Yvonne into my office today to tell me that she had been
unable to upload her photo metadata to the external web server.
Why? Was it down? She had been able to upload the photos themselves to Digital Ocean, so it wasn't the net connection to
the outside world. But I couldn't even pingwww.lemis.com.
Was it down? No, no issues from eureka, only from lagoon. Routes? Probably
not if I could access Digital Ocean, but checked for explicit routes. No, nothing.
Ran tcpdump on eureka's external interface. Ping out, ping reply.
And lagoon saw nothing. A bit of searching: the firewall! This line stopped it:
ipfw add 00021 allow ip from mx1 to aussie-gw
I suppose that makes sense, but how did it work before? I had already puzzled about this
rule when I rebooted a couple of days ago, where the rule looked even stranger:
ipfw add 00021 allow ip from mx1 to eureka
That could never have worked because no traffic gets routed to eureka over the global
Internet. All traffic goes to aussie-gw. So the new rule means that nothing finds
its way to the natd further down in the rule set.
But it used to work. At the very least it seems that I made modifications to the
configuration that didn't find their way back to the config file. I won't make that mistake
again (I hope).
How to fix it? First I need to think my way through ipfw. For the time being,
Yvonne has synced her metadata, so I have time to think.
The other networking issue I have is with distress, my Microsoft 10 box. Compared to
firewall debugging, this seems to be swamp debugging. I still suspect that it has suddenly
decided not to be compatible with older versions of SMB, but there was one
workaround to try: mount the NFS-mounted copy of eureka:/Photos from lagoon.
Does that work? Yes! So I'll leave it like that until I have finished
upgrading eureka, hopefully some time in this decade.
So a 100 W equivalent (“ENERGY”) globe produces 1600 lm (or Lm, as they prefer to falsify
it), and a 60 W equivalent produces 800 lm. That doesn't seem plausible. But it doesn't
really answer the question, so off to check what I had:
What does this tell me? It suggests that the brighter globes are also more efficient (maybe
some overhead?), and that there are two different groupings of efficiency, one round 102
lm/W, the other round 85 lm/W. Coincidence?
I was quite happy with the Viltrox
NF-M43X “speed booster” (apparently a trademark of Metabones) that I received last month. The build quality is considerably better than the other adapter, and you
can see past the rather silly tripod adapter that's part of the device.
Or can you? Today I used it to take the photo of the Strelitzia reginae. And there I
ran into a problem:
The tripod adapter fouls the ARCA Swiss plate that I normally have mounted on the camera.
Yes, I can take it off, but it's a bit silly to have to. They could at least have made it
demountable, or done the logical thing and omitted it altogether. You can't even mount an
ARCA Swiss plate on it, because that would foul the camera body.
The obvious problem is inadequate depth of field. I could have stopped down a little, but
not much. What I need is focus stacking, but they won't keep still long enough for
that—yet. I wonder if they look any worse when they're dead.
Yesterday I discovered that I could access an NFS-mounted FreeBSD file system with SMB. But that was only up to a
point, and when I tried more complicated processing today I got various hangs. Why? I
don't know. I can still do all the processing I want via dischord, the Microsoft 7
box, so it's not urgent. More urgent is continuing the system upgrade for eureka.
On Sunday I got my eureso upgrade to the point where I had a file system with
an up-to-date kernel and userland—on dereel. Next step is to upgrade the ports, and
for that I needed to boot it. New machine? Virtual machines should work for that.
OK, fire up my eureso virtual machine. Oh. Can't do. Some network problem,
error VERR_SUPDRV_COMPONENT_NOT_FOUND. What's that? I've seen problems with VBox
networking before. From prior recollection (and documentation), it was something to do with the kernel modules that were
needed: /boot/modules/vboxdrv.ko and /boot/modules/vboxnetflt.ko. But they
were loaded. Found a /boot/modules/vboxnetadp.ko, so loaded that too, with no
change.
OK, Google
search. Once again no useful answer. OK, unload the kernel modules and reload them.
Suddenly I had 14 modules instead of 11! But no, the Id of the modules doesn't have to be
sequential. In fact there was only one new module, ng_ether.ko. The process of
loading the other two must have loaded it too as a dependency. And after that things
worked.
So why didn't ng_ether.ko get loaded at boot time? No idea. But that's a FreeBSD-specific module, so it's probably not surprising
that I didn't find any useful answers.
But that wasn't the end of the matter. After NFS exporting the file system
on eureso, I was able to tar copy the file system—for about 400 MB of data.
Then the VM crashed. Restarted it, restarted the copy, and I got a spontaneous reboot,
which further progressed into a reboot loop that did about 3 iterations before
continuing—and dereel carried on copying as if nothing had happened. That way data
inconsistencies lie. How do I copy the file now? From the other end?
One of the strangest things I've found about Strelitzia flowers is that they're so
varied. Yesterday'sStrelitzia reginae flower seemed completely normal, but I've seen others. And now
the Strelitzia nicolai has produced a flower with at least 4 spikes:
How did that happen? These cans feel strange, rough-surfaced, almost as if they were made
from plastic. Did I put the slab down too forcefully? No, it seems. The leak was tiny and
on the side of one of the cans:
That can't have been my fault: it was in the middle of a 6 pack, pointing inwards, and the
cardboard of the pack was undamaged. Put it in a beer glass into the fridge, where it
stopped leaking, at least while I was watching.
More trouble with my mobile phone today. Various buzzes. Is that its way of saying “you're
receiving a call, but I'm too polite to make the noises you asked for”? Tried calling
myself. No “ring” tone. Was the original buzz Yvonne trying to call me? Tried
pressing the phone symbol on the log to call her back. Nothing happened.
While messing around, about a minute later, my VoIP phone rang. I had pressed the phone
symbol on the wrong log entry and called myself—but it had taken the phone a whole minute to
get the request through its thick head!
OK, reboot. And how about that, though this is Android and not
Microsoft, it worked! And the ring tones worked again.
On IRC people suggested that you shouldn't need to reboot a phone. But this is a Nokia 3, a cheap and nasty phone only about
50 times the speed of a CDC 7600, the 1970s supercomputer.
The year is coming to an end, and one of our tasks is to write a new Christmas letter (really an end-of-year letter, since we
want to include Christmas in it). And for that we need a photo.
In the past we've chosen different places each year:
Where do we take this year's photo? It's difficult to get something from inside because of
the disturbing background. But there's one room with a sofa and a plain background: the
“library”.
But: it's been almost a year since my cousins Brendan and Robert Herbert brought me many boxes of
old photos and documents from my mother, and the room is still full of them:
I must finally get round to scanning in the photos. Or should I? I don't recognize a large
number of them, and I still have lots of old photos from the 1980s and 1990s that I haven't
scanned in. At the very least I should get rid of all the old frames in the second box,
which are really of no use to me.
The place for the photo is only one part of the setup. How do we differentiate ourselves
from previous years? Clearly horses are out of question in the library, so it's back to our
faithful Borzois. And in this
year of the pandemic, we should
all wear face masks—the dogs too.
How do we do that? Yvonne has borrowed a couple of greyhound
muzzles from Whitey (Tim Winter), our pet food supplier, who is into greyhound racing:
43 years ago today we had a family reunion. I was back
in Australia for the first time in 6 years with Doris, my wife of the time, and my parents
had held not one, but two events, a barbecue and an evening reception. It was the first
time in a long time I that had seen many of my relatives, and some I saw for the first time.
And for many, it was also the last time; in the following 20 years I only returned to
Australia once, and I didn't visit many of them then.
But one of the things I found in my mother's old documents was this pack of gold-rimmed
invitation cards, somewhat over the top even for her. Looking at the amount left over, she
doesn't seem to have used many, especially since they came from the Singapore era that finished nearly 3 years
earlier:
How about that, the second party of 43 years ago. But the date was 9 December 1977. How
could that be? Did I get the date wrong? I wasn't keeping a diary at the time, but I did
keep notes of fuel consumption. For early December I have (written on the back of a punch
card):
The format is mileometer reading (trip counter in brackets), both in miles, / litres /
price. Look at those fuel costs! Between 16.5 and 21.9 ¢ per litre!
I'm pretty sure that we arrived in Australia in late November, and the fact that we didn't
drive anywhere until 6 December is not of any particular consequence. And on 9 December we
were in Frankston.
The next fill wasn't until a week later, in Bendigo. The distance traveled (245 miles,
or 392 km) clearly includes the journey from Frankston to Bendigo (210 km), and the other
180 odd km don't seem much for a week's driving, so that doesn't preclude the possibility
that we drove to Bendigo on the afternoon of the 6th.
Still, there was a barbecue, which started round midday, and I was pretty sure that it was
on the same day. My aunt Freda was there, and she lived in or near Frankston. But I'm sure
she didn't travel with us. So I'm still not sure.
Yesterday the latest issue with my eureka upgrade was eureso crashing
while copying files. OK, tar the file system to a file on (the real) eureka and
untar it into the second file system on eureso.
It still crashed! It wasn't file system corruption on the destination file system: I had
run newfs on it first. What's the issue here? Old, worn-out VBox? Old, worn-out kernel on eureso? No,
that kernel is the same as the one running on eureka.
But clearly I need an alternative. One of the pile of 5 old real machines on the office
floor? I don't even know which ones are still functional. One of the other VMs
on eureka? Tried firing up one called 12-stable, which couldn't
boot. 12-unstable could, so I hung the file system on it and was finally able to
copy things, but by then it was evening.
Time to tidy up old, worn-out machines, both real and virtual.
Appointment with Paul Smith today to discuss last month's blood test. Dammit, do I really have to go to Ballarat just for that?
No. One of the positive things about COVID-19 is that we can
now perform many doctor's appointments by phone, and today was a good example. Somehow it's
not the same, though; maybe we should have a video link.
Results were unspectacular. Vitamin D is now in the normal range, though the test was taken
only a couple of days since the end of the 3000 IU treatment. So for a while I'll take a
more normal 1000 IU. And HbA1c seems to be climbing, from 5.9 in January to 6.1 in June to 6.3 now. The soft limit is 7.0,
and the hard limit is 8.0, so though Paul is concerned, it's not clear why. Even if things
continue at this rate, it'll be a couple of years before I hit the soft limit.
What are these numbers? According to the Wikipedia page I shouldn't be over 5.7%. I had to drag out my prior records (later)
to find that the numbers Paul quoted were %, and they correspond to 41, 43 and (probably) 45
mmol/mol, an obfuscatory way to say 4.1 ‰ etc (Per mille, itself an obfuscatory way to
describe tenths of a %).
But they're both % values! What is this? Ah, blood sugar is too important to have only one
unit of measurement. It seems that there are at least three. I don't really want to know.
And there's the issue. I don't have a copy of last month's results. Normally Paul prints
them out for me, but that clearly didn't work. Still, he can send them to me by email
(reminding me that I'm still waiting for X-rays by email from Mario Cordioli from two weeks ago).
I also promised to send him some information on my visits to Dr. (Mr.?) Turner on 24 June 2015 and 19 August 2015. But the mail didn't go through:
<reception@healthfirstballarat.com.au>: host
d205870.a.ess.uk.barracudanetworks.com[35.176.92.113] said: 550 permanent
failure for one or more recipients
(reception@healthfirstballarat.com.au:blocked) (in reply to end of DATA
command)
Blocked? What's that? Did I have the wrong email address? No, it's correct. Why is it
blocked? Called them up and spoke to Amanda, who sent me an email message, to which I
replied successfully. Well, except that she couldn't open the attachment. What attachment? Oh yes,
there's a digital signature. But it seems
that her MUA is not set up to read straight RFC 822 mail messages, though she did find it.
So what's going on here? Clearly Amanda doesn't know, but it would be interested to
investigate.
So when did we have the family reunions in December 1977? I had thought 3 December 1977, but my mother's invitation said 9 December. Nothing for it: I needed
to find additional evidence.
The obvious place was to look at the photos that I took at the time. The scans, taken
weren't very good, but since then I've been able to significantly improve the results. So
instead of scanning my mother's old photos, as planned, I started scanning my old photos, in
the process discovering many more, both unscanned and (badly) scanned. There's a total of
about 320 of them, and it takes half a day to scan a single film.
Petra found a mud wasp nest next to the front door today, and accordingly removed it. But
the wasp wouldn't take no for an answer. During the course of the day, it (they?) rebuilt
it at an amazing rate. I started observing relatively late, but between 17:33 and 19:10 it
increased like this:
Spent most of the day scanning in slides from December 1977. I've scanned slides before,
but I still run into issues I hadn't expected.
The good news: the images are much better than the best I could get out of that
horrible Canon 9900F. But
they weren't the same! After scanning in all images that I think could have belonged to the
barbecue on 3 December 1977, none of the images matched.
The explanation came with the next film: more photos of the same event or events,
interspersed with newer images up to the middle of the month. It seems that I only scanned
in that film in 2004. It's also clear that we had two
cameras—here's Doris with the other one a week or two earlier:
Or is it? Where's the photo she took there? And almost none of the photos appear to have
been taken by Doris. Why would I take two rolls of Kodachrome 64 in two different cameras at
the same time? And which would the other camera have been? My second camera (an Edixa-Mat Reflex D)
was stolen the year before, and I never replaced it. So
on the face of it it would have been Doris' camera, probably a Miranda F, except that there's every reason to
believe that I took most or all of the photos. I'm still puzzled.
The scan results with the Epson were an order of magnitude better with the Canon, but it's
not perfect either. In particular, apart from the clunky software interface, it also seems
to have issues recognizing the size of a slide, at least when it's in portrait mode. This
is the display after previewing:
The first image (top left) clearly has a different aspect ratio from the others. That's
because it's in portrait orientation. Turning it round and repeating the preview gave the
full width, as these two crops show (the second rotated back for comparison):
That's actually easier than putting them all in the right way up. It takes less time, and
rotating them during postprocessing is much easier. But it also shows a general issue with
framing: now the images are clipped in the other direction.
The mud wasp nest was clearly finished when I took my last photo yesterday evening. It looked just the same this morning. OK, remove it and watch
what happens.
But that's as far as it went. After I took these photos, it stopped. I think it decided
that my proximity was not good, and it's either going to wait for me to go away, or more
likely build a nest elsewhere.
Off to Ross Creek today for lunch with Danielle Teo and family. We arrived too early, of course (for the
record: driving time 22 minutes), so off with Yvonne to take
a look at Wotan, who has ended his wanderings here:
Danielle is Hokkien, and so of
course was the food. My experience is more with Cantonese food, and there was some
difference. Now how many of these dishes can I copy?
Back to yesterday's question: why did I take two separate sets of photos of the family
reunions in December 1977?
Why, did I take two separate sets of photos of the family reunions in December 1977? After
more careful comparison, including of who was there and what they were wearing, it seems
that we had more reunions than I thought. What I have now is:
Probably 3 December: we had a relatively small barbecue. The only people I got on film
were two boys round 10 years old:
It's quite probable that these are Brendan and Robert Herbert, my cousins who live in
Bendigo, so their presence
didn't need to mean much in the way of parties.
If I'm right, this also shows how things had changed the next time they visited:
That's my father staring bravely into the cannon, my uncle Bob Herbert on the left in
the first image, his wife Lorraine on the left in the second.
Then, almost certainly on the next day, we had a barbecue worthy of its name: in
particular Bob and Lorraine had come from Armidale, 1200 km away. This
barbecue was also with members of the Lehey family, though I don't see any here:
I suppose it's a sign of the times that I don't recognize who this is. But clearly the
evening photos were taken on the same day. For some reason there seems to be no
representative of the Leheys in the photos:
So it seems that all these photos were taken sequentially in one camera over two days. If
Doris took any, she probably kept them when we broke up. And it strongly suggests that we
didn't have a reception on 9 December, as the invitation cards suggested, or if we did, I
didn't take any photos. The date also clashes with one of the few possibilities where I
could have taken photos at my uncle Max's place, which I had set for 9 December.
What a pain this scanning is! The slides weren't nearly as well-marked as I thought. Film
number 7 proved to be mainly purchased slides of the Alhambra, now amazingly faded, which we
presumably bought in October 1982, along with 13 slides of Singapore. And my propensity to get 37
shots on a 36 exposure film meant that some of the slide carriers (also for 36 shots)
overflowed. The carrier with the Alhambra and Singapore also had a 36 and 37 with a proud
mother and baby, and neither of us can recognize who they might be.
On the positive side, I'm now in a position to compare the scans of 16 years ago:
It's not as simple as just being underexposed: I couldn't find a way to enable 48 bit mode
on the Canon, so I couldn't fix the scans. And the thing was too stupid to find the frames
correctly (something, however, where the Epson also has problems).
I've been having various issues with the wireless keyboards that I use instead of a remote
control for teevee, my TV computer. Spent some time messing around with batteries,
without any final results.
But Yvonne wanted to join in the fun. She came to me and
said “I think my mouse needs a new battery”. I don't think so:
But yes, the mouse was dead. It has a red LED, but it was off. How old is it? It seems to
be a relic of a bygone millennium. But things were worse than that: the keyboard didn't
seem to work either, though the NumLock key came on. It took me a while to realize
that I had dislodged the monitor connector while checking the mouse.
But the X display still didn't work. Switch to a
vty? Yes, worked, sort of: the display driver scrambles the screen, but input still works.
Back to X. The LED came on and it worked, sort of: I could move the cursor, but some button
functions didn't work. Window manager problems?
Restarted X. Same problem. Try with other mice. Similar problems. Mouse buttons 2 and 3
worked, mouse button didn't. Dammit, Microsoft time: reboot.
And reboot failed with an unexpected error:
Dec 7 17:41:28 lagoon kernel: (da1:umass-sim0:0:0:0): SERVICE ACTION IN(16). CDB: 9e 10 00 00 00 00 00 00 00 00 00 00 00 20 00 00
Dec 7 17:41:28 lagoon kernel: (da1:umass-sim0:0:0:0): SCSI sense: HARDWARE FAILURE asc:44,81 (Vendor Specific ASCQ)
Dec 7 17:41:28 lagoon kernel: (da1:umass-sim0:0:0:0): fatal error, failed to attach to device
Dec 7 17:41:28 lagoon kernel: g_access(958): provider da1 has error 6 set
That's the photo backup disk that I installed only 3 months ago. Is it really dead? It's possible, of course, but it seems
strange that it should have occurred right now.
Disconnected and reconnected the disk. Finished booting, doing the lengthy fsck in
the background, and started X. Still mouse problems. The old Sun mouse is clearly
defective; I checked it in teevee and got the same results. But the Logitech
wireless mouse that I put in in its place also didn't always work: in particular, mouse
button 1 only worked sometimes, as I was able to confirm with xev.
So what's the issue? My best bet is that the Sun mouse failed and took lagoon's USB
bus along with it. That would explain the problems with /photobackup. Time for a
new machine. What do I have lying around, apart from junk? My list shows:
My guess is that the second system is defective, and that's why I don't have complete
information. And the first one is considerably slower than the current lagoon,
itself not a ball of fire. Time for yet anotherThinkCentre. The good news is that the
M93p is now available for $159, including postage, so we'll hope that we can hold out until
then.
And a new mouse? OfficeWorks has them
for $4 and upwards. But maybe Yvonne will find that a wireless mouse isn't that bad after
all.
We've had the occasional visit of a heron for at least 3
months, though we've had them on numerous occasions previously. I was idly wondering
if it was always the same one, but today we got at least a partial answer:
Are they a pair? Yvonne is concerned that they would eat the
goldfish, but I don't think that's easy. What they do seem to be eating are the numerous
snails around the garden at the moment.
Another grid power failure this morning, starting at 08:40:53, and lasting nearly 16 minutes. Fortunately it was in
the daytime, so despite the low state of charge of the battery, we didn't have any problems.
That wasn't the only one; at 23:30:50 we had another brief outage, which I normally wouldn't
mention.
Yesterday's Christmas photos weren't the best we have ever taken, so today we had
another attempt. It wasn't much better. Yvonne was able to
get Leonid to sit down, but only briefly. In
the end we once again didn't get anything of much use:
So why can't distress (Microsoft 10) access file systems on eureka with
SMB? There was an
issue with the protocol version, but I fixed that months ago. Has distress forgotten the settings? Off to check, with the help
of this page. Well, sort of help. It's dated 23 May 2018, over
1½ years old, so Microsoft has changed their maze of twisty little menus, all different.
But when I checked, yes, SMB 1.0 client was enabled. The message I had was:
OK, more searching, and came up with this page, which said “Pinpointing the direct cause of the problem you are dealing
with is one of the most important steps while troubleshooting”. That sounds like my kind of
page, but what followed was a series of suggested solutions without first knowing what the
problem was. First I should enter:
C:\Users\grog> net use * /delete
That sounds a bit drastic. How about
C:\Users\grog> net use P: /delete
Yes, that worked. Next, adapted,
C:\Users\grog> net use P: \\eureka\Photos /user:grog Vomit
Left hand, right hand. On the left it discovered EUREKA (its name for eureka), but
on the right it can't find the “network path” \\EUREKA.
tcpdump time. It's sending requests to eureka on port microsoft-ds
(445), and though it's getting answers, nmap can't access microsoft-ds, and
the tcpdump shows RST responses:
Clearly dischord can access the system, so my guess is that distress is
ignoring settings and trying to use SMB client > 1.0. OK, it's nothing new that
Microsoft is broken. High time to get back to my eureso system upgrade.
That should probably be “still in flower”. I had already noticed it in May and thought that it could be an Acacia stricta, except for the
flowering time.
But it seems that it has been flowering for a lot of the intervening time. Maybe not such a
bad choice for our garden. Surely Acacias should be able to survive here.
My previous attempts at huevos a la
tigre left the egg not completely cooked. Maybe the “air fryer” approach isn't the best. How about a
good old microwave oven?
Yesterday the first wide-scale vaccination against SARS-CoV-2 started in the United Kingdom, not without a bit of
theatre. Neave Barker of Al Jazeera reported:
More than half of the people who have died of the virus in the UK are over 80. They're
getting the injection first.
So it wasn't surprising that one the first to receive the vaccine was William Shakespeare,
though I have to wonder whether Donald Trump's logistics team might have had a hand in the matter.
Clearly I went off on a tangent to try to fix SMB communication
between distress and eurekayesterday. I should have been continuing with my upgrade of eureka to a
modern version of FreeBSD. So that's what I
started on today. I had a root file system with up-to-date kernel and relatively new
userland. Boot up in eureso and upgrade the ports.
That wasn't the problem I expected, sort of. It didn't even want to delete Emacs, only, to my surprise:
Installed packages to be REMOVED:
chromium: 79.0.3945.130
firefox: 72.0.2_2,1
OK, I can reinstall them later. Spent much of the day with the install. Then back to
install the browsers:
=== root@eureso (/dev/pts/0) /usr/src 9 -> mailme pkg install chromium pkg: No packages available to install matching 'chromium' have been found in the repositories
=== root@eureso (/dev/pts/0) /usr/src 10 -> mailme pkg install firefox pkg: No packages available to install matching 'firefox' have been found in the repositories
Huh?
=== root@eureso (/dev/pts/0) /usr/src 11 -> pkg search chromium chromium-bsu-0.9.16.1_1 Arcade-style, top-scrolling space shooter
chromium-gn-85.0.4183.121_3 Gn meta build framework
=== root@eureso (/dev/pts/0) /usr/src 12 -> pkg search chrome chrome-gnome-shell-10.1 GNOME Shell integration for Chrome
rubygem-chromedriver-helper-2.1.1 Easy installation and use of chromedriver
xf86-video-openchrome-0.6.0_4 X.Org openChrome display driver
=== root@eureso (/dev/pts/0) /usr/src 13 -> pkg search firefox firefox-esr-78.5.0_4,1 Web browser based on the browser portion of Mozilla
What has happened there? pkg database corruption?
=== root@eureso (/dev/pts/1) ~ 7 -> mailme pkg update Updating FreeBSD repository catalogue...
FreeBSD repository is up to date.
All repositories are up to date.
That's what I expected, but how about using force?
=== root@eureso (/dev/pts/1) ~ 8 -> mailme pkg update -f Updating FreeBSD repository catalogue...
Fetching meta.conf: . done
Fetching packagesite.txz: .......... done
Processing entries:
Newer FreeBSD version for package p5-POE-Component-Server-SOAP:
To ignore this error set IGNORE_OSVERSION=yes
- package: 1201000
- running kernel: 1200502
Why should the current kernel, built last week, have that kind of inconsistency? After a
bit of messing around, discovered that both kernels were considerably out of date:
Huh? I just installed a kernel on this thing a few days ago. Where did it go? I'll have
to fire up instable-12 again to find out what I did.
Why is this all such a pain? Clearly I'm ignoring what I wrote 25 years ago about making
installation easy, but then in those days we were talking about new installations, and we
didn't have the massive dependencies of the Ports Collection to contend with.
I've been scanning slides of our Australia trip in 1977 and 1978 for 5 days, and finally I'm done—I think. Multiple things complicated the matter,
including deciding how to organize the files. I started with a single directory per film,
but then it turned out that the slide trays didn't correspond one to one with films. But
now I have directories ~/Photos/Film/19780101 to ~/Photos/Film/19780108, along
with one that didn't belong there at all: processed in March 1977, and from my preliminary
examination seems to have been taken in Mexico.
That's confusing, because I thought we traveled to Mexico in January 1979. But on
recollection it was a decision we made at relatively short notice: we had planned to travel
to Singapore for my sister's
wedding, but she called it off. And those recollections are related to our time in
Schermbeck, which we left in
October 1976. I had joined Karstadt in May 1976 with the agreement that I
could take my holidays for 1977 and 1978 together over the Christmas 1977 break—as, of
course, I did. So why other holidays in 1977? My guess is that it was (probably
deliberately) left-over holiday time from 1976, which would have expired in March.
So how do I process the photos? Decided to put them all into one directory, process them,
run my “contact print” process and then move the selected images to a more appropriate date.
Repeat until they're all done. So far I only have around Melbourne and the Adelaide trip.
Why did I not find my new kernel and userland on euresoyesterday? Put the “disk” back into the instable-12 VM and checked. Yes, the
kernel is there!
[root@instable-12 /destdir]# ls -l /destdir/boot/kernel/kernel -r-xr-xr-x 1 root wheel 37262464 Nov 29 22:12 /destdir/boot/kernel/kernel
What's going on? Checked the pathnames to the “disks”, not helped by the too-small window
(which, however, I was able to enlarge. Yes, same pathname. But then a transient display
popped up:
That's a feature, not a bug, just not a feature that I was reckoning with. What it
basically means is that I have to be very careful when moving “disks” from one VM to
another. And what do I do now? Procrastinate again and consider what to do. The obvious
thing would be to complete the work on instable-12, but first I need to think it
through.
That's nonsense, of course. This was on lagoon, a FreeBSD box. And the phone number wasn't even valid:
it's clearly a US telephone number. But I couldn't get rid of the page! They had modified
the cursor movement so that I couldn't close the tab, nor even copy the URL:
When I last investigated photovoltaic power, one of the things I did was to check how much
power I could expect from a panel. Various web pages suggested that round here I could get
the equivalent of 3.5 hours of sunshine on average throughout the year: the average daily
output of a 1 kW panel would be 3.5 kWh.
I've been keeping an eye on that. We have 10.8 kW of panels, so I should expect an average
of 37.8 kWh per day. In fact, we almost never got that much. Today (sunrise 5:56, sunset 20:39), for the
first time since the middle of the year, we finally generated 50.01 kWh, the equivalent of
4.6 hours of power. But that's a real exception. I should find a way to sum all the power
we have generated, but I'd expect the average to be no more than 2.5 hours, maybe only 2.
For the past three days I have had no ABC radio
signal at all. Admittedly only two radios, one connected by a real Yagi antenna, the other just a
strip of wire behind my bed. At least no reason to believe that there's anything common
between the two, and TV reception was normal.
Bad ABC! Their coverage page showed no outages. That's not the first time, and
previous experience suggested that I should be prepared before calling them. So I asked on
the DereelFacebook page and got almost no response.
OK, what could the alternatives be? Could it be that the bedside radio was connected to the
antenna after all? Then a loose connector from the radio antenna could be the problem. Out
to take a look:
To get the photo of the antenna I used my Leica DG Vario-Elmar 100-400 mm f/4.0-6.3, which doesn't normally see much use. And
on my way back into the house I was greeted by this sight:
I had thought that it's an Aquila
audax, the audacious largest species of Eagle, known as the wedge-tailed eagle. But
somehow the tail doesn't seem just right. I took other photos, but they were even worse.
Even these are barely acceptable. But what else could it be?
The new ThinkCentre 93p for
Yvonne has arrived. It came with a virgin install of
Microsoft “Windows”Pro, so before giving it
to Yvonne I tried setting up file sharing to see whether it would far any better
than distress.
First problem: what should I call the system? Checked /usr/share/dict/web2 and came
up with the possibilities disagree, dislike, disarray, disaster,
discard, discord, disdain and disturb. And in the DNS
configuration I already had disgust, which I had used last time. In the end I
selected disgust, though it's probably not the most appropriate.
It's only been 5 months since I set up distress, but already I had forgotten the details:
everything Microsoft is so counter-intuitive. Still, I have a HOWTO page to tell me what to do.
Oh. I started updating it round that time, but never got very far. As a result it wasn't
very much help. Still, there's plenty of rant in my diary, and that helped.
One thing that it made clear was that the process was different this time. In particular,
though it was the same version of “Windows 10” (2004), it didn't want a Microsoft ID from
me. Did it somehow sense that it was near other Microsoft boxen? It certainly didn't say.
Took a number of screen shots of the process, something that I didn't do last time, and in
principle things worked without too much pain. I was able to disable their firewall and
enable remote desktop (differently from last time), but I ran into the same issues trying to
mount SMB file
systems from eureka. But I noticed a point in the diary at the time: it didn't work then either until I did something that
made it work. And frustratingly, I never found out what it was.
About the first thing I do every morning is to let Piccola out of the laundry and outside—not directly through the laundry door, but
through the front door, because that's her routine.
But looking through the laundry door I found a dead bird:
What is it? I thought it was a swallow, but Yvonne told me that it was a swift. She's usually right, but I went
checking anyway, and came up with the interesting fact that there are almost no close-up
images of swallows and swifts (especially swifts) on the web. It wasn't until much later
that I found my bird book, which showed images of barn swallows that look almost identical.
So that's what it is until proof of the contrary.
How did it die? None of our animals were involved as far as I can see. It seems uninjured.
Hopefully not some contagious disease.
I've now scanned something like 10 films, but I've only processed part of the first one.
The processing takes more of my time, thus the discrepancy. But before I lose track, I'd
better try catching up. No more scanning for the moment.
Somehow I've been doing lots of things lately. But not today: time to catch up with things
I've been doing, and to do a bit of photo processing. It's good to slow down a bit.
I haven't paid much attention to the garden lately—the lack of progress is very
discouraging. But clearly some plants are suffering from lack of water, so off to check the
sprinklers. Yes, we have a number of leaks. Plugged some of them, but the others are
hidden under surprisingly prickly hebes, and I'll wait for the cooler weather forecast later in the week.
Paul Gallagher along this morning for the fortnightly lawn mowing and continued weeding. He
also made himself invaluable by reconnecting the radio antenna:
Today was Chris Bahlo's birthday. It's been a while since we stopped having dinner every
Saturday, but today was a special occasion, even if the birthday cake involved some effort:
Well, what's “technically”? Clearly the 14 mm lens is rectilinear, and the field of view
follows different rules from fisheye lenses. Wrote an answer, which was well received, but the original poster had a follow-up question:
“Can you explain those formulas again and put them in a form i can save to notes like super
simplified.”.
Good question, I suppose. What does Wikipedia say? Nothing useful, it
seems:
For lenses projecting rectilinear (non-spatially-distorted) images of distant objects, the
effective focal length and the image format dimensions completely define the angle of
view. Calculations for lenses producing non-rectilinear images are much more complex and
in the end not very useful in most practical applications. (In the case of a lens with
distortion, e.g., a fisheye lens, a longer lens with distortion can have a wider angle of
view than a shorter lens with low distortion.
I have such problems with these statements. First, the author is confusing distortion and
projection. I put these two images in the reply:
The first is the fisheye projection (note the curvature of the frames at the top of the
image), and the second is rectilinear (note the extreme elongation at the edges). Which is
“distorted”? It's interesting to note that some news videos now appear to be taken with
fisheye lenses, maybe because the projection is less obtrusive.
More complicated? For a rectilinear lens, the angle of view is
2 * atan (d / (2 * f)). For a fisheye lens,
it's d / f. Clearly the fisheye is simpler.
Off looking for something better, and came up with this page, by Bob
Atkins. It looks quite reasonable.
Oh. He disagrees with me about the formula for
fisheyes. 4 * asin (d / (f * 4)),
not d / f. Why?
In fact, the disagreement is elsewhere. He has gone to the trouble to show the formulae for
fisheye lenses with different projections. He claims, without evidence, that most fisheye
lenses have an equisolid projection, for which the formula is—he
says—4 * asin (d / (f * 4)). He gives the same
formula as I have for equidistant projection,
the one I'm using. He also comes up with two further projections: for ““orthogonal”
projection (I think he means orthographic)
it's 4 * asin (d / (f * 4)), and for stereographic it's 4 * atan (d / (f * 4)).
So who's right? What projections do most fisheye lenses have? How can I find out? Until
proof of the contrary I'll stick with equidistant.
And what's this equisolid projection? I haven't found a good description yet.
Yesterday I did some thinking about fisheye projections. It's a confusing issue:
firstly, people seem to think that there's only one projection, and that it's a form of
distortion. In addition, most references to projections seem to have to do with map projections. Yes, they're the
same thing, but the viewpoint is somewhat different. And finally, there seems to be no
consensus about the names of the projections. So far I have:
wikipedia
Bob Atkins
alternatives
rectilinear
gnomonic
stereographic
conform, planisphere, azimuthal conformal
equidistant
equidistance
linear, linear-scaled
equisolid angle
equal-area
orthographic
orthogonal
orthogonal
I'm sure that there are others too, but I can't find them.
Did a bit of searching and came up with this page, by Paul Bourke, which
explains the background of the mappings. And the Wikipedia fisheye pages is also not as bad as the
angle of view page.
Sadly, the illustrations on both pages are not really compelling, notably not the one on Wikipedia.
Probably the take-home thing is: the equidistant projection seems to be the best choice.
Probably the most common is the equisolid angle projection,
But Paul Bourke writes:
By far most fisheye lenses manufactured today for cameras, and usually with the intent of
being mapped into equirectangular images, are of the equidistant type.
Who's right? Maybe both of them at the time they wrote. The Wikipedia page gives
projections for a number of fisheye lenses, all old; the newest Olympus lens is the 8 mm
f/2.8 Zuiko, presumably a circular fisheye for full-frame film cameras. It is apparently
equal-area (equisolid). Are the newer Olympus lenses also equisolid?
The obvious thing is to take a photo of a geometric subject and compare areas of the photo.
I did this three years ago:
That doesn't look like the bricks are all the same size. But that's clear on reflection:
the equidistant projection is equidistant: the sizes are the same when mapping a
spherical surface. And clearly this wall is plane, not spherical.
So how do I find out? A relatively detailed Google search revealed nothing. Sent a
question to the M43 Tech
Talk page on Facebook, but I don't
expect any useful answer there. How about considering what happens when the direction of
view changes? With rectilinear views (like Google Maps Street View) the change in aspect at
the corners is very pronounced. If I understand it correctly, things should be distorted in
the other direction with equisolid and (especially) orthographic projections, while
equidistant should show no change. From Paul Bourke's page:
Yvonne off shopping today, time for me to change the hardware
of lagoon, her computer, hopefully without her noticing. Since last week's crash it seemed certain that the USB subsystem of her current machine was
defective, so I bought a new one.
They're both ThinkCentres,
the old one an M91p, the new an M93p. All I really needed was to change the disk and the
display card. How long could it take? 10 minutes?
It took half an hour. ThinkCentres are interesting because they need almost no tools, but I
still have difficulty opening them (each model differently) and getting the disks out of the
carriers. And of course there's the cabling that gets in the way. But the new one came up
with no issues—except that I still had the same problem with the Logitech mouse that had
convinced me that something was wrong with the USB system. So in all probability there's
nothing wrong with the old machine after all. But the new machine has USB 3, so the
transfers to the external disks will be much faster.
Now how do I ensure that Yvonne doesn't notice the swap? Put all the stuff around the
machine back the way it was before. Yes, the box looks marginally different, but will she
notice?
No, not directly. But I had moved the USB connectors round, with the mouse dongles at the
back, and while I was at it I plugged her camera in to charge (off the USB subsystem). And
she noticed that. And somehow the new el-cheapo wireless mouse ($7 from OfficeWorks) still has issues with not waking up.
But as far as I know, she still hasn't noticed that the machine has changed.
I've been observing the mud wasp nests outside the front door for nearly two weeks now. Only the following day it looked as if I had chased it away. But now it seems to have
brought reinforcement. Here on 4 December, 11 December and today:
I've been wondering where the dead birds came from recently, but today Piccola brought a clue in her mouth, this time some kind
of wren. It was still alive, so I took it for her and put it outside on the table on the
verandah.
It didn't look good. It looked as if the tail may have been broken at the base, and it was
panting furiously. Yvonne got some kitchen paper, put it in
a plastic container, and put the bird on top. How about that, it stopped panting, and an
hour or so later it was gone. Yes, I can't exclude the possibility that Piccola found it
and processed it somewhere where we couldn't find her, but I don't think so.
So: was she responsible for the other two birds? It's quite possible that she wanted to
kill them but not eat them.
I had barely put Piccola's bird away when
somebody knocked on the door. Lucy from the BeeacCFA, with a big brochure and lots of advice on how to
survive the fire season. One suggestion: install the VicEmergency app on my phone. My
response: “Why? It's useless”. I had installed it a couple of years ago, and though it's still installed, I don't get notified. She
agreed, but she had to say it.
Why don't people do something about these useless apps?
It rained overnight. How much? I have the choice: 4.0 mm (west rain gauge), 6.6 mm
(weather station) or 7.0 mm (east rain gauge).
I've had issues with the two real rain gauges showing different results; in fact, they
almost never show the same results. But this is very extreme, especially since the weather
station almost invariably shows less than either of the other two. Can it have something to
do with the way the rain fell?
Another dead bird outside the laundry this morning, something like a sparrow. That's
four dead birds in six days. But this one was different: it had clearly been chewed on.
Clearly Piccola. And the fact that I found
it in the same place as two of the others suggests that she had had a paw in their deaths
too.
Yvonne thinks that it had something to do with the hay
delivery on Monday. She had seen Piccola climbing up over the hay to the swallow's
nests. Problem: the dead swallow was on Saturday, and the dead fairy wren was on Sunday. And the other three birds didn't seem to be related to the swallows' nests.
So the hay might make it easier for Piccola in the future, but it doesn't explain how she
caught these ones.
After taking a photo with my Olympus OM-D E-M1 Mark I today,
I switched back from auto-ISO to standard ISO 200/24°. And then Piccola started stalking something.
What went wrong there? The Exif data shows correct ISO setting, but the exposure (1/60 s at f/4.1, EV 10) was far too long.
Only 3 minutes later, after power cycling, I took another photo in a somewhat darker area
with 1/320 s at f/5.6 (13.3 EV). The problem was visible in the viewfinder (one of the
great advantages of an electronic viewfinder), but it went away when I power cycled the
camera.
Into the office this morning and checked the nightly backups, as usual. Where's the backup
for lagoon? Nothing there. A quick check showed a cron job started at 21:00,
but waiting on something. And then nothing.
OK, first start a dump and put the question on hold. It ran normally, but I got two mail
messages. Two dumps!
day 17 month 12 weekday 4
dump -2uf - / | bzip2 > /dump/lagoon-FreeBSD/2/root.bz2
DUMP: WARNING: should use -L when dumping live read-write filesystems!
DUMP: Date of this level 2 dump: Thu Dec 17 21:00:00 2020
DUMP: Date of last level 1 dump: Fri Dec 11 21:00:00 2020
...
DUMP: level 2 dump on Thu Dec 17 21:00:00 2020
dump -2uf - /home | bzip2 > /dump/lagoon-FreeBSD/2/home.bz2
DUMP: WARNING: should use -L when dumping live read-write filesystems!
DUMP: Date of this level 2 dump: Thu Dec 17 21:43:05 2020
DUMP: Date of last level 1 dump: Fri Dec 11 21:01:58 2020
...
DUMP: level 2 dump on Thu Dec 17 21:43:05 2020
That all looked good. But why didn't I have that in my inbox first thing in the morning?
Blame it on Microsoft: it sets the system clock to local time. It took me quite a while to
notice the date: it was in the future. It seems that ntpd took one look at the
system clock and gave up, writing an error message to standard output where nobody would
ever see it. The result was a system clock 11 hours in the future.
The other thing was that Yvonnedid seem to notice
that something had happened to her computer: I left a slot bracket behind after putting the
display card in the new machine, and I didn't replace the stuff round her computer carefully
enough. Next time I'll take a phone to be sure.
On Tuesday I did some thinking about fisheye projections. As expected, I haven't had
any feedback about what projection my fisheye lenses use. And I had already established
that a single photo wouldn't help me decide.
Clearly the images at the edges are smaller for the fisheyes. For an equidistant projection
the sizes should be the same. So maybe it is equisolid after all. It's not clear
that my photos are accurate enough to do exact measurements, but here's what I got for the
ratio of size at the edge and in the middle for each lens:
Lens
Width
Width
Ratio
(centre)
(edge)
M.Zuiko fisheye
1324
1156
0.87
Zuiko Digital fisheye
1396
1256
0.9
M.Zuiko 7-14
520
940
1.8
M.Zuiko 7-14
1272
1796
1.41
The last line is for a different pair of images taken closer to the wall, rather proving my
concerns about the accuracy.
Three grid power failures this morning: one-second outages at 04:53:42 and 08:42:19, along with a nearly 40 minute
outage at 07:47:07. Fortunately the sun was up and the air conditioner wasn't running, so
during the outage the battery state of charge didn't change.
Yvonne in from the paddocks this morning barely able to
stand. Somewhere outside she had been overcome by weakness and nausea, though without any
obvious description. She spent much of the day in bed, but by evening was feeling better
again.
They haven't expired yet—they still have 4 months to go—but I shouldn't just leave them
lying around. Why haven't I eaten them yet? They're clearly wrapped in nice single-portion
nests, but how much does one weigh after cooking? And how long do you cook them for?
I'll believe that when I've checked it. The 6 nests weighed an average of 60.5 g, which
makes the 30 total come some way short of the 2 kg that the bag is supposed to weigh;
possibly they have dried out a little. Still, cook until not quite soft. 7 minutes. And
the 363 g gave 902 g of cooked noodles, close to 5 of my portions. So it was worth the
effort, and showed that I can't just rely on one nest per portion. And of course the
noodles tasted fine.
I think I'm going to have to give up on trees and bushes and concentrate on things that
regenerate every year. But then our maltreated Clematis “Edo Murasaki” is still catching
up with spring:
While chasing my fisheye projection questions, came across this page, which might help. It
comes up with yet another projection, “Thoby fisheye”, which I still need to follow.
This page also looks interesting. Potentially there's other interesting stuff on
that site.
I established last weekend that a number of drippers in the sprinkler system are not working
correctly. I fixed one or two, but found a reason to postpone the rest. First it was too
hot, then there was some other reason. But we risk having plants die, so today I finally
gave myself a kick and replaced them. With one exception it was completely straightforward,
but what a kick I needed!
Over to Graeme and Linda Swift for a barbecue lunch today. About 6 years ago Graeme had
looked over the fence at the house, which was just being built, and invited me for coffee
when it was done. So today was the day:
We were a little concerned that they might serve unpalatable Australian sausages, but no,
Graeme doesn't like them either, so they tried out something new on their guests:
Those are pretty much exactly the chicken and beef kebabs that we have tried a couple of
times in our air fryer, most recently
two months ago. How did Graeme's barbecued kebabs compare? Much better! I think
I'm going to have to accept the fact that the rotisserie on the “air fryer” isn't the way to
go.
They appear to recently have created a bar room in what we termed the “middle sanctum”
(inner sanctum was their house, currently being renovated):
As usual, we ate too much. Well, I ate too much. Yvonne managed to surpass herself by eating only one kebab and a little potato salad—for the whole
day! But I had had enough to do without an evening meal too.
As I've been saying for some time now, the garden is not doing well. I think that I'm going
to have to accept that much of the work isn't worth the time or money. In particular, trees
just don't thrive here. My carefully coddled Paulownia kawakamii (that's the
parent tree on the Wikipedia page) has now died:
Should I be sad? It was barely a metre high after over 5 years; it should have been 5 m
high years ago. Even if it had survived, it would be nothing like it should have been.
Other trees are not doing well.
The Buddleja × weyeriana
that I planted two months ago that I salvaged and then planted a couple of months ago has died. The
Corymbia ficifolia that I bought earlier in the year, and which flowered happily until the first frost, doesn't
seem to have recovered from the winter:
On the other hand, a couple of trees are bucking the trend. The Schinus molle that I was so concerned
about two months ago now
seems to be returning. Here then and now:
Once we get rid of the weeds round its base, it might still grow. A related plant (I think)
is the Acer negundo (“Box
elder”) that we planted nearly 3 years ago, and which never did well. We had thought it dead, but there's
hope yet:
The Hibiscus
rosa-sinensis “Uncle Max” that I planted two years ago isn't dead, and it's developing buds, but it's barely any bigger than
it was when I planted it (first photo):
I think it's just too cold for it where it is. Now that it's looking vaguely OK it's
probably time to take a couple of cuttings for planting inside.
My suspicion is still that trees grow to a certain point, after which they hit the clay
substrate at about 80 cm depth, and then they just die. But there are a couple
In a more positive light, the bulbs and tubers and things are flowering well:
The flower stems are over a metre long, and they remain (there's still last year's stem in
the middle).
Part of the issues is almost certainly watering, and in particular I've wondered whether
that was the issue with the Paulownia and the Corymbia ficifolia. One indication is these hanging baskets, where I was a little
lax with the daily watering. Only one plant survived, but now seems to be doing well:
Where did all that water come from? It can't have been from rain; it hasn't rained in days.
It looks like the fixes allowed new leaks to spring up. How I hate this technology!
Yvonne still hasn't completely recovered from her fall
two months ago, but at least the doctor has allowed her to ride again. Today Chris
Bahlo came over with Yvonne's horse Carlotta, who had been visiting Chris while Chris'
horses ate down our spring grass, and off they went:
Today was a completely different matter. The maximum temperature was 13.3°. I'm reminded
of 25 December 1977, where the temperature in Bendigo was 17°, and in London it was
14°—still warmer than here today.
As if that wasn't bad enough, it poured with rain all day, a total of 33 mm.
Time to cook a batch of baked beans today. I add pork rind, but how much? I didn't note it. Today I put more in than is
probably needed, 120 g for 750 g of dried beans. We'll see how that pans out. I've also
decided that I don't need as many beans as before. At one point I had 160 g per serving,
which I had lowered to 140 g, but it still seems to be too much. This time the portions are
120 g.
My weather station is horribly unreliable,
and my software is full of heuristics to capture valid data anyway. In general, the station
returns I/O errors several times a day, and I have to recover.
But today it didn't work. The debug printfs that I have put in the software return things
like:
./wh1080 starting
Raw rain: 0.000000, max: 3.000000
Previous rain 0.0, rain now 940628436634351921645591195969126400.0, difference 940628436634351921645591195969126400.0
Raw rain: 940628436634351921645591195969126400.000000, max: 3.000000
That's not the station. It's my heuristics, and clearly using floating point doesn't help.
I have the distinct feeling that I have an uninitialized variable somewhere, but I can't
find it for the life of me. The fact that it's happening now suggests that it's related to
the continuing rainfall, but the printfs don't bear that out.
In my mail saw an update to youtube-dl. OK,
upgrade.
pkg: Repository FreeBSD load error: cannot open sqlite3 db: Not a directory
Now isn't that helpful? Which sqlite3 db? Where is this stuff? Was this
related to a recent NFS hang? What are sqlite3 dbs called, anyway? locate didn't help
much, but then I gradually recalled /var/pkg/db. OK, what's in there?
=== grog@teevee (/dev/pts/7) /spool/Already/Series 215 -> l /var/db/pkg/ total 145
-rw-r--r-- 1 root wheel 158 22 Dec 09:28 FreeBSD.meta
-rw-r--r-- 1 root wheel 89,997,312 22 Dec 09:34 local.sqlite
-rw-r--r-- 1 root wheel 54,493,184 15 Nov 16:10 repo-FreeBSD.sqlite
-r--r--r-- 1 root wheel 6,471,021 21 Dec 03:05 vuln.xml
Off to compare this system (teevee) with eureka. 994 entries! Here's some of
it:
drwxr-xr-x 2 root wheel 512 19 Oct 2012 redland-1.0.15_1
drwxr-xr-x 2 root wheel 512 30 Jun 2013 renderproto-0.11.1
-rw-r--r-- 1 root wheel 56,098,816 1 Nov 2018 repo-FreeBSD.sqlite
drwxr-xr-x 2 root wheel 512 9 Jan 2012 resourceproto-1.2.0
Where have they gone? Looking at the timestamps, somewhere else; this is an old, worn-out
directory.
OK, try again to get the error messages more carefully:
=== root@teevee (/dev/pts/5) /var/cache/pkg 52 -> pkg upgrade youtube_dl Updating FreeBSD repository catalogue...
FreeBSD repository is up to date.
All repositories are up to date.
Checking integrity... done (0 conflicting)
Your packages are up to date.
In other words, the problem had gone into hiding. And the update hadn't made it to
the packages system yet.
HOW I hate bad error messages! But my guess is that it was probably talking
about /var/db/pkg/local.sqlite (and not /var/db/pkg/repo-FreeBSD.sqlite),
which was modified round the time. Now how did it recover?
For years, the Frankfurter Allgemeine Zeitung, one
of the best German newspapers, had a slogan „Dahinter steckt immer ein kluger Kopf“. I
can't think of a good translation, and of course neither
can Google
Translate: “There is always a clever mind behind it”. The image makes things clearer:
The important error in the translation is that the article is talking about a head, not a
mind. It's certainly a better slogan than „Bild denkt für mich“ (“Bild thinks for me”), a
slogan of the Bild-Zeitung,
even if the Bild slogan is easier to translate. It also seems to have lasted better:
a Google
image search brought only 20 hits, all false positives, and 17 of which were from this
diary for October 2016. It does, however, show me
that I've done this comparison before, though with a different emphasis, and it makes me wonder
whether I have misquoted the slogan.
But that's all background. This year has been one where (fake?) news is particularly
important, and I've been reading not the German newspapers, not the Australian ones, but the
US Washington Post and
New York Times. Even
Yvonne has joined in. Both have subscription models, but
give a few free articles per week, and they send free news summaries.
Still, that's not always enough, and when the Washington Post offered 12 weeks for $1 for
full coverage, it seemed that I could afford it. The sting's in the tail, of course: at the
end of the three months I can cancel the subscription or continue at $10 per 4 weeks.
But I continue to be bombarded with offers like $29 for a year or (best value!) $59 for two
years:
OK, Washington Post, why is $59 best value? That's $29.50 per year. Or do they really not
intend to continue these offers? I can understand that the $1 for 12 weeks is probably a
once-off, but I'm sure that I've seen the $30 offer frequently. My mail backups show a
“last chance, sale ends tonight” offering exactly the same offer, except that it ended 3
weeks ago. If my backups went further back, I would probably find similar offers.
Still, it leaves me in a dilemma. My current $1 subscription runs until the end of January.
Should I let it run out and hope that the $30 offer is still there, or throw away $2.60 and
sign up now?
There are other newspapers, of course. I've tried reading the The Guardian, which is free, a good
price for the content. And then there's the Courier (Ballarat), our local
newspaper (amusingly one of two Couriers in Australia; the other is the Mount Barker Courier, our
local newspaper when we lived in Echunga). They provide a daily email with headlines and—they claim—free access to
COVID-19-related coverage. Some
of the headlines look good, but they don't offer any free views, and they want $3 a week for
access, comparable to the standard rates for the Washington Post, and round 4 times the
price of their specials. That's too much. I wonder how much the print version costs, but
they're not saying.
An interesting new gadgetannounced today: an adapter for legacy lenses to a mirrorless camera. There are
plenty of those available—I have a number myself—but this one includes a focus helix
connected to the camera's focus logic, so it can autofocus any lens, even ones that never
had autofocus.
That's quite a clever idea. It's billed as an adapter from Leica M (of all things) to Nikon Z-mount, but obviously the make
of the lens isn't important. The make of camera might be: clearly the maker has found out
how to access the autofocus logic of the Nikon. I wonder how many other cameras it could be
adapted to.
Interesting photos going round the media lately, like these:
The explanation: France has closed
the border with the United
Kingdom because of the discovery of a mutation of SARS-CoV-2 (“Coronavirus”, as if there were only one
kind), which somebody thinks is up to 70% more contagious than the current version that we
all know and love. So France has closed the borders.
And the British are up in arms! Goods traffic has ground to a standstill! It's not fair!
They could be right. It's not clear what the real issues are around this new mutation. But
what is clear is that, bar a miracle (in short supply this year) this is going to happen
anyway next week when Britain regains its independence. In fact, the photos above, from the
Independent and the Evening Standard, are not about this issue at all, but about Brexit. And the the first (with photo) was
written a couple of days before the latest contagion scare, explaining that the reason was
people trying to jump the gun on the Brexit chaos. That's something that I haven't heard
mentioned in conjunction with the virus mutation.
Huh? There are at least two things wrong with this picture. First, there's a leak in
irrigation circuit 1, and secondly, its time slot is 4:00 to 5:00, long before I saw it.
Out to take a look. The relay board (sprinkler.lemis.com) showed:
So it seems to agree with the LEDs. Program? Not beyond the bounds of possibility, of
course, but in this case the program is so straightforward that I can't believe that that's
the problem. It never sets more than one relay at a time. Network issues? That, too,
seems hard to believe. Got as far as formatting the output of the program (HTML from the
relay board, the only way it's prepared to talk to me) so that I can read it, but analysis
is going to take a while.
Moreton Bay bugs are
generally only available round Christmas, so we planned them in for this evening's dinner.
How do you prepare them? I've been there before, not coincidentally on the same day 6 years ago, so I had something
to go on.
But somehow it wasn't quite enough. It seems that I cut the wrong side of the bug, the
carapace rather than the tail. Today it seemed easier cut down the sides and at the join
between the main body and the tail:
Turkey breast roast for dinner this evening. It's a compromise between my wish for turkey
and Yvonne's wish for something else. The remains can be cut
into slices and eaten with German-style Abendbrot.
Instructions from ALDI: thaw, cook
in foil at 170° for 40 minutes, turn over. 30 minutes before the end (a polite way of
saying “after 60 minutes”) remove the foil and brush with oil to brown. End temperature
should be 82°.
And how about that, it really took exactly 90 minutes to reach 82°. But:
That's not exactly browned. And it was cooked inside. What to do?
The obvious thing was to let it rest (where the temperature went up to about 86°, which I
think is normal) before cooling again. Then put it under the grill for a while:
We've been buying our eggs from a local producer for some months now, and they have been
excellent, big, fresh and tasty—until the last lot, which appeared to have been a batch that
had been forgotten somewhere for a while. Half the yolks broke on opening, and this morning
Yvonne found this:
But that's still really not good enough. Apart from the fact that the dogs wouldn't look at
the camera, there are perspective issues that make things look bad, and worse still, I cut
off my feet.
Yes, I don't recognize that either. In fact, the message is almost completely useless.
Google has been unhelpful in the past, but now it proves no information whatsover.
Does the blue tick and “This device” mean that the unrecognized device was the one
displaying the page? Hard to say, since they don't. And if so, is it the “device” on which
the web browser is running (dereel.lemis.com), or the device on which the browser is
displaying ( (eureka.lemis.com)? To add to the confusion, it showed the same
information on other browsers and on other systems.
And the display breakage at the bottom? I checked on other systems, and it rendered
correctly there. Another reason to finally upgrade eureka.
Today was the first day of Christmas. What does that mean to us nowadays? When we were
children it meant presents and maybe ceremonies like Midnight Mass, but we're anything but
children now. Good music (part of the ceremonies)? I tried the Christmas Oratorio yesterday, but
today I couldn't be bothered.
And a ham for Christmas. Yes, tasted good. But we cook good food every weekend, so the
biggest difference was that it was on a Friday. Somehow there's nothing special left to do.
How does the Olympus E-PM1 compare to
the E-30? From the size and weight
point of view they're significantly different (E-30 on the left, E-PM1 on the right, both
with 14-42 mm f/3.5-5.6 kit lenses):
Neither is new. The E-30 was announced in late 2008, the E-PM1 in June 2011. They both
have 12 MP sensors. How much do they really differ? The first thing I found was that the
E-30 sensor maxes out at 3200/36 ISO, while the E-PM1 goes to 12800/42 ISO. In
principle, the E-PM1 is a better camera. Those almost 3 years made a big difference, not
just because of the loss of the mirror.
I found something similar when my Olympus OM-D E-M1 had to go for
repair in September 2014. What to use for my panoramas? At the time I had the choice of the
E-30 and the E-PM2,
which looks almost the same as the E-PM1. And the E-PM2 does HDR! So I used that and
not the E-30.
Does the E-PM1 do HDR? Checked the manual, which didn't mention it. Checked the menus,
which didn't mention it. Check the E-PM2 menus, which do mention it. So the E-PM1
doesn't have that advantage over the E-30.
But while trying the E-PM2, I ran into a different problem: the lens seems to be defective.
It's the M.Zuiko Digital
ED 14-42 mm f/3.5-5.6 EZ with this irritating “electric zoom” feature that ensures
that you can't zoom quickly. But today it ensured something completely different:
It shouldn't do that, and it no longer works. The zoom ring does nothing, and the camera
doesn't display anything. I've tried it on at least 3 cameras, so it's not the camera. It
looks like the lens is defective, but why? The last time I used it was on Sunday, where it worked as well as it ever did. And since then it's just been on
the shelf. What's the cause? Electrics? Contacts? Dust?
I've been observing the power output of our photovoltaic system for some time now,
most recently two weeks ago. What's the maximum power that we can get out of it under ideal
circumstances? That should be something like yesterday, 4 days after the solstice with
sunshine all day long. We got 50.43 kWh of power, the equivalent of 4.67 hours of direct
sunshine. By contrast, on 9 December it was 50.01 kWh, on 11 December 51.77 kWh (4.8 hours)
and on 14 December 50.73 kWh.
That looks like as good as it gets. Even at this time of year we can get far worse results,
like the 8.9 kWh on 22 December. How can people claim 3½ hours? It would presumably
require movable panels.
But looking at the power generation graph for 14 December is interesting:
Firstly, except for a single spike, output was limited to 6 kW. I have 3 panel arrays of
3.6 kW each, two facing in the same direction. So the maximum should have been 7.2 kW.
Presumably this is a limitation of the inverter; it's a good thing that they didn't put the
third array facing in the same direction.
But where is the third array? It should show a bulge to the right. Yes, there is
one, but it's not very pronounced. At this time of year the sun sets at 20:43, and the
power cut out at 19:34, when the sun was at an elevation of 11° and an azimuth of 250°.
That's far beyond the view of the north arrays, which must have been unlit round 16:10. But
the power generation started at 7:10, when the arrays were also unlit (azimuth 110°,
elevation 12°). They wouldn't have had any direct sunlight until 8:20. It would be easy to
extrapolate that at the other end and say that power from the north arrays should have
stopped 70 minutes after the arrays were unlit, at 17:20. But that ignores the elevation
(12° at 7:10, 37° at 17:20).
So somehow I need to understand how much direct sunlight and how much light from the sky
contribute to power generation. I suppose I should wait for a sunny evening and disconnect
the west array for 10 minutes or so.
One thing's clear: despite what I thought was a cloudless day, there was significant power
loss between 11:40 and 15:00. Without that, there would have been more power. But how
much?
We've been making this dish for decades; even the online recipe dates to 4 February 2006. In the meantime, YouTube has come along, and there's now a recipe by Joël Robuchon that looks surprisingly
similar:
There's also a recipe at the Académie du
gout, which bakes the lasagne topped with Parmesan cheese.
Problem: there was music, but the voices were inaudible. I put it in the article anyway,
and Yvonne read it before publication. She had no problem
with the sound, though she wasn't very impressed by the recipe.
So why didn't I get any sound? Still problems with one channel? A bit of messing
round with the el-cheapo amplifier that I got in early September. It looks essentially like this:
Swapping the inputs confirmed that one channel was not making it out to the loudspeakers.
But why? I had already confirmed that the Volume knob on the right didn't make much
difference, nor sense, especially as there are also buttons Vol+ and Vol-
under the silly LED display. It was set to minimum. Turn it up. The left channel came to
life! So is it a balance control? Sort of, but balance appears to be equal when it's set
to full.
There had been some evidence of battery leakage when seen from the outside, and in fact it
wasn't much worse. But it had (presumably) led to the detachment of one of the contacts
between the batteries:
I certainly didn't plant any there. but in principle anything that will survive there is
welcome, especially as not much flowers there at this time of year.
Paul Gallagher was due to continue weeding today, but he didn't show. No message. And it
seems he was due to do some pruning next door at the Marriotts. They're going walkabout
again tomorrow, so over to see what needed to be done, in the process speaking to them in
person for the first time in over 9 months.
Finally got some work done on the sprinkler system. Why can't I apply myself more?
The monthly PV battery
calibration today, this time longer. Last month it only took 6 hours; today it was still running at midnight after nearly
10 hours, this time discharging completely:
Mark Murray wants people who can recognize him to do a virtual PGP key signature via Zoom. OK, there are a couple of
things to challenge me. My last key signing was 14 May 2005, and
I've been meaning to set up Zoom for some time.
OK, off looking. https://zoom.us/. Why .us? Anyway, it
wasn't nearly as straightforward as I thought. But finally it seemed that what I needed was
the free zoom client. How do I get it? First sign up (free!). And
above-average stupidity in the process. First, another bloody CAPTCHA (what's a crosswalk? As I have
commented before, nothing to do with the Stations of the Cross). And
the usual stupid password restrictions:
OK, let's try “Funny d0main”. No obvious complaint, but pressing “Continue”
did nothing. Broken web site? After much experimentation, no. Broken password
requirements, broken description, broken error reporting. It doesn't want spaces in the
password, but not only is it too polite to say so in advance, it's too polite to say so on
rejection. The red colour of the text is the only indication.
OK, choose another (completely unrelated) password offensive characters. “We have sent you
email”. But none came, not even a local delivery attempt. Much searching to discover
(higher-order MX):
Huh? Followed up http://www.lemis.com/grog/workspam.html, now http://www.lemis.com/grog/workspam.php, to discover that I once decided to block
a number of TLDs, including .us. Why? At the time I wrote
In my experience, all mail from some some TLDs is spam.
Clearly that's not the case any more. workspam has done its work. workspam can go. And a
resend of the message works. But I'm still puzzled by some details. It seems that I
created this file on 19 May 2015, after moving to Stones Road, and
long after converting my web pages from HTML to PHP. So why the reference to .html?
The reasoning is a few days older. Also, I have noRCS control file that
includes these rules.
Back to Zoom. How do I use the thing? Clearly somebody else needs to send up the session.
They have some suggestions on introductory documentation, but it's all far more complicated
than I expected. For example, the link to the documentation is via the (unlinkable) profile
page. Mañana.
The last year has not been good for my Hibiscus rosa-sinensis. The
original cutting that I propagated 12 years ago is now
dead, and the only plant I have left is planted outside and suffered considerably in the
winter. But finally it's flowering again:
The sunlight in the distance finally faded at 20:46. According to the NOAA Solar Calculator it should be
20:50, corresponding to 19:23 apparent solar time, but that's reasonable enough.
I sometimes idly wondered how far the term “fish” would stretch, or if the version with fake
fish would be cheaper. But today I found proof of their claims:
Clearly it's exotic. It looks almost like Dill, but not quite. My guess is that it's Fennel, like the plants in the Marriott's
garden (from which it might have migrated, though that's over 500 m away).
More interesting, though, was that there were more of these lilies right next to them:
Huevos a la tigre for
breakfast again (a sure sign that we had a broken egg to use up, and in this case also
cooked potatoes). I've been having issues with the consistency of the result, and today I
decided to try a completely different approach: cook for a considerable time in the “hair
dryer” “air fryer”. It wasn't until I was
done that I discovered that it wasn't so different from the previous attempt after all. Instead of 15 minutes at 150°, I chose 20 minutes at
200°. That made a big difference:
While editing our annual newsletter, I
looked back to previous years. Our photography has really improved, and I felt inspired to
go back and reprocess some of the worse ones.
Prepared for processing on distress, the Microsoft “Windows” 10 box. Boy, was it
slow! It had been up for nearly a month, and was clearly running out of steam. OK,
Microsoft solution: reboot.
When it came back, it couldn't access the file systems on lagoon, including the photo
disk! I had exactly this problem with eurekaa month ago, but not with lagoon.
Dammit, what is wrong with this horrible “operating system”? Off again to check what
was going over the network, in the process having to install wireshark from source to avoid having to break
avidemux. How about
that, dischord tries to establish contact with lagoon on
port microsoft-ds (445) and gets a RST. Why that? A quick (OK,
not-so-quick) nmap showed
that lagoon was listening on netbios-ssn (139) but not
on microsoft-ds,
Configuration issue? Off to google, and discovered that microsoft-ds is not
directly associated with the 2.0 version of the SMB client. In
fact, eureka also claims to support microsoft-ds.
OK, is smbd running on lagoon? No! It seems that it gets started
by inetd, and only for netbios-ssn. Start smbd manually, and all was
well.
Well, partially. “It works for me” (grog), but not for Yvonne. Why? More digging
required.
Today was Piccola's 12th birthday, and of
course New Year's Eve.
Time for celebration? Drink (French « picoler ») a (small) bottle of bubbly (German
„Pikkolo“)?
But no, we're too old and boring for that sort of thing. This photo was taken the following
morning. Once again we only waited until midnight in New Zealand.
There's another bloody year gone by! Here's a summary.
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.
