to display on this flaky combination of FreeBSD, firefox and Flash—at least for the moment.
Also took a look at the software on the DVD. There's a new version of Ashampoo Photo Optimizer. I have been using version 1
for some time and find it quite useful. I
had already considered upgrading to the new version, but some of the advertising made it
sound less interesting. Trying it out today confirmed that impression: it now offers
brightness, contrast and gamma adjustments, but without any kind of numbers, and other
software already does that. And it has even more of a toy interface than the previous
version. Conceivably there's something there of interest, but I can't be bothered going
through the pain of mangling the interface.
A few bills to pay today with ANZ's emetic
“Internet” Banking. I've complained about this before, but it still infuriates
me every time I use it:
I have to enter the sum as a number of digits, followed by a decimal point and two more
digits. The customary commas every three digits, for example 1,234,456.00, are
not allowed. But that's the way the sum is displayed on the confirmation screen.
The dates must be entered in an ambiguous format with obligatory redundant
leading zeroes. For example, today's date is not, as you might expect, 2 February
2009 or 2 Feb 2009, and also not 2/2/2009: it must be
entered as 02/02/2009. Today's one of the few days where the format isn't
ambiguous: is tomorrow 02/03/2009 or 03/02/2009? You might say
“We're in Australia, so 02/03/2009 is correct”, and in this case
you'd be right. But firstly ANZ presumably has many American customers, who might make
mistakes with this format, and secondly this same web site has other Americanisms like
reverse chronological listing. So
why insist on such a stupid format, and what earthly use are the leading zeroes?
Two of my bills were for car insurance, which I pay by a system called BPAY. For this I need a biller code and a reference.
For RACV the biller code is a 13 digit
number, printed in the format 12345 1234 1234. It makes sense to split up a
number like that: it's easier to see if you make a mistake. But this horrible ANZ
software refuses to accept the spaces: you must enter a single 13-digit number.
It's non-trivial just to check that you entered it correctly.
Ten years ago I thought that people might listen to my suggestions about how to communicate
more effectively. Now I know they won't. Email messages which look more like random junk
are more prevalent than ever before, and I've had people who couldn't find what they were
looking for in the mess they created. Send two questions to somebody in the Microsoft space
and you can expect that they will answer only one.
One thing that I've become more and more aware of, though, is the inability to reply to the
people who are involved. It would seem that the Cc: header is unknown in the
Microsoft space—but it's not. All Microsoft and Apple MUAs seem to be able to handle
it well, even if Apple manages to put the text Cc: at the bottom of the box it
builds to make up for the tiny window it uses for the response. So why is it? In the last
couple of days I have had several mail exchanges with various people, where it's clear that
all involved should be kept in the loop—and in every case the responses I have
received have been addressed only to me.
You can't blame this on the software vendors (indeed, much of this behaviour is not the
fault of even such emetic MUAs as Microsoft “Outlook”): both
“Outlook” and Apple's mail offer equally simple buttons “Reply” and
“Reply All” [sic]. This seems to be a case of following the sheep in
front of you. But it's really amazing that almost everybody in the Microsoft space seems to
do it.
I've been using the term “Microsoft space” a lot lately, and today I spent much
of the day trying to define it. It's not easy; about the only thing that I can say is that
I use the word “Microsoft” in recognition of the leading part that Microsoft has
played in establishing this view of computing. It also includes Apple and much of Linux,
and it relates to some of the things I mention in my paper Why I hate OpenOffice. It's also much longer than
I intended, and I probably need to sit back and think about what I want to do with it.
Also spent some time listening to the tutorial on the GIMP on the DVD from c't's digital photography special. The
information is not bad, but it's a pity it relates to GIMP 2.4, while the software version
on the same DVD was 2.6.
More chopping of branches off the conifer, this time not because of any danger, but because
it looked funny. Also more work on protecting the flowers from the kangaroos; our Gazanias
are looking unhappy again.
Another power failure in the middle of the
night! I'm beginning to seriously consider solar power as at least a backup.
Up early to go to Melbourne to do our quarterly shopping and pick up Carola, who was booked
into the Hotel Spencer, a backpacker's hostel. First called up the hostel, where we were told
that she wasn't booked in. I asked if they had the correct spelling of her surname
(Schlanhof), and they looked again and found the booking, but said that she hadn't arrived.
So what were we to believe?
Off to Melbourne and to the hostel, in Spencer St. I suppose it's normal that that kind of
place is a dump, and this one lived up to it. After a while Yvonne found somebody, they checked the room where Carola should have slept, and
nobody was there.
Off to the Queen Victoria Market, which was
just round the corner, and called the number (in Austria) that we had been given by the
hostel; potentially it's her mobile number, and I got voice mail. Then (on the third
attempt) Carola got through to Yvonne and told her she was at the backpackers, and confirmed
the address (475 Spencer St).
Back to the hostel—still no Carola. Decided that the value of 475 might be a variable
and headed south. On the way, Yvonne got another call, from a German speaking man, who told
us that she was on the corner of Spencer St and Flinders St. Down there and picked her up.
It seemed that she didn't know she had ended up in the wrong hostel.
It would be nice here to give links to the “My map” that I created for the day's destinations. But somehow
there are a lot of straightforward things still missing from Google Maps, and this appears to be one of them. I wish
they'd fix them before taking photos of every street on Earth.
Then out to Mediterranean Wholesalers in Sydney Road, Brunswick, where the harmony was less
than perfect: Yvonne had bought a liqueur called “Perfetta Armonia” there on
previous occasions, but they no longer carry it. From there to Footscray, having dinner at
a Vietnamese place on Hopkins St called (I think) Dong Que. The food was acceptable, but
barely—Yvonne's “seafood noodles” had only three pieces of squid and
nothing else.
Then across the road to the market, where we were sorely disappointed. The fish was
uninteresting and not the freshest, and the vegetables were downright bad. I found some
completely soggy ginger in one place, but nothing much that we really wanted to buy,
except—accidentally—some green tomatoes. Green tomatoes are Mexican, and
normally you wouldn't find them in a south-east Asian context; but here they were on offer
as “ripe” tomatoes.
Home, annoying ourselves that the APCO petrol station in Lara had been offering petrol for
$1.069 a litre on our way to town, and on the way back they had gone up to $1.279. We
should have struck while the iron was hot.
In the evening, dinner on the verandah—the March flies are here already,
unfortunately.
For Carola's sake a wallaby, presumably the one we saw on Monday, showed up, and it was even
joined by a kangaroo, the first time we've seen both kinds of animal together:
Huevos rancheros for breakfast
today, making clear how horrible these Safeway tortillas are. They're so tough that you
can't even cut them with a serrated knife, and I ended up using a pair of scissors:
Only later did we realise that it was a day of Total Fire Ban, and we weren't allowed in the
forest—I think: I can't find any regulations that forbid riding during a Total Fire
Ban, but it does make sense.
It was too hot to do much outside today, so spent some time playing around with the idea of
using a single source file both for my diary and the
RSS feed. There's really only one file with two
different names, and the code recognizes what it has to do in the time-honoured way, by the
name. It required surprisingly little code; hopefully I haven't introduced too many bugs.
Another stinking hot day—about 45° and thus nearly as hot as last week. In Melbourne
it was even hotter, a high of 46.4°, compared to only 45.1° last week. This is also
apparently the highest temperature ever recorded in Melbourne, if the Bureau of
Meteorology is to be believed.
Total fire ban and a warning of conditions similar to those of Ash Wednesday. Another chance
to check out the appalling quality of the DSE bushfire site. One of the fires reported appeared to be at the Australian Equine Behaviour Centre (reported as
“Clonbinane Equestrian Centre”), run by Andrew Maclean. Kept a careful eye on
that one, for at least half an hour, after which it disappeared from the list! What
a disaster!
And the site is so slow, probably because of the continual refreshes that it performs, and
navigating it is such a pain! The only good news was that there were no fires in our area,
and that by evening the temperature had dropped to under 30°, though not as suddenly as
in
Melbourne, as this graph from the Bureau of Meteorology shows:
The horizontal resolution is 18 pixels per hour, or 3â…“ minutes per pixel. At 16:13(â…“), the
temperature was 45°. In 10 minutes it dropped to 31.7°, a drop of 1.3° per minute. And it
dropped almost as fast a second time later in the evening.
Part of the fire danger today was the result of the high winds, which made themselves felt
in the garden. They were so high that Hugin was not able to create one of my panorama shots: presumably the leaves were too different from one photo to
the next.
Round midday I turned on the sprinklers again (normally they run in the small hours of the
morning), but that wasn't enough, and a couple of hours later some of the plants were
looking decidedly limp: jasmine, tomatoes and even the Cape daisies, so I gave them another
go.
Spent most of the afternoon cooking—somehow I had underestimated how long it would
take. Carola is a near-vegetarian (she also eats fish), so I did a lot of Indian and Malay
food, including writing down a recipe for ikan
goreng and not yet writing down a recipe for deep-fried cauliflower.
Also tried out the rice cooker again, this time with Basmati rice. Complete disaster: it
stopped cooking in the middle, leaving the rice just a little cooked, and I had to complete
the operation manually. I suspect that the dish wasn't properly seated, and the temperature
sensing mechanism decided that it was hot enough. That unreliability makes it worse than
useless.
In the evening, Diane Saunders and her friend Lynn arrived from South Australia to spend a
couple of days; on the way they passed a serious bushfire near Horsham. Lynn is thinking of
buying a horse from Chris.
Finally it's cooler, and the news came trickling in about yesterday's bushfires. The
warnings were correct: the situation was not only as bad as Ash Wednesday, it was a lot
worse; by the end of the day we heard reports of 75 deaths, compared to “only”
47 in Ash Wednesday, and the number is sure to rise further.
Barbecue today—there was a Total Fire Ban, but seems that Total Fire Bans don't
include gas-fired barbecues close to a house, so we were in the clear. Apart from Carola,
Diane and Lynn, the Yeardleys came along with some friends from Singapore, so for the first
time we had a really large number of people—11—on the verandah.
That's not counting the flies, of course. We engaged in a typically Australian pastime:
Watching TV in the evening—“Rain Shadow”, apparently the last episode. It's not a very good series,
but it was filmed in Callington, not
far from where we used to live. In this episode, somebody went to great lengths to make a
fake map with new names for all the places in the area,
including Mount
Barker, which they called Blackford—not once, but twice, with different levels of
renaming:
Other places are Slate Hill (Wistow), White Flat
(Kanmantoo) and Paringa (Callington).
It's amazing that they'd go to such trouble for what was presumably a low-budget series.
Another bloody power failure! This is
getting to be really annoying.
Off to Sovereign Hill today for the first time. CJ and Sue are volunteer participants and got us in for half
price, and were happy to show us around. That was quite worth while: CJ knew all the things
that were worth seeing, and dragged us from one to the next at the right times. On the
other side, of course, he also knew the people running the place, with some unexpected
results for Carola, caught panning for gold without a license and clapped into irons:
Normally it's only the staff who get this kind of treatment.
One of the more interesting things I discovered was
that Lola Montez, better known in a
Bavarian context, spent some time in Ballarat and caused quite a stir here as well. Saw a
reenactment of the case where she whipped the editor of the Ballarat Times:
On the way out, Yvonne managed to cause a stir by tripping
over a box in the main street. The methods of treating her scratches seemed a little
draconian:
Into town to talk to Peter O'Connell today, taking Carola with us to do some shopping.
After that along to Avalon Nursery,
only to discover that they are closed on Tuesdays. To Mount Buninyong instead, braving the
drizzle to climb up the lookout tower; even from the top, there's no panoramic view. A
little disappointing.
Jason (I think) and mate from Skybridge in this evening to install a new, improved satellite modem—I thought. What they
brought with them was another IPX 3200.
I told them to take it away again, but they told me that they had already registered the
changeover with IPStar, who had closed for the
day a few minutes earlier, and they couldn't get any other modem to work until tomorrow
morning at the earliest. Contacted Wideband and spoke to Matthew, who told me that the new Beta software (RC 018), not exactly
reassuring terms, was much better than before, and that they had had no more problems with
data corruption. So they installed the thing. At least it seems to work—so far.
We'll see if this problem with the BST is still there. One way or the other, I'm not at all
happy, and I made that clear when signing the docket.
Jason told me that the web server now worked with other web browsers than Microsoft
“Internet Explorer”, but I can't see any evidence of that. Even with
“Internet Explorer” it doesn't work right, but then my “Internet
Explorer” appears to have been installed without my permission by Telstra's wireless software, and for some reason it
kept bringing up the wireless menus, though there's no longer an interface there. What a
mess this Microsoft is!
Into the office this morning to discover that the satellite modem didn't hang, at any rate,
but the display suggests that the connection is now even slower:
It's difficult to see, but it looks as if the “TCP speed” has dropped slightly
after the installation.
The reliability of the link is also not significantly better. On installation, I started a
ping to my external web server, and stopped it about 18 hours later:
Those duplicates look like a worry; I haven't seen them before. And there's still
2.7% packet loss, and the average ping time of 1.28 seconds is really unacceptable even on a
satellite link. I think it's time to find an alternative to this form of Internet access.
The temperatures are still low—today we had a maximum of 15°, which is 32° less than
the all-time high only a couple of weeks ago—and the bushfires are gradually less of a
threat. They've brought a surprising number of people out of the woodwork, though:
yesterday Yvonne's sister Vera wrote to hear what was going
on, and today I got a phone call from Conny Wölk, with whom I worked about 15 years ago.
He's now in Switzerland, half retired. He reports of a surprising number of top-notch
people we know who are disillusioned with the current computer marketplace (and that's what
it really is!) and have found alternative things to do.
I've had many occasions recently to complain about the SBS web site. One ongoing problem is that the link for the “Regional”
programme took me to a page for Tasmania with the emetic, space-embedded URL http://www.sbs.com.au/schedule/2009-02-12/SBS Tasmania. OK, I can guess past that
breakage, and I've been downloading a page http://www.sbs.com.au/schedule/2009-02-12/SBS Regional instead. But lately that
hasn't been working; I get a page with three sections saying “No program [sic]
data available”.
Today I called up SBS by phone (not an easy number to find; true to webmaster form, never
the twain shall meet. For future reference, the number is 1 800 500 727) and was told that
there is no difference between the programmes in various parts of Australia. That begs the
question why they make this distinction in the first place; possibly it's yet another
indication that web masters live in a parallel universe. She told me that she would report
the problem; I won't hold my breath to see it fixed.
Off early today to the Great Ocean
Road. After some discussion, didn't go to Otway
Fly, but tried Melba
Gully instead. It's a bit run-down, but quite worth visiting. I think I've decided I
don't like the commercial atmosphere of Otway Fly, but it's not clear whether Maits Rest or Melba Gully is better. Melba Gully probably wins because it's much
easier to get to. One annoying thing was the mention of “Triplet Falls” on the
information at the entrance, with no indication of where it might be:
There was also no mention of the place in any of the maps, nor anywhere else we looked. It
wasn't until I got back home that I was able to search the web and find the Visit VictoriaTriplet Falls page—another one with a URL not intended for human consumption.
It's just on the other side of Lavers Hill, on the road to Beech Forest—even the
“Visit Victoria” page refuses to divulge how far. So if we want to see it, it's
another good 180 km round trip and a bit of searching.
Then on towards the 12 Apostles, finding a wildlife park on the way where Carola could
finally stroke her wallabies:
This is now the third time I've been to the Apostles, and it's gradually becoming clear that
there's not much to see there. They're very impressive, of course, but they don't change
much. Took the opportunity—again!—to take photos with my new telephoto lens,
this time another one. The photos aren't bad, but with
the exception of better clarity on some shots, they're not much different from last time.
Then on to Port Campbell for lunch—at 15:30—and got some fish and chips at a
place called “Splash” on the east side of the main street—the food was
considerably better than what we got
at Apollo Bay18 months ago.
Back via Camperdown, and
again up to
Mount
Leura to take a look around the surrounding countryside. This time the weather was
good, but it's surprising how difficult it is to find a lookout that doesn't have some
obstacle somewhere in the 360° view.
My fears about the new satellite modem have confirmed themselves: it's showing the same
problems of TCP hangs that I've seen from the very beginning. My link monitoring software shows:
This is a very specific issue: I'm unable to set up new TCP connections, but existing ones,
even to the same site, work with no problems. ping speeds (ICMP) and UDP are also
unaffected. Looking at the raw data for the same time, I see things like:
1234514699 3.09 # Fri Feb 13 19:45:02 EST 2009
1234514767 75.01 # Fri Feb 13 19:47:22 EST 2009
...
1234515894 75.01 # Fri Feb 13 20:06:09 EST 2009
1234515962 75.01 # Fri Feb 13 20:07:17 EST 2009
1234516029 16.94 # Fri Feb 13 20:07:26 EST 2009
1234516097 2.17 # Fri Feb 13 20:08:19 EST 2009
1234516166 2.58 # Fri Feb 13 20:09:28 EST 2009
The time 75.01 is the time it takes for the TCP connection attempt to time out; currently my
scripts have a bug which makes it difficult to determine when the connection fails, but they
did here. There is a difference, however, in this behaviour: it recovers from the
situation. With previous firmware revisions, only a reset would help. Now you just need to
wait for 20 minutes of no network access, and Voilà ! it comes back again.
What a crock!
Matthew's promise of less data corruption also doesn't seem to have come to fruition:
5257 files to consider
Received disconnect from 203.10.76.45: 2: Corrupted MAC on input.
rsync: connection unexpectedly closed (9 bytes received so far) [sender]
rsync error: unexplained error (code 255) at io.c(632) [sender=3.0.4]
Fri Feb 13 09:27:33 EST 2009
Called up again and spoke to Nathan, who wasn't able to access the modem either. He
promised to investigate and send me mail on the outcome. He also came up with a new excuse
for some of the breakage: “The web browser CGI was written for Microsoft Internet
Explorer”. So it's not a bug, it's a feature. The sad thing is that this probably
means that people have no intention of fixing the bug.
High time for more rain. Other indications are the state of the shade trees to the north-east of the house. One in particular
looks as if it has had it. Here the appearance a year ago and today:
More pain with the satellite modem today. How can anybody install such a load
of junk? Called up Wideband and spoke to
Jay, who told me that the firmware version I had (RC 018) was not the best—in direct
contradiction to what Matthew said on Tuesday—and that they would install RC 029 on
Monday. Who knows if that will bring any improvement. I am getting thoroughly fed up with
these people. At the beginning they were helpful, but it's unacceptable that such basic
problems as continual loss of connection and the inability to communicate with the modem are
still with us after over a year. And this excuse “the modem is designed to work with
Microsoft ‘Internet Explorer’” is an insult to any normal definition of
interoperability. I've never seen a site on the web that would only work with
“Internet Explorer”. I think that one way or another, a formal complaint is
necessary.
Also mail from Peter Nicholson, enclosing a graph of data from his IPStar link. He's with Aussie Broadband, arguably a
different ISP, so the comparison of his data (left) with mine (right) is particularly
interesting. Unfortunately he doesn't have the TCP data.
Both these graphs are from the same time frame, so you'd expect the dropouts to be the same.
But they're not, not by a long shot. In particular (and rather unusually), I didn't have
any link dropouts during that period, while Peter had 15. So maybe this does have something
to do with the ISPs after all. I wish a few more people would send me their data.
Carola left today to visit Laurel Gordon in Tasmania. Problem: she had some seeds with
her, Eucalyptus
and Calendula. They were to take back
home, but there are quarantine regulations in Tasmania, and we didn't know whether she could
bring them in to the state. Took a look at the AQIS web site, where I didn't even try to find out more than a contact phone number.
That proved to be easy enough: it's 1-800-020-504. And it's only open during normal office
hours.
With a bit more effort found the airport phone numbers. Melbourne is 03-8318-8200 and
continually engaged. Got through to Hobart on 03-6214-6020 and discovered that there are,
indeed, restrictions. It's up to the individual AQIS officer whether he's prepared to let
it through on the understanding that they'll not be unpacked in Tasmania, and also whether
they're prepared to hold them until she leaves again. How completely unacceptable.
Then we had to go off riding before Carola left, and even Yvonne was getting a bit tired. Smoke haze on the way back home—the weather forecast had
warned of that, presumably from last week's bushfires—but I
didn't like the way the haze seemed to be concentrated in the area coming from our house, so
off back home to find, of course, that nothing was wrong. It's strange you can't smell this
smoke.
This satellite connection is appalling! It must be an order of magnitude
worse than when I got rid of it a year ago:
I count a total of 86 failed TCP connections in the data files for the graphs above, meaning
at least 90 minutes of inability to communicate. During that time, as before, existing TCP
connections worked normally, and ping showed “only” 1.5% packet loss. This
upgrade to the next flaky firmware had better show significant improvements.
The kangaroos are learning to tip over the wire mesh we have put around many shrubs, and
they've eaten all the leaves off the oak sapling yet again. Time to connect it to the
electric fence, I think.
I've been spending far too much time lately trying to choose a new tripod for my camera.
How difficult it is! They all seem to be too short, and I can't find any useful information
about the heads at all. Why should a ball head be better than one with levers? How do you
pan with one? Maybe that's all taken care of, but no documentation describes it.
Today somebody posted a good example of the extremes of modern GUI software, apparently Microsoft “Word”. I suppose
that's just the result of extremes of configuration, but it certainly makes a point.
And of course the link is now dead, or at least non-responsive.
The problems with the satellite modem continued all morning, and finally I called up
Wideband, spoke to Michael, and asked him
when the firmware upgrade would be done. He promised to call back, and did so, saying the
modem would be upgraded within 30 minutes, and that it could be off the air for up to 30
minutes. It really did happen within the time frame, but the “off the air” was
so short that it hardly registered in comparison with the “normal” dropouts:
The update was the last dropout, just before 12:30, and it lasted only about a minute.
The only immediate result of the update was the confirmation that the web application is
still broken. My copy of Microsoft “Internet Explorer” wouldn't look at it
either, just wanting to save it to disk. Decided to upgrade to the latest version, which
was fraught with difficulties. The “Software Update” appeared to hang, so I
went to the Microsoft support
site and follow the link to “Internet Explorer 7”—which came up in
German! It also didn't give me the option of downloading a different version. Set the
country to Australia and tried again—still the German version! But this time I got
the option of downloading the Australian version, and did that.
The pain wasn't over—it then went out looking for “Internet Explorer”
updates. After about 10 minutes of no activity, installed wireshark, which showed no obvious external
activity. But I couldn't stop the search! The stop functions were unclickable. In
disgust, put the box to sleep.
In the meantime, took a look at why this page doesn't work on any other browser. It proved
to be written in relatively legible
Javascript. Callum Gibson pointed me at the “Error Console” for
firefox, which told me
(requiring me to painfully copy each message):
Warning: Expected identifier for pseudo-class or pseudo-element but found ' '. Ruleset ignored due to bad selector.
Source File: http://sat-gw.lemis.com:8080/cb.css
Line: 5
Warning: Error in parsing value for property 'CURSOR'. Declaration dropped.
Source File: http://sat-gw.lemis.com:8080/cb.css
Line: 26
Warning: Error in parsing value for property 'CURSOR'. Declaration dropped.
Source File: http://sat-gw.lemis.com:8080/cb.css
Line: 29
Warning: Error in parsing value for property 'FILTER'. Declaration dropped.
Source File: http://sat-gw.lemis.com:8080/xWebGateway.cgi
Line: 0
Warning: Unknown property 'moz-opacity'. Declaration dropped.
Source File: http://sat-gw.lemis.com:8080/xWebGateway.cgi
Line: 0
Error: uncaught exception: [Exception... "Not enough arguments [nsIDOMHTMLTableRowElement.insertCell]" nsresult: "0x80570001 (NS_ERROR_XPC_NOT_ENOUGH_ARGS)" location: "JS frame :: http://sat-gw.lemis.com:8080/xmlcode.js :: update_tab_control_view :: line 220" data: no]
What does that mean? Clearly the last is a bug, probably the reason the thing doesn't run.
Does “Internet Explorer” simply ignore the bug? And I suppose it's typical of
this kind of broken software that cb.css contains:
The following error was encountered:
Unable to determine IP address from host name for www.dcita.gov.au
OK, that's common enough on this flaky link, but this time the problem persisted. Further
investigation showed that the agency has changed its name to the even more emetic Department of Broadband, Communications and the Digital
Economy (DBCDE). “Department of Broadband” indeed! Who's responsible for
Narrowband? Who's responsible for narrow minds?
And they have taken down the old web site, instead of redirecting! They still
have an archive web site that could
have handled a redirect. In fact, it would be even simpler than that: both the DCITA archive
site and DBCDE are running on the same machine (Apache/Red Hat), so it would have been as simple as
a VirtualHost entry. What stupidity! They're still in the phone book (with the new web server
name, which I suppose they think is a URL), so it's clear that they don't think of the web in
the same way as the phone system. How can I expect to get any technical sympathy for a
department which so clearly shows its complete and utter incompetence in technical matters?
Gave up in disgust.
So far it looks as if my new satellite modem firmware is working better than the old one.
Finally managed to install “Internet Explorer” release 8, which worked, but then
came up with the startling message:
Internet Explorer 7: Jetzt herunterladen-Telstra BigPond Home Internet Explorer.
Where on earth did it get that from?
After I got it working, it still offered to save the file from the satellite
modem—this was a clean install, with the exception of importing defaults from
firefox. Reset everything, didn't
import the defaults, and it finally worked. Even more surprisingly, I checked
the configuration and found NetBIOS
enabled. No idea what for, but I know I don't need it, and it's a potential security issue,
so I disabled it, forgetting that I'm in the Microsoft Space, so it rebooted.
To my amazement, the TCP sessions survived the reboot. Yes, they'll do this on a real modem
too, but the modem maintains state for BST, so it's to be
expected that it loses this state when rebooted. I wonder if it has found a way around
that, or whether it wasn't a real reboot.
More horrible GUIs
People have shown me some more horrible GUIs. Time to create a page to collect them.
In the afternoon, more research into tripods. Documentation is still an issue. Finally
came across a supplier in
Shanghai with at least well-documented offers at prices that don't look too bad, with
an amazing diversity of prices and pictures which ultimately proved to be the same
tripod offered in three different currencies and with three alternative combinations of
purchase price and shipping costs, as well as Buy It Now and auction options. The prices
ranged between $49.90 and $120.59, but after factoring in the postage, they all came to the
same thing. Normally I'd avoid a vendor like this, but the feedback was surprisingly
positive, in particular in relationship to the quality of the product, so bought one.
Which price? There's a refund option if you don't like the product, but of course you lose
the shipping costs. So the free shipping seems to be the way to go, especially as it was $1
cheaper. Hopefully things will be OK.
Also did some investigation of telephoto lenses. I need something that goes beyond my
current 70/300 mm lens, and to make it worthwhile it should be at least 800mm. At that
length, there are a few options:
A real telephoto lens with autofocus and automatic exposure. B&H offer only one such lens, a Sigma 300-800 mm zoom for $7,199—way out of my price range.
A cheap telephoto lens with manual focus and manual exposure, such as the Samyang 650-1300 mm
lens I have seen around. Apart from the obvious problems of manual focus on a modern
DSLR, which isn't really designed to help, this thing has no diaphragm, so you can't
even stop down for more depth of field. It's claimed to be “f/8-16”, but
that's dependent only on the focal length. The price is OK, but the reviews I read of
it were horrible.
Use a teleconverter. That sounds like a good idea, but Olympus teleconverters are very
expensive, and the strongest one is only 2x, so I wouldn't get beyond 600 mm. It would
also drop the maximum aperture by 2 stops.
Use a screw-on magnifier lens, which fits on the filter thread of the lens. This
maintains automatic focus and exposure, and doesn't (I think) alter the maximum aperture.
They're cheap, but I've read bad reports about them too, though not as bad as the Samyang
lens.
Stick to what I have now, using a Hanimex Pentax thread 300 mm f/5.5 lens with 2x and 3x
teleconverters, which even gets me as far as 1800mm f/33. This one has the great
advantage that it costs me nothing, but the disadvantages of alternatives 2 and 3, and
also the disadvantage I've already noted that the photo quality is worse than “digital zoom”, just using the Olympus
lens at 300 mm and cropping.
Decided that the only reasonable alternative at the moment was alternative 4, and found a
suitable-looking lens for $129 on eBay from
onlinedigitalshop. Send
a message stating what I want to use it with, and received a reply telling me it should be
fine, so bought one. I'm still not sure that was the correct thing to do, but we'll see.
It, too, can be returned, limiting my exposure to about $30.
For reasons I forget, I was reminded of a series of rants I had with Fritz Jörn, starting over a quarter of a century ago. Fritz
has some good ideas, but in comparison with me he's GBS's “reasonable man”:
The reasonable man adapts himself to the world; the unreasonable one
persists in trying to adapt the world to himself. Therefore all
progress depends on the unreasonable man.
-- George Bernard Shaw
Fritz was the marketing manager for Tandem Computers GmbH in the 1980s, and he sent out
various press releases by email—not for reading, but for printing. The result is
one of my oldest rants.
Mail today from Max (surname unknown) from Albany WA, identifying a few more of my
mystery plants. Gradually we're coming
closer to solving the mysteries—only 5 left, and one of them is half solved.
More problems with the satellite connection today:
Just before the end of the first outage, called up Wideband, and Nick was unable to confirm what was
wrong, but suspected that the southern beam for Victoria was out. So nothing to do with my
satellite box, just general unreliability. I wish these people would get their act
together. And this time, of course, my TCP sessions went to hell, so yesterday's experience
must have been related to a soft restart.
The kangaroos are becoming a real nuisance; today found that they had attacked
the Hardenbergia on the centre
column of the verandah and defoliated it up to about a metre. They had also tried to pull
away the stem, but fortunately it was wired on, so the stem and the upper leaves survived.
The wire was stretched by about 10 cm, though. I suppose I should plant a large surface
of Gazanias near the fence: they grow
like fury, and the kangaroos like them. First we need some rain, though.
My free copy of “Beautiful
Architecture” arrived in the mail today. After I submitted the last set of
corrections, they didn't send me a new draft, and I had feared the worst. With
justification; there are still things in there that I had tried to correct several times.
This isn't the first time I've had trouble with the O'Reilly production team. It's really frustrating.
I suppose there's some grounds for cautious optimism that the latest version of the
satellite modem firmware is working better; I haven't seen any further evidence of purely
TCP hangs. From that point of view, I now have a new modem which makes less noise, but is
otherwise no better than the one I had before. And the packet loss rate is still
unacceptable:
Time to do some work in the garden, but it's so hot again. We still have temperatures in
the mid-30s, and the rainfall this year has been negligible, only 8.4 mm:
The bad part are the 10 and 30 day rainfall averages (red and green). We desperately need
rain!
How did that happen? Is it serious? Inside is a metal connector, and the thing still seems
to work, but for how long? It's still under guarantee, so called up Kenwood service at 1
800 126 659 and got the phone number of their Ballarat repair people, John Thomas & Co.:
5331 3099. Doubtless it will not be too difficult to fix, but it has shaken my trust in the
brand more than a little.
The weather was more bearable today, in fact quite pleasant, with the exception of these
damned March flies. I did some work in the garden adding more irrigation—the drought
has shown clearly where not enough water is getting through—and had to carry a can of
fly spray to ward off the flies.
How to kill a kangaroo
How do you kill a kangaroo that you don't see, or at least get rid of the bastards? As the
weather gets dryer, they are becoming more and more invasive. This evening I saw three of
them in the garden, not 10 metres from the verandah, and chased them away before it occurred
to me that I should have got some photos. They didn't go happily, either—it took
Yvonne a separate attempt to chase them back to the swamp.
They're doing a lot of damage to the garden. I've thought of poison, electric shocks and
shooting them, but all seem to be a lot of work. There must be a simpler way.
Why does my weekend photo processing always take so long? It should be becoming more
routine now. But somehow I still end up spending 2 hours every Saturday.
Our next building project is already on its way, building another shed for horses and hay.
The way the weather's been lately we'll probably need to plan for more hay than we had
hoped. Round to CJ's place to take a look at one he has built; he also has some second-hand
building materials that we should be able to use.
Killing kangaroos is probably not an option, but Juha Kupiainen came up with one that
potentially might be: drive them away with ultrasonics. The web site is rather strange (no pricing or other information, just
“enter your phone number in this form in the format that we think is right”),
but I'm still dubious after my experience
with kangaroo whistles. Still, it's worth following up on.
A bit of culinary experimentation in the evening: we have this Vietnamese mint, but we don't
really know what to do with it. Faked up a chicken and pine nuts recipe, which tasted surprisingly good. I won't put any thyme
in next time, though.
ABC Gardening Australia has a new
moderator since the end of last month, Steve Ryan. Today he opened up his garden under the
open garden scheme, and we decided to go
and have a look. Just finding the place was difficult: he lives
in Macedon, on the road from
Melbourne to Bendigo, but all the
documentation put the address in Mount Macedon, so of course
Google Maps didn't find it. The Open Gardens web site included a Melway reference, but it proved to be a
small-scale map that covered Bendigo to Ballarat, and was completely useless for finding the
street. Finally found it in VicRoads, which
didn't do a very good job either—we found at least three errors on the way, roads
which were supposed to be there but which were not.
Found our way there without much difficulty, and, as I had feared, found that we weren't the
only ones—they came in their busloads, thus solving the problem of finding the place.
Spent about an hour looking around and getting a number of ideas for our own garden. Also
noted that not everything that Steve touches is automatically successful. This one appears
to be a Buddleja globosa (the
dried-out flowers in the middle of the photo). It does very well in our garden:
He also had plants for sale, at horrendous prices. To be fair, he specializes in rare and
exotic plants, and they have their price—beyond what we currently want to pay. We
were quite impressed by the pond, and Yvonne liked some of
the succulents:
Took a different way back, via Mount Macedon, Woodend
and Daylesford, mainly
because I've never been there before. Discovered that Daylesford has a reputation of being
a gourmet centre, and as we went through—at 13:00—found that the weekly Farmer's
Market was still in full swing, so in to take a look:
Yvonne found some of the succulents she liked the look of, as well as others, and we also
found some Victorinox kitchen knives, so we
were $50 poorer by the time we left.
Back home and spent some time assessing the work we need to do to improve the layout of the
garden.
I've complained frequently about the ridiculously low speed limits in Australia, and
particularly in Victoria: on a straight, multilane freeway such as the Prince's Freeway
between Melbourne and Geelong, you are
typically limited to 100 km/h. On less well built roads, such as the Gippsland freeway,
which is only two lanes and somewhat twisty, you're allowed 110 km/h. You'd think this were
contradictory, but it seems to be part of a plan. When it's icy, you're called upon to
drive at 40 km/h:
I suppose that says more about the level of ice that they expect than anything else, but in
a state where the drivers are bludgeoned into obeying what the authorities want (today we
were continually stuck behind people doing 20 km/h or more below the speed limit), it's
downright dangerous.
It's still hot! And today they were predicting more bushfire activity, though it didn't
seem nearly as bad as two weeks ago. Still, we did get 5
new bushfires, including one near Daylesford, where we passed through only yesterday. If
only it would rain! I did note that there was some rain almost exactly where the Daylesford
fire was, but it was almost certainly not nearly enough.
Doing some idle thoughts about electronic flash. How do you
measure the intensity? For portable flashes, such as on-camera, it's a
guide number, which makes perfect
sense to the person using it manually, as long as the manufacturer hasn't lied about the
conditions. But studio equipment measures the power input to the flash tube, in the strange
“unit” watt-seconds, a reinvention of
the Joule. That's a more honest but less
useful measure: you can be sure it's correct.
But how do you compare units rated in these ways? There are many steps between the power
input to the tube and the light hitting the subject:
First it needs to be applied to the flash tube. Most units can apply only part of the
power available, but that's not as important as the maximum power that can be applied.
The tube converts some of this power into light, some into heat, and some into sound. The
proportion of light is important, but I've never seen any numbers.
Then the light has to be directed at the subject. In general this is done with
reflectors, but they can be of different materials, and they can be direct or indirect,
especially with studio equipment. Reflectors can be adjusted (usually only
approximately).
Round about here things start getting more complicated. Guide numbers also take this
sort of thing into account—it's one of the ways that manufacturers lie about the
output of their units.
So how do you compare them? If you're on a forum,
you say “you can't compare” or “This is like converting weight of a car
to miles per gallon”. This really means “I don't know how to compare
them”, and it's no help to anybody.
Clearly there is a correlation, if you define your reference conditions. For
example, for a specific reflector/diffuser setup, a studio flash will have both a guide
number and a power input, and with the appropriate factor you can calculate one from the
other. It's not a direct conversion: firstly, Joules are Joules, while guide numbers depend
on the unit of length in use, and they're proportional to the square of the light output.
So if, for sake of argument, 100 J corresponds to a guide number of 30 metres at 100 ISO,
400 J correspond to a guide number of 60 metres (or a little under 200 ft, if you're still
using those units) at the same sensitivity.
Still, it's worth having a rough idea. I've checked the specs of some units and found:
A 150 J unit with a guide
number of 30, implying a guide number of about 21 for 100 J.
A 1000 J unit with a guide number specified only as 394 ft, though it's clear that
was converted from the rounder sum 120 m. That implies a guide number of 38 for 100 J.
A 200 J unit with a guide number of 115 ft (35 m), suggesting a guide number of 25
for 100 J.
My own studio flashes, rated at 110 J or 32 m, implying a guide number of just over 30 for
100 J.
So I assume that the conversion 100 J = guide number 30 is a good rule of thumb. It's not
much more than that: when taking close-up photos, I put my lights at a distance of about 1.5
metres, and I typically use f/11. At that distance with two lights with a guide number of
32 (combined guide number 45), I'd expect to use about f/30. But that's bounced off a
couple of umbrellas and diffused through a light tent, by which time the “guide
number” for an individual unit has dropped from 30 to 12. There's a reason why guide
numbers don't work in that environment.
Also more thoughts on how to structure the garden. One thing I need to factor in is that
this will, of course, take years. How do I ensure that it looks good in the meantime? One
of the things we're also considering, of course, is the choice of plants. One thing that
has become abundantly clear is that we need wind breaks, and we'd like something that
attracts birds, but I can't find a good overview. Is
Grevillea tetragonoloba
as good a choice as its sellers claim? Would it make a good hedge?
In the evening, we had another attack of
Corellas, unfortunately too late in the
evening to get good photos:
People can say what they want about resolution, but it's time for better low-light
performance. These photos were taken at 1600 ISO, and the quality is only barely
acceptable. It's time for sensors with 100,000 ISO (and not 102,400 ISO, as the current
trend seems to be) with better quality than this.
All three of my recent eBay purchases are now
here: a memory card from Hong Kong arrived in the post, and the tripod and the supplementary
telephoto lens have arrived at the
Sebastopol post office.
It's interesting that the memory card took a week from Hong Kong, while the tripod took only
5 days from Shanghai, and the lens, from Queensland, took about the same time.
So today looked like a good day to go into town and do various odds and ends that were
waiting. Had a haircut, picked up the parcels, and took the Kenwood mixer to John Thomas
for repair, where they told me they thought the thing had been dropped, possibly before I
got it. That sounds unlikely to me; certainly we never dropped it. Strangely, they weren't
certain about the warranty, though it clearly stated “6 months manufacturers
warranty”. They promised to get back to me, but didn't.
Back home and unpacked my new toys, which weren't as satisfying as I had hoped. The tripod
cost 5 times as much as the old one, and though it's probably better, there's not much in
it. I still don't understand the advantage of a ball head; it makes panning much more
difficult. Maybe I'll buy a different head.
The supplementary telephoto lens was a whole different matter. It was a complete disaster,
an order of magnitude worse than my worst fears. It has extreme chromatic aberration and
flare, and it can't focus to infinity. The problems are evident even in most of the
thumbnails below; click on the photos to enlarge them and the enormity of the problem
becomes apparent. To quote “Andys”, one of the semi-anonymous people on IRC,
they look like they've been taken through the bottom of a coke bottle.
The photo above (whole frame, then a detail from top left) was taken at maximum focal length
(3x300 mm). It shows unbelievable aberration. The one below (whole frame, three details)
was taken at minimum focal length (3x70 mm). It seems to be worse at the edges: the last
detail, from the middle, is not quite as bad.
The furthest focusing distance appears to be about 15 m. Here it is set manually; it can't
quite get as far as the post. Beyond this point, the lens no longer focuses.
That kills one of the advantages I
attributed to this approach, that I could still use auto-focus. But after focusing
manually, it's still much worse than the Hanimex lens I already have, the one that I was
trying to improve on. And I know I can improve on the Hanimex: just use the Olympus
telephoto lens by itself and crop the result (“digital zoom”). The following
photos show:
The photo taken at 3x300 mm with the supplementary lens, manually focused
A photo taken with the 300 mm f/5.5 Hanimex lens with a 3x teleconverter. This photo was
taken on a windy day with a shutter speed of 1/20 second, so it's not even as sharp as it
could be, and it's clear that there's a colour problem as well, which I would normally
compensate for during postprocessing. But it's still much better than the new lens.
A photo taken with the Olympus lens alone at 300 mm, cropped to simulate 900 mm. It was
taken the following day, so the flower doesn't look quite the same, but the difference in
sharpness is obvious:
In summary, then, the supplementary lens is worse than useless. It makes worse photos than
when it's not used.
So how could this happen? Surely nobody can be happy with this kind of
“quality”. What good is a telephoto lens that can't focus further than 15 m?
It must have something to do with the lens I've attached it to, something I thought of in
advance. Thus this mail exchange with the vendor (tidied up from eBay's usual emetic format):
> I'm interested in this item for use with an Olympus E-510 and 70/300 Olympus telephoto (58 mm filter
> thread). I've heard some horror stories about supplementary lenses; do you see any problem with
> this combination?
The 3x Telephoto Lens will fit to the 58mm Lens with an adaptor which will be included in the
package. There will not be any problem. We have not received any complaints from our past customers.
So I wrote another message asking for suggestions, and got the startling reply (original
format):
The 3x
Telephoto Lens will fit to the 58mm Lens with an adaptor which will
be
included in the package.
Clearly another case of inability to read email, and also an indication that this is a
standard response. I sent another message back again. There's still no need to leave
negative feedback, nor even to mention the name of the vendor; it could be a simple
misunderstanding, though he's stretching my patience. How many people try this sort of
thing on a long tele lens? Still, that's why I asked the vendor, and he'll need to do
something.
More playing around with lenses today. Got another message from the vendor of the broken
supplementary lens, not offering any explanation, but offering a refund. I suppose I'll
take that, but I'm still not happy. I don't have a standard lens on my Olympus with which I
can test the supplementary tele—the filter thread on the ZUIKO DIGITAL ED
12-60mm F2.8-4.0 SWD is 72 mm—but my old Pentax Z-1 with the 28-85 Takumar has
the right size thread, so put it on there. There's no film in it, and I wouldn't buy one
just for that, but it's clear that the lens wouldn't work well there either: there's extreme
vignetting up to about 60 mm, and beyond that the same chromatic aberration patterns show
up.
Also found a couple of old video wide angle supplementary lenses that I had bought decades
ago. One of them—a fisheye—also fitted the Pentax, and from what I can see the
quality was better. I think that's the answer: these lenses are designed for video cameras,
where they may work well, but this vendor is advertising them for digital SLRs, and they're
a catastrophe there.
Yvonne into town, bringing back Natalie, a friend from Byron
Bay, and an Olympus EC-20 teleconverter kindly lent to my by a friend in Alice Springs. That gave me the opportunity
to do a reasonable comparison of the various options at my disposal:
The Olympus lens with the EC-20 teleconverter, effective focal lengths 140 to 600 mm.
The Olympus lens with the 3x supplementary lens, effective focal lengths 210 to 900 mm.
The Hanimex lens with a 2x teleconverter, effective focal length 600 mm.
The Hanimex lens with a 3x teleconverter, effective focal length 900 mm.
The Hanimex lens with the 2x and 3x teleconverters, effective focal length 1800 mm.
In the process discovered the advantages of having two tripods. The new one isn't much
better than the old one, but one holding the camera and another propping up the lens makes
it an order of magnitude more stable. Also noted that I need to cover the viewfinder in
these slow lens photos; otherwise the exposure is greatly falsified. I'll write up a
separate page over the next couple of
days.
On the way home this afternoon, Yvonne saw a pillar of smoke. Checked the DSE site and found yes, indeed, there had been a bushfire in Dereel, but it was last
week and now controlled. It wasn't until much later that they reported another fire, today, in Enfield—also “controlled”.
A couple of things worry me about this:
Both of these fires were in Enfield State Forest. Is this arson? If so, will the
arsonist strike again? Coming Friday promises to be a
dangerous day, and any fire that gets here would probably come from that direction.
How big were these fires? If you believe the DSE—not a wise idea—they're
minuscule: last week's Peppermint Track fire is reported as 0.01 ha, and the one in
Enfield was 0.002 ha—and they're both “controlled”, the Peppermint Track
fire since over a week. How difficult is it to put out a fire covering 20 m²? That's the
size of an average room in a house. My take is that the size of the fire has been greatly
falsified, possibly by people who can't count.
Mail from the vendor of the supplementary telephoto lens, telling me something that I had
already discovered:
5. Do not use the camera wide angle mode while telephoto lens is on camera
6. This will cause black borders and tunnel effect around lens
Shouldn't this have been part of the advertisement?
Spent most of the day working on my telephoto comparisons, which clearly showed that the supplementary telephoto is of no
use whatsoever. It also confirmed my opinion that the Hanimex 300 mm lens is so much worse
than the Olympus that it's no use either. About the only thing that's still in the running
is the EC-20. I should probably do more testing there.
How much water does a Salvia
microphylla need? 5 months ago I made the decision
to prune back the one round the old petrol pump as new flowers appeared, which they did
quite nicely, and now the new growth is as high as the old growth was. But it's looking
straggly; here the appearance in October and now:
On the other hand, the soil is still moist to the feel, at least under the mulch. Put a
dripper line in there anyway; we'll see if that improves things.
More cooking. Spent a surprising amount of time looking for a good recipe for seekh kebab, not to mention the reason why it's
spelt “seekh” and not “sikh”. It seems it's a Farsi
(“Persian”) word, and has nothing to do with
the Sikh people. Results weren't too bad;
for the first time I got a consistency that didn't fall apart on me.
Another day of extreme fire danger! Spent a lot of time looking at the DSE fires site, which told a grim story: 8 new
fires, and unlike Black Saturday, as they seem to be
calling the event, they were mainly to the west of Melbourne—three or four of them in
the Portland area. And once
again the site showed how completely useless it is: we had another couple of fires of 20 m²,
and the fires in Portland disappeared without trace later in the afternoon. Surely people
can handle this information better.
Daniel O'Connor sent me some patches for FreeBSD months ago, but we've changed our repository to subversion, and I still haven't got my head around
it. Spent a lot of the day updating swamp.lemis.com, my test machine, to the latest
version of FreeBSD 8-CURRENT. Soon I should finally get round to committing the
patches.
The first stage of the telephoto lens
comparison is over now, but I still need to play around with this EC-20 teleconverter I
have on loan. Took a number of photos, again far too late in the day; the results were
mainly useless, and only these two showed any merit:
The main problem is focus: the lens can auto-focus, some of the time, but it's so
slow that I'd prefer to set the focus and then change to manual focus to keep the distance.
And there's the issue: depending on where the bird lands, it'll be in focus or not. And
under these lighting conditions I needed full aperture (f/11.2, as it reports). More
thought (and light) required.
The situation can be improved by postprocessing, though the problems remain. Here is
what I did out of the box 15 years later (run the cursor over an
image to compare it with the previous image):
The remains are barely visible against the background of mulch. About the only consolation
is that it's not the first time, and it has recovered before. They're unlikely to have
another go at that one for a while, but I'd rather they ate more of
the Nasturtiums, which recover better:
Photo day again today, giving me a chance to try out my new tripod in earnest for the first
time. I had already puzzled about the utility of a ball
head, and today I got my answer, I think: they're not nearly as good as the old pan
and tilt heads. The one advantage is that you can easily tilt the camera on its side, which
isn't easy with a pan and tilt head. The significant disadvantage is that you can easily
tilt the camera on its side even if you don't want to. Taking my panoramas was a lot more
difficult as a result. Looks like I need a new head.
One of the things it did demonstrate is a significant disadvantage of the new xD card: it's slow! Taking
the panoramas of the verandah stalled because the write buffer for the card was full. On
further examination, it's a type M card, which only manages a write speed of 2.5 MB/s, only
about 13% of my CF card. I suppose it's fast enough for most things, but it's obvious
enough to be annoying.
My Olympus E-510 camera has a rather interesting feature: it has support for panorama photos. For reasons
best known to Olympus, it only works
with an xD memory card. Until last week, I didn't have one, and instead I my weekly
panoramas of the outside of my house using hugin. It's not clear what additional functionality the camera provides, but
it is a pain the whole way:
To use it, I need to set “SCENE” mode, and it insists on “Live
View”, presumably so that it can display a frame for lining up the photos. But
Live View is quite a problem outside in the sun, where it can often be too bright to use
it.
The overlap required appears to be fixed, and is very small, only about 10% of the total
picture width. As a result I only took 4 photos for the test shot, where hugin
had shown me I needed 7, and even then it covered a wider angle. Based on my prior
experience, I had my doubts about whether it would be enough.
Once having taken the photos, I had to process them with this horrible OLYMPUS Master
2, which took several attempts just to download the images from the
camera—and then I couldn't find the way to get it to load only some of them, so it
took about 20 minutes to load the entire contents of the card. About the only
reasonably clever thing it did was to select the images automatically when I clicked on
“Panorama”—but I still needed to double-click on each one to get it to
accept it.
Finally it created a panorama of sorts. As I had feared, the overlap wasn't enough, and the
exposure was uneven. According to the instructions, the focus and exposure remain unchanged
for the set of photos, but it doesn't look like it. Here the “normal” panorama
created by hugin, followed by the Olympus panorama, the latter uncropped:
The joins between the individual photos are very clear, while with hugin I have to look
carefully to see them. Here are three details of the preceding photos:
Here the hugin image (the first) shows an offset column, but that's all. The Olympus
photo shows duplication of one of the Canna
flowers and severe discontinuities in the roof structure.
I can't see anything wrong with the hugin image here, but the Olympus image shows two
problems: firstly the obvious discontinuities, and secondly the missing detail at bottom
centre, caused by lack of enough images.
Here there's a minor discontinuity in the hugin image,
and—surprisingly—none in the Olympus image. But the exposure is completely
different; possibly the light changed between the two images, and it couldn't handle it.
In sum, then, another “feature” that I can forget about. I can't see any
advantage in it.
To make things complete, looking through the MythTV programme information, I discovered that there was no information for
ABC2. For some reason, shepherd hadn't loaded it.
How do you debug this kind of problem? With great difficulty. Shepherd starts by
refusing to run more than once a day. You can work around this with the marginally
documented --notimetest option, but only if you run shepherd directly. It
doesn't work in its normal incarnation as tv_grab_au. I've already worked around that
by changing the contents on /usr/bin/tv_grab_au:
Then shepherd insists on checking for updates Every Single Time it is run,
taking about 15 seconds to complete. There's an option --noupdate, but it doesn't
work for me.
Those are just problems in running shepherd at all, of course. Why didn't it load
any data for ABC2? That sort of information is in the configuration
file shepherd.conf, but that's full of status information and gets changed all the
time. diff shows hundreds of lines like:
It's not clear what Adelaide channel 31 has to do here; it can't be from when I was in the
Adelaide area, since this is a completely clean install. And I have no idea what TVS is,
but the changes seem completely gratuitous; they're just changes in sequence. There's so
much stuff in there that it's impossible to guess what change might relate to the problem.
So I ran shepherd --configure to check if the channel mappings had changed, and
found, amongst other things,
Guide data sources:
( 0) (no guide) ( 7) SBS
( 1) ABC HD ( 8) SBS HD
( 2) ABC1 ( 9) SBS News
( 3) ABC2 (10) SC10
( 4) Imparja (11) WIN
( 5) Prime (12) ImparjaHD
( 6) Prime HD (13) SC10HD
MythTV channel -: ABC2 ?
That MythTV channel - indicates that it thinks there is no “channel
number” (a remnant of obsolescent analogue technology) associated with ABC2. But why
not? Probably left behind by one of my attempts to add tuners to the system. A quick query
showed:
So there were two entries, a valid one and an invalid one. There are a number of invalid
ones, and I've been scared to remove them in case I break something else in MythTVs house of
cards. But there was a valid one too; why didn't channel 22 show up? Went off looking for
what queries shepherd makes, but couldn't get the mysqld to trace; presumably
it was built without the corresponding trace code. Instead (this is Linux)
used strace to look at the traffic. That's a bit of a pain too; you can get it to
dump the complete buffer read in, but you have to do it by file descriptor. There appears
to be no way to tell it to dump for all file descriptors, or just for all reads. If there
is, it's well hidden. And in the process managed to get a couple of hangs and had to kill
the strace process and mysqld, which restarts automatically.
Finally extracted the following three queries:
SELECT j.id, j.chanid, j.starttime, j.inserttime, j.type, j.cmds, j.flags, j.status, j.statustime, j.hostname, j.args, j.comment, r.endtime, j.schedruntime FROM jobqueue j, recorded r WHERE j.chanid = r.chanid AND j.starttime = r.starttime ORDER BY j.schedruntime, j.id;
SELECT chanid, starttime, lastupdatetime, recusage, hostname FROM inuseprograms;
SELECT name,callsign,channum,xmltvid FROM channel;
Clearly the one I'm looking for is the last one, and of course it looks as if it should find
all channels. And it does. So why doesn't shepherd find the channel number?
I still have no idea. In the end, I just assigned channel number 4711 to the second record,
and it worked, as I could see from the verbose output of running shepherd. But
I still had no data for ABC2 in the database. For that, it seems, you need to
run mythfilldatabase and let it invoke shepherd via tv_grab_au. After
several hours, I had worked around the problem. What a pain!
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.
