A message arrived from Aussie Broadband yesterday evening after I had gone to bed: you have an unscheduled NBN outage! Well, it
didn't arrive immediately, because we had an outage. But yes, there was one, about 25
minutes long. Impact on me: my
IRC TCP connections
reset, and I had to restart.
More interestingly, though, this is the second unscheduled outage this (last) month. And
the last scheduled outage was nearly 2 months ago, a real record. Have they just changed
their reporting method? Both were in the middle of the night, but yesterday's was before
midnight, something that I haven't seen for nearly a year.
Yesterday's infrared comparison left one question open: why was
the E-30 so much less sensitive than
the other cameras, when 5 years ago my E-30 was more sensitive? One reason might be that this E-30 is a
different camera from the one I had 5 years ago, but should that make a difference?
Off to reprocess the photos I took then, setting the white balance this time. Here a
comparison with the E-PM2 and the old E-30.
The second image is clearly darker than the first, but it's correctly exposed: the sky (top
left) is barely correctly exposed. The problem is that the first one doesn't include the
sky. OK, increase the exposure of both of them until one point (to the right of the crush,
bottom right) barely starts to be overexposed. For the E-PM2, that's 1.23 EV more, for the
E-30 it's 1.86 EV more, pretty much exactly ⅔ EV difference. Yes, they're both lighter than
I would choose, but the important thing was to get the same level of exposure:
So this lightness corresponds to an exposure of 20 s for the E-PM2 and 4 s for the E-30,
both at f/8 (1.7 and 3.9 EV respectively). It's fair to say that that E-30 was a good 2 EV
more sensitive than the E-PM2.
But that's not what I saw yesterday: there the E-PM2 was roughly 3 EV more sensitive
than the E-30. The E-PM2 shot was taken at 3200/36° ISO, which confuses the issue, but it
did mean I could have hand-held it. Reducing to 200/24° ISO and f/5.6, the exposure
corresponds to 1.2 s (4.4 EV) compared to 4 s (3.0 EV) for the E-30. So that's a total
difference of 3.4 EV between the photos of 5 years ago and now. The E-PM2 hasn't changed,
so that would suggest that the exposures are consistent. Are they? We have:
I've adjusted the exposure for the E-30 5 years ago in line with the adjustments I tried
today. But it still doesn't help. The issue is almost certainly that yesterday's scene was
brighter. So if I adjust the old exposures by the 1.4 EV difference of the E-PM2, we get:
Yvonne wanted lasagne al forno for dinner this evening. Not an easy dish. It has four
components: ragù bolognese, itself
quite complicated, salsa di
pomodori, sauce béchamel (for some reason almost always in French) and the noodles themselves. Based on experience
with shop-bought lasagne, I decided to make fresh noodles. About the only time-saver was
that we had salsa di pomodori in the deep freeze.
Got as far as making the ragù, mixing the dough for the noodles and preparing
the buerre manié for the béchamel. Thaw out the salsa di pomodori. It
was lumpy! Turned out to
be ratatouille: the label had
fallen off, and we had assumed it would be the salsa, but it seems that we had run out.
No lasagne tonight! Quick change of plan (Wiener Schnitzel). We can make the salsa tomorrow and eat the lasagne then.
Dereel has many birds, mainly parrots
and magpies.
Sulphur-crested
cockatoos appear by the hundred, irritatingly flying across the sky where I can seldom
get good photos:
Black cockatoos are much rarer, but since we moved to the Stones Road property, we often see
them too (yellow-tailed
black cockatoos), though there are fewer in a mob, only round 50 on average.
It has rained heavily over the last few days, to the point that we could hardly go outside.
And it has been colder than usual. But despite all that,
our Corymbia ficifolia, a
summer flowerer, is still flowering:
That's not very many flowers, but it's more than I recall; yet another effect of much more
fertilizer, I think.
But the plant of most interest was
the Robinia pseudoacacia
that we transplanted two weeks ago. It certainly doesn't look as happy as on transplanting (first photo):
Yvonne was worried that it might die—something that I
considered quite possible—but in fact the impression is misleading. Yes, the old leaves are
dying off, but new ones are already coming out in no uncertain manner, and new shoots are
coming out next to the old, dying ones:
As planned yesterday, Yvonne made a new batch of salsa di pomodori today, and I continued
with making the lasagne. Not good. The
noodle dough got caught in the noodle machine, and I couldn't extricate it:
That's not the first time with this machine, but I've never had it before that. There must
be something wrong with the machine; I won't use it again. In the end I had to give up with
the fresh pasta and salvage the dish with the dried “instant” pasta that I so hate. Tried
boiling it first, which didn't seem to make much difference to it, but potentially made it
less dry than it might otherwise have been.
And the béchamel didn't mix well
either. I had to use a pot that I don't normally use, and it didn't go well with the mixer,
with the result that it burnt on the bottom:
Somehow I've been doing nothing round the house over the last couple of months. Something
in the office or kitchen is always holding me up. But in the last couple of days I've had a
little time on my hands.
OK, spread some fertilizer in the front garden. First “native”
(Proteaceae) fertilizer for
the Grevilleas
and Leucadendra, and then the
remainder of a bag for other plants. Finished that off and couldn't be bothered to open a
new one, so left it at that. After all, it's (finally) dry, not the best time to spread
fertilizer.
And how about finally installing the other shelves in the garage? I installed one
six months ago and then left the other two for “later”. But then I could move out
some of the stuff in my office and tidy away some stuff that I put in there temporarily 5
years ago when we moved in. Set to, in the process recalling why I had procrastinated: these
components are really very difficult to insert correctly, and each of the 24 pieces required
multiple attempts to fit correctly, as I noted on the second day of assembly. Today I had deliberately made things even more
difficult: instead of the intended 5 shelves, I had decided on 7 for this unit and 8 for the
other. I had bought a third unit just for the shelves. Got the 5 shelves installed with
much cursing, and decided that that was all I could subject Yvonne's ears to. Again mañana.
Received on my mobile phone today, as relayed to my Real Computer:
Incoming - AusGov (No contact specified)
Message text: Coronavirus Aus Gov msg: Help us to keep you safe and ease restrictions by downloading the COVIDSafe app now: aus.gov.au/app (4/5/20 19:19)
Now wouldn't you think they would have known that I had already installed it? Maybe it's to
instill a false sense of security in people concerned about their privacy.
CJ Ellis along this morning to help install the gate opener, which we've already had for a
month now. The critical part was the alignment of the mountings, and the instructions had
stipulated using clamps. But that didn't make sense. There's only one place we can mount
the big end of the opener, on the inside of the gatepost. So after careful measurement it
made more sense to mount it there.
Somehow it was slow going. Yes, I mislaid a few things and had to look for them, and the
maker had helped by not supplying enough washers; I'll have to buy some more. But just
aligning things and drilling the holes seemed to take longer than I had expected.
That was the hard part. After breakfast I went out again to do the electrics.
Oh. Where are the cables? The terminal blocks? The screws? Where should I put the
battery charge controller? The instructions said nothing. All I had was a box, two
batteries and the controller:
Clearly the batteries were intended to fit inside the box, though there's no way to tie them
down. And the charge controller? It's clearly not designed to be mounted externally, but
there's no obvious place to put it. The battery box and its cover, however, were much
deeper than necessary, and I found I could put it—again loose—inside the box and still screw
it closed.
Or I would have been able to had they been friendly enough to include screws. As it is,
there's no way to close it. Established that they're the same M4×10 mm screws used in the
controller box. By sheer coincidence found some M4×15 mm screws, too long for this box, but
it seems that they would fit in the main controller box. But I'm going to have to buy some
terminal blocks anyway, so I might as well get the right screws too.
In passing, though the controller is actually attached to its protective box, it doesn't
look as if they were designed for each other:
So now we have the gate opener hardware mounted in place. All that's left is to mount the
solar panel, batteries and the control unit.
Control unit first. No problem. But first I need to connect things up, so removed it again
and turned my attention to the solar panel. It's really a strange looking thing:
It looks as if it was stuck together in some cottage industry. Maybe it was. But more to
the point, how do I mount it? Opened the bracket package and found:
Now the bracket isn't marked as being intended to fit to the solar panel, though I can't see
where else it would fit. But clearly the screw holes don't match.
Dammit, there are just too many problems with this thing. Time to send the seller a
message listing the problems. To my
surprise, received a reply only 13
minutes later. Sadly, it wasn't very helpful. Here a summary:
Please find the following:
- No instructions beyond wiring diagram for batteries and panel.
I will send you instructions to your email shortly.
- No way of fixing the batteries in the box that I presume is intended for them.
The battery box should be able to fit 2 batteries, that is the standard box provided by supplier.
- No screws to close the battery box.
The screws should be in the battery box or the control box, or in your installation accessories kit.
- No indication of where to put the solar controller.
Some customers make their own storage box for controller, or you can try to put it in control box if there is space.
- No way to mount the solar panel.
Solar panel mounting post is not included.
- No cables to join the components.
It only comes with motor cable 1.5m, it is a 5 core cable connects the actuator to circuit board. Rest of the cables are available from your local store or your electrician.
- A bracket that I can't identify. I thought it was for the solar panel, but it doesn't match.
There is a supporting frame comes with panel, like a bracket.
In summary: yes, we don't supply everything, including cables and a place for the charge
controller. And yes, the frame that comes with the panel, the one that doesn't fit, is the
one intended for the panel. And the missing screws should have been there, but no offer to
send new ones.
So about the only thing of use were the PDF versions of the documentation, which showed that
I had received the complete documentation, but also that I had forgotten one of the sheets
of paper:
And how about that, that worked, and it matches the diagram. So why didn't I think of that?
Because it's wrong! The panel is designed for vertical mounting, as the label and
the connector on the back show:
The outlet from the junction box should be at the bottom, which matches the label. It also
matches the orientation of our neighbour's gate opener solar panel, for what that's worth.
So I claim that I'm right, and they're wrong.
Apart from that, the screw threads are on the bracket, so I first need to remove the larger
frame, attach the angled bracket, then screw the bracket back on again without being able to
access the screw heads. And how do I open the contact box?
OK, there's OPEN written at the top. Just push down? Doesn't work. Looking at the
other end, where the cable should come out, there doesn't seem to be any way to slide it.
And at the top end there are a couple of tags that clearly indicate that the
thing should be slid in that way.
Lift the cable end and then pull? That might work, but an attempt didn't show much
willingness on its part. How much force should I use?
Yvonne brought back screws and junction blocks as requested. It seems that finding M4×10
screws is anything but simple, and she had to go to three different places before getting
some at a specialty shop. But are they what I want? She had taken a sample with her
(left), but what she got was on the right:
The ones she got (30¢ each!) are shorter and have a wider head. And the head doesn't fit in
the battery box! It does fit in the controller box, however, and I'm sure I'll find a way
to fix things. But why didn't the people at the specialty shop notice that? What
a pain this all is!
The shelf that I started assembling on Monday was blocking the garage, so while Yvonne was
away I finally got round to finishing it. There are a number of issues to note, in
particular that it's much easier to put in the shelves one at a time before adding the
frames above them. But the big one is just getting the locking tags inserted in the
columns. The side members look like this:
But for that the top ones need to stand a little clear of the body of the member so that
they can catch in the slit. OK, pull them out a little with a screwdriver (run the cursor over an image to compare it with its neighbour):
And that causes untold pain and cursing. I think I'm getting the hang of it now, but I'll
still need a while to drum up courage to build the third unit.
Somehow the Clematis that we bought
2½ years ago have had it hard. The “General Sikorski” near the dog run suffered
multiple insults, but somehow has always come back. The “Edo Murasaki” on the wind break
north of the house grew, but not profusely. I've been watching it produce a couple more
buds, but only a couple, due any day now.
In the middle of the night (about 2:00) I woke up and heard a rushing noise from the lounge
room. It was difficult to localize, but finally I traced it to the outside tap, near the
corner of the lounge room and not far from the “Edo Murasaki”. Somehow it was turned on,
and the hose that had been attached had become dislodged.
OK, outside and turn it off. How much water had we lost? It wasn't running when we went to
bed round 22:30, so it would have been a maximum of 3½ hours. How much water comes out?
Less than 5 l/min. So that would be a worst case of about 1,000 litres. We can live with
that, at least at this time of year.
But while out there, saw something in
the pelargonium bush. It wasn't
there any more in the morning; it had moved further north:
That's the wind break, still with the clematis buds. No chance of them flowering any more.
The winds last night had torn the wind break from its mountings (on the left in the first
photo) and torn not only the clematis apart, but also
the Jasminum polyanthum from
their roots.
It's autumn now, so doubtless both will come again in the spring. They might even end up
bushier as a result. In the meantime, it'll give me something to think about in mounting
the solar panel for the gate opener. How do I avoid this happening again?
In passing, who left the tap turned on? I'm very careful not to do that, so it must have
been Yvonne. But she said she hadn't used it for weeks, it
must have been me. In the end we blamed Nikolai, because he was standing next to us.
It was too windy to mount the gate opener components today, but I finally got round to
spreading some more fertilizer in anticipation of the rain promised for this evening. 10
kg, and I didn't even get the garden finished. Hopefully it will make a difference in the
spring.
The really interesting thing is the relationship of those times. I've been with Yvonne for over three-quarters of the last 50 years. The time between
meeting and Sue Fortescue and marrying Doris seemed long, but it's just the same time as we
have spent here, and I haven't even got round to unpacking things yet.
Today was the last day where I could finish installing the electrics for the gate opener: in
the afternoon it was supposed to rain.
But I got almost nowhere. I established that the frame and bracket for the solar panel were
built in such a way that I would have to attach the frame first and then mount the solar
panel when the frame was in place, not the easiest way to mount things.
But the real issue was how to attach the cables. I had already established that the cover
for the contacts on the panel showed no obvious way of opening. In principle you'd expect
to just push down on the marking:
And how do I connect? No tags. I'll have to wrap the (supplied by me) wire around the
screws. And the other end? The control unit has thoughtfully been provided with spade
connectors that fit exactly over the battery terminals. Only that's not the way things are
intended to be connected:
What's wrong here? I really hate cutting off terminals, and possibly there's some
other way to attach things. It took me a while to establish that these terminals would be
exactly what I needed to connect to the battery if using the supplied mains charger. Is
there a way to use the main controller to charge from the solar panel? No, it doesn't look
like it:
There's no connection for a solar panel there; ADAPTER is for the mains adapter.
Presumably there's a good reason not to connect the solar panel there. And in addition I
received an image of strange aspect ratio showing me that my wiring guess was in fact
correct:
I'm left with the distinct feeling that the opener itself is of good quality, and the
battery and solar installation is a badly thought out addition. But this, too, can be
solved.
In passing, when I opened the accessories bag for the battery compartment, I found, apart
from the (correct!) cable joining the two batteries, the M4×10 screws that I had missed.
What a place to put them!
When Yana was here over Christmas, she took exception to me making frijoles de la olla with kidney beans. They should be
made with frijoles negros. Yes,
of course, I explained, but where do you find them? She set off
into Ballarat and came back with some,
labeled (I think) “Black turtle
beans”. Wonderful! But I had a big batch of frozen kidney beans in the deep freezer,
so it wasn't until now that I finally got round to looking at them.
The ones on the right are the (red) kidney beans. The ones in the middle are from a package
marked “Adzuki beans” and also
Đậu đỏ.
Are they the same thing? But then I found the ones on the left, loose in a container.
They're certainly black. Off looking for confirmation, and found little. The Spanish
Wikipedia page contained the information:
Las semillas son negras opacas, alargadas, de tamaño pequeño, con 18 a 25 g/100 semillas.
OK, I can understand enough of that to recognize that a seed should weigh between 180 and
250 mg. Counted out 25 of them and weighed a total of 4.36 g, or 174 mg per bean, somewhat
lighter than the specification. But the other beans (Adzuki?) weigh about 129 mg each, so
that's the closest I have so far.
So what does “Las semillas son negras opacas, alargadas, de tamaño pequeño, con 18 a 25
g/100 semillas.” mean? I translate it as “The seeds are opaque black, wide, of
small tamaño, with 18 to 25 g/100 seeds”. What does tamaño mean? Put it
through Google
Translate, which for some reason chose the mobile phone version of the page, coming up
with:
The seeds are opaque black, elongated, small in size, with 18 to 25 g / 100 seeds.
So does tamaño mean size? My not-very-good Spanish-German dictionary translates it
as „derartig, so groß, so klein“, suggesting “thus”, but it seems that as a noun it does
indeed mean “size”. Only I had to follow up to be sure, and it doesn't seem to be
appropriate in all cases.
But the real thing that hit me was this part:
The black beans, black beans , black beans , black beans black or black Zaragozas are
black beans...
The weather today was pretty terrible, a good thing under the circumstances: I spent most of
the day in the kitchen.
For dinner we had enchiladas verdes. I've had a recipe for ever, but the details are unclear. It seems that the last time
we tried them was ten years ago, and even then I had my doubts.
Today I started by comparing recipes for salsa verde, a little late. Yvonne had bought
a capsicum, part of the recipe, but it seems that most recipes I've found online don't use
it. My recipe also asked for cream, which I have decided is US American. In the end I
changed it significantly. In the end I had:
The jalapeños should have
been serranos, and we could have
had more (in other words, Yvonne didn't complain), but we
didn't have any serranos. The enchiladas didn't taste bad, though the tortillas almost
disintegrated. Next time I'll consider baking them alone for a couple of minutes first.
And somehow I can't recall having eaten them before.
The weather wasn't good enough today for mounting the gate opener stuff, but at least I
could work my way through the wiring. The first issue proved to be a problem that I had
identified without recognizing the consequences:
Yes, I need a terminal block there. And the thoughtfully included spade connectors are
needed to connect to the batteries. But how do I attach cables to them? The correct way is
with a crimping tool, but if I once had one, I don't know where it is. Buy a new tool for
two connections? That way madness lies.
In the end I removed the insulation and soldered wires to the connectors. How do you remove
the insulation? I tried conventional tools and also kitchen utensils with equal lack of
success and danger to my person. But they're heat sensitive, so I was able to get them off
with the help of the soldering iron.
Only later did I realize that I did have a couple of cables with terminals on them. They're
in the same photo above. But the cable was far too long, and I hate cutting them, so I
probably would have done the same anyway.
Still, that seems to be the last hurdle. Tomorrow the weather will be better, so I can
hopefully finish things then.
Do the batteries for the gate opener have any charge? Yes, one has 13.13 V, the other 12.98
V.
Why the discrepancy? I had measured the second one with the leads swapped, so it really
displayed -12.98 V. For the fun of it, swapped the leads. 13.13 V. Both batteries read
the same, but the multimeter showed a discrepancy of 150 mV depending on which way round I
measured them.
What can cause that? It's a cheap multimeter, but I hadn't expected that kind of issue.
Presumably it has something to do with the circuitry that turns the voltages around.
As promised, the weather was reasonable today, so I finally got round to installing the
remaining parts of the gate opener, only 6 days after CJ and I installed the motor.
First, the solar panel. Yes, that worked. As expected, getting the screws in was a pain,
but somehow I found all the ones I dropped into the grass:
How can I charge the batteries with the charger (which was supplied)? Ah, said
the slip of paper, only joking. Look at the main manual, which has another wiring
diagram:
Is there any sequence in which I should connect the wires? There are no switches at
all. But no, no information in the instructions. I connected the panel first, which
illuminated both LEDs (“Charge” and “Over-discharge”) on the control unit. The latter
made sense if it couldn't find the batteries, and it went out when I connected them. It
wasn't until I checked the photos I had taken that I found, on what is now the top of
the control unit:
As I knew, there's no way to anchor the components in the battery box! They fit inside,
sort of, but only just, and I needed adhesive tape to hold the controller in place until
I closed the box:
But finally that was done, and there was just the controller to wire. That was
straightforward enough, though the devil's in the detail: the screws for the terminal panel
at the bottom are underneath, so it's much easier to attach the wires before screwing the
box to the post:
And yes, there's no power supply attached; that doesn't make sense unless we really run into
trouble with the PV charger, and the power supply isn't waterproof.
Next, associate the “remotes” (remote controls). The instructions were a bit vague or even
contradictory:
Press and release the CODE SW button, the CODE LED will be ON, then press the key in the
remote twice in 4 seconds, the CODE LED will flash for 3 seconds and then to OFF. Now the
remote has been programmed successfully.
NOTE: The button of the remote control should be pressed and hold for more than 2 seconds
while programming.
But the 2 seconds didn't seem to be necessary. What the instructions didn't say was that
immediately after associating the first remote control, the motor started running. But the
gate didn't move. I had thought that this was because I had disengaged it, but it seems
that the motor automatically reengages when it starts. The noise it made was like a screw
going off the end of its range.
Why? I had installed exactly as instructed. OK, disengage the motor and half open the
gate. Bingo! It ran. And didn't shut completely. The whole careful adjustment of the
closing position hadn't worked, and I'm pretty sure that I'm not to blame (for once). In
the end adjusted the position of the motor with the adjustments that they provide. Here
last week and after adjustment:
Then put the retaining pins in for the joints at the end of the motor. Surprise! The one
at the gate end didn't fit. It seems that, though it's smaller than the other end, it takes
the longer pin.
But finally it was done. We had to remove the dropper at the “open” end, since it opens
further than that (I wonder what people do if there's a wall in this position), and the
second time it closed it wasn't as exact:
The position of Limit Switch A was fixed in factory, do not adjust it again. Plug on the
power to running gate opener, use a screwdriver to loosen the screw of Limit Switch B,
slide Limit Switch B to the desired closed position and fix it. Limit setting for Gate 1
is finished now. NOTE: Always place the magnetic ring between the Limit Switch A and B.
And then there's this:
WARNING: Ensure the gate opener is Power Off when you make any adjustment. Keep away from
the gate, while setting the system in case of any unexpected gate moving. Carefully adjust
the DIP switches to avoid the risk of machine damage, injury or death. Always ask the help
of professional technician /electrician if you have any question.
OK, people, how do I Power Off? There are no switches! I would have to open the control
box and physically disconnect the batteries. And which DIP switches must I “adjust”? There
are three: Push/pull operation (never needing “adjustment”), auto close and photo beam.
But it's working. Our neighbours had noted that theirs was slower than they would have
liked, but that's not the case for ours, though I don't see any likelihood of it attacking
me while I'm adjusting it. I'll consider shut point adjustments later if that proves
necessary, and I most certainly won't disconnect the power to do so.
For some weeks now I've had the internal unit for the weather station in the lounge room.
On the one hand, the temperature in the lounge room is more interesting than in the office,
but the real reason was to have a line of sight to the outside unit, in the hope that the
communication would be better.
The hope seems to have gone to fruition, but a couple of days ago I had a strange graph
(blue line, obscured by the text; the bigger version is clearer):
What kind of malfunction could have caused that spike round 16:00? It's not a single
incorrect reading: it took 90 minutes for the temperature reading to return to normal.
How much are old cameras worth? People are still doing a roaring trade in 35 mm film
cameras. Even I bought an Edixa
Reflextwo years ago. But what about more modern cameras?
Are DSLRs old enough to be collectable? It
occurred to me that the oldest real DSLR (one that was designed as a DSLR from scratch,
rather than an evolution of
film SLRs) was
the Olympus E-1, introduced only 17
years ago. Looking at the specs, it's amazing how much things evolved between then and when
the E-3 was introduced, only 4 years
later.
Last year I bought an E-30,
effectively an E-3 with less environmental protection. Even the manufacture date in the
“secret” parameters is November 2007, a year before it was announced, but matching the
announcement date of the E-3 quite well.
Looking at the specs, a lot had happened between September 2003 and November 2007. Apart
from considerably better functionality, notably “live view” (the precursor to mirrorless
cameras) and image stabilization, the sensor resolution had increased from 4.9 MP to 12.3
MP, and the highest ISO sensitivity had gone from 800/30° to 3200/36°, and the rear LCD
screen had more than doubled in area.
Wouldn't it be fun to have an E-1, just for comparison? Unlike the Edixa, I could even take
photos with it. So when not one, but two showed up on eBay for a starting price of $100, it sounded like a good
idea. How much should I offer? The seller had a “Buy It Now” of $250, but nobody believes
that. A maximum of $150 should be enough.
To my surprise, five other people bid on them, and they finally went for $306. The winning
bidder must have cursed not buying them for the “Buy It Now” price. But what did the people
want them for? For collectables they're too expensive, and for real use the specs are
insufficient.
How have things changed for us as a result of the curfewquarantine
lockdown? Hardly at all. We bought a couple of face masks, and I wore one in town
2 weeks ago. But why not use it as the title photo for my diary for the duration of
the restrictions? So Yvonne came to take a photo of me with
the mask.
Not easy. For the first couple of shots the flash didn't fire:
Why not? I've had issues with the triggers in the past, but this time I had to wipe the
contacts on the camera with my finger. We've probably never taken flash photos with it
before. OK, finally we get a flash:
Huh? Oh, yes, Yvonne's camera is set to auto-ISO, and the camera didn't know about the
flash, so it took the photo at 3200/36° ISO, 3 stops too bright. Set to 200/24° as
intended. Finally we got a photo:
There's something about the Covidly web site
that browsers don't like. I've had to shoot down browsers on multiple occasions, but today
things were worse than usual. On teevee the whole display hung, including the mouse
cursor. The firefox process was
using 100% CPU time, and I couldn't stop it: it was in STOP state. I thought that
only happened when it was in a debugger, but clearly that wasn't the case, and sending
a SIGCONT didn't help. In the end I had to reboot the system Yet Again.
Is this hardware? That's what I thought last time. But it's looking more like a software
issue now. How can I debug this? Take a processor dump before rebooting? My tools are
long worn-out, even if they're still described
in gdb(4).
Using them would first require bringing them up to date.
VictoriasCOVID-19
restrictions have been eased. I forget most details, because they don't apply to me. Now
there can be up to 30 people present at a funeral! And, I think, 5 people can get together
at a time, and up to 10 people can play sport together, carefully planned so that games like
football and cricket are still out of bounds.
And what did I do? Nothing worth talking about. I should, of course, have been
upgrading eureka, but once again my enthusiasm wasn't up to it. What the hell, why
not have a day doing nothing?
Today was the first really cool night this year. The weather station measured a minimum of
1.4°, which probably meant that the ground temperature was round freezing. And for the
first time since the replacement of the controller, the air conditioner had to perform a de-icing cycle.
I've grumbled about the poor duty cycle in the past: in cold weather it only heated for half the time. But today that was
different: it ran for fully 90 minutes before de-icing, where previously it would start
de-icing after less than 30 minutes. Even now, though, it took 10 minutes to recover. Why?
But it does seem that the behaviour of the new controller is somewhat different.
Call from Karen Rickard (?; a man) from Ballarat Solar Panels today. He wanted to
come and remedy the issues that Daryl Buchanan had identified last month, notably the missing labels. He brought with him a document that included
the following text:
“CUSTOMER UNSURE OF SYSTEM INSTALLED AND THE CORRECT FUNCTION OF THE CHANGEOVER SWITCH”. Of
course I know what system is installed and how the changeover switch works. It was this
idiot Buchanan who operated it incorrectly, twice, after I had told him how it
worked. I find that particularly offensive when he had caused us considerable pain by not
listening, especially since he didn't need to operate it in the first place: he didn't
understand the wiring even after I explained it to him.
To be fair, Karen (or whatever his name was) was confused too. It seems that these strange
“hybrid” systems are not very well understood, nor frequently installed. But why not? It
seems the most obvious way to do things if a grid connection is available. We needed some
discussion to clarify the function of the switch. For me, it's straightforward: it's a
bypass switch for the inverter. In the UP position it feeds the grid into the inverter, and
the inverter feeds power to the house. In the MIDDLE connection, everything is disconnected
(that's the one that Daryl Buchanan selected not once, but despite my admonishments,
twice). In the DOWN position the inverter is out of circuit and the grid feeds directly to
the house.
But what use is the big one? How can it explode? How do sparks or flames cause danger? My
understanding is that Lithium-based batteries die from electrical mishandling, not the
presence of sparks or smoke in the environment. What happens if my car generates smoke when
it starts?
In general, all these labels don't help much. I had to explain the component layout to both
Daryl Buchanan and Karen. What we need is an overall circuit shown somewhere, not all these
individual components.
And almost as a sting in the tail, Buchanan complained that “pips” (the screw covers on the
switch housings) were missing. Yes, he was right:
That shows him measuring the wrong side of the PV array switch and wondering why the voltage
was wrong (something that I, UNSURE CUSTOMER, had to explain to him).
One thing that he mentioned last month, but didn't include in his report, and which I
thought would make sense, was some kind of protection to avoid driving the car into
the battery:
While talking to Karen, he mentioned that he had taken a course, and some inspectors also on
the course didn't like the regulation that electricians should not use ladders made of
conductive materials. It seems that aluminium ladders are by far the lightest, but the
regulations prohibit them. I don't really see why they should be a safety concern, but
clearly inspectors should adhere to the regulations.
It's been a while since I ate huevos
a la tigre, my take
on Huevos a la flamenca.
But today we had a broken raw egg to process, and recently I've discovered
a mandolin (or is that mandoline?)
that cuts good julienne strips. So
off to try it out:
Problem: it doesn't look like it, but it was (more correctly they were) in the way. Somehow
I managed to knock one off the wall and break the glass. Petra Gietz managed to kill
another. And now Yvonne has outdone both of us and broken
another two.
Why? Somehow it's easy to hit with a broom handle. The last one is still working, without
the glass, but I've finally decided that it's time to mount it higher. Here then and now:
A few days ago I commented on the prices that a pair of
old Olympus E-1 cameras brought, and
wondered why. But then another one popped up: $522! This is a single, used body of a
camera that's really only of interest to collectors. Chris Bahlo bought her OM-D E-M10 Mark
II a few years back, brand new, for $499.
It was long enough ago that Olympus don't want to know about it any more, but basically
it's almost identical to the E-M10 Mark III
And I bought my second E-30 last
year, with two lenses, for $250.
Alright, this camera comes from Japan, and I've seen a surprising number of surprisingly
expensive cameras offered in Japan, but I still don't understand how they can get that kind
of price.
But then I saw a real classic camera for sale,
a Nikon F with Photomic viewfinder (a
pentaprism with built-in TTL exposure metering), something that I lusted after when I was a
boy. OK, not the best of condition, and it looks as if the Photomic isn't correctly
mounted:
Still, that's a really classic camera. I found another one (in excellent condition,
however) going for $1,460, with only a conventional pentaprism. Watched it for a while, and
90 minutes before end the bids were still only at $60, so I put in a snipe—and won! $96. I
wonder if the experts can see defects that I haven't. At least I established, based on
this list, that the
camera was built between January and March 1970. Also found this page that gave me some information about the Photomic prism, though I still
don't know (nor care too much) whether it was a Photomic FT or a Photomic FTn. Maybe it's
written on the head itself, though this video on the subject doesn't seem to suggest it:
What makes it stand out (apart from the fact that it's flowering now, while most acacias
flower in spring) is the strong perfume, which hits us every time we walk past. Wouldn't
that be a nice shrub for the garden?
But what is it? I took a twig to examine more carefully:
Then an online search, which proved to be very unhelpful. The closest I came
was Acacia stricta, but they bloom
in spring, nobody mentions the perfume, and the flowers grow close to the stem, as this
photo from VICFLORA shows:
One of my new toys is a mandolin. Or is that a mandoline? My spell checker opts for
“mandolin”, but this is a kitchen utensil, not a musical instrument.
And according to Wikipedia, it's
spelt mandoline. According to
OED, it's mandolin, with the comment
Frequently mandoline.
More to the point, though, is the pronunciation:
Brit. /ˌmandəˈlɪn/, /ˈmandəlɪn/, /ˈmandl̩ɪn/
U.S. /ˈˌmændəˈˌlɪn/
But Wikipedia is strongly US-centric, so their pronunciation was
/ˌmændəˈliːn, -ˈlɪn/ or /ˈmændəlɪn/;
OK, I can fix that. Oh. This easy-to-use Wikipedia
uses MediaWiki, an easy-to-use
substitute for complicated markup languages
like troff
and TeX. Simple, right? Just paste the IPA
into the editor window.
After 30 minutes I gave up. The idea is a template to lead the user to a relatively useless
description of IPA for the language in question (in this case English). Each Individual
Glyph needs to be separated by a vertical bar, so to get /ˌmændəˈliːn, -ˈlɪn/ you
need to enter:
Naughty boy! You entered a, which isn't valid. Never mind what OED thinks. It
didn't complain about the l̩; it just repeated the previous IPA snippet.
Life's too short to mess around with this junk. I considered just ditching the update, but
in the end I just put in the raw IPA without the template stuff. That will probably upset
somebody (probably US American) who finds that it violates some Wikipedia policy, like
the time some well-meaning person reverted a fix to
the cron page, made
by Doug McIlroy and submitted
by Ken Thompson (the author
of cron) because there was no published reference.
Over the course of time Yvonne has bought no less than three
tubs of garlic butter, which I almost never eat, and which she only eats with beef filet.
One is in the deep freeze, an opened one expired about 2 years ago, and there was a full one
that expired only two months ago.
Nothing for it: we'll have to eat beef filet, and for once I ate it with garlic butter. And
accompaniments? Broccoli (Yvonne's favourite) and pommes soufflées. I had made some
comments in the past that they didn't work very well, but this time I decided to take photos
and consider what went wrong.
They look nice, but they didn't blow up. It wasn't until we ate them that we discovered
that, despite the appearance, they weren't really cooked. Clearly they had been fried too
hot.
This really does need considerable preparation. I haven't even decided what approach to
take next time, but a bit of research might help.
How is our house oriented in relationship to the compass? In the past I have had the
impression that the plots of land in the area were measured with magnetic compasses and
without a clear understanding of the deviation. Even
the City of Melbourne suffers
from this problem.
In principle we're oriented with the long side of the house east and west, but it's clear
that it's turned slightly clockwise. I had guessed a deviation of 12°, but there's a way to
find out: wait for the sun to cast a shadow parallel to a wall of the house and use the
NOAA Solar Calculator to
show me the azimuth at the corresponding time.
The sun shone parallel to the west wall of the lounge room at 11:51:45, which today
corresponded to an azimuth of 8.29° at -37.80024, 143.75095. So that's the value, to be
confirmed on some other occasion.
I don't look at the system log on eureka that often, but the last time I did, I saw:
May 17 08:28:59 eureka grog: ssh w4 terminated
May 17 08:29:34 eureka last message repeated 14 times
May 17 08:31:39 eureka last message repeated 50 times
May 17 08:41:42 eureka last message repeated 240 times
May 17 08:45:22 eureka last message repeated 89 times
That's clearly one of my keep-alive scripts: when logged in to a remote system, I can time
out thanks to the NBN. OK, do it in a loop,
wait a few seconds and then try again.
But this one isn't going to work. w4.lemis.com was a temporary name for what became
(I think) lax.lemis.com. So I can just stop the ssh loop.
Or can I? Where are they running?
=== grog@eureka (/dev/pts/19) ~ 89 -> ps aux|grep w4 USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
grog 7535 0.0 0.0 52604 4328 30 I+ 12Feb20 0:00.04 ssh -A w4
grog 7548 0.0 0.0 52604 4328 46 I+ 12Feb20 0:00.06 ssh -A w4
/dev/pts/30 and /dev/pts/46. OK, where are they?
I didn't find them. That's not difficult: I have 5 X displays, with 18, 46, 53, 36 and 11 windows, a total of 164 windows. Sure, I just need
to look at the xterms, but I still didn't find them. OK, what else is running on
those windows?
=== grog@eureka (/dev/pts/19) ~ 90 -> ps ut30 USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
grog 7535 0.0 0.0 52604 4328 30 I+ 12Feb20 0:00.04 ssh -A w4
grog 36280 0.0 0.0 17888 2244 30 Is 14Sep19 0:30.52 /usr/local/bin/bash
=== grog@eureka (/dev/pts/19) ~ 91 -> ps ut46 USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
grog 7548 0.0 0.0 52604 4328 46 I+ 12Feb20 0:00.06 ssh -A w4
grog 49740 0.0 0.0 17888 2244 46 Is 24Oct19 0:17.38 /usr/local/bin/bash
OK,
=== grog@eureka (/dev/pts/19) ~ 94 -> kill 7548 === grog@eureka (/dev/pts/19) ~ 95 -> ps ut46 USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
grog 33584 0.0 0.0 8272 1932 46 S+ 4:23pm 0:00.00 sleep 5
grog 49740 0.0 0.0 17888 2768 46 Ss 24Oct19 0:17.38 /usr/local/bin/bash
That makes sense. That's my loop. OK, dammit,
=== grog@eureka (/dev/pts/19) ~ 99 -> kill -9 49740 === grog@eureka (/dev/pts/19) ~ 100 -> ps ut46 ps: /dev/pts/46, /dev/tty46, and /dev/46: No such file or directory
That should do it. But there must be an easier way.
This was the first weekend since the relaxation of the quarantine laws
for COVID-19. For me, nothing unusual.
Linda Swift came to watch Yvonne work with her horses, and I
later saw her talking to Diane next door.
But then I had to go to Chris Bahlo's place for some photos, and saw two sets of neighbours
in Bliss Road also doing something outside with what appeared to be visitors. And even on
IRC nothing was happening. What's all this fascination with being outside?
Yvonne's new Olympus OM-D E-M5 Mark III has a feature that I haven't seen on any other camera: it can take 1920×1080 video at 120
fps. That could be quite useful for examining horses' gaits.
Today Chris Bahlo was doing some jousting training, along with Amber Fitzpatrick and
(exceptionally) Yvonne. Over with her camera to take some video.
Note to self: familiarize yourself with the buttons before taking video. The Olympus
cameras have a dedicated button for video, so you can interleave video and stills. But the
E-M5 Mark III has its button further back than on my cameras, and to complete the confusion
there's another button where the video button is on my camera, which doesn't seem to do
anything. It's marked “Exposure Compensation Button”, but it's not necessary for setting EV
offsets.
And once again there's an issue of focus. It seems that 120p doesn't allow continuous
autofocus (presumably the processor is too busy doing other things). In my situation that
wasn't a big issue: f/8 at a distance of between 50 and 100 m doesn't require refocusing.
But somehow I still managed to get some shots which were badly out of focus. Clearly a case
for RTFM.
I spent a considerable part of my youth in what the British call
the West Country, really the extreme
south-west: Somerset, Devon
and Cornwall.
And in those parts there's a dish
called Cornish Pasties. I've eaten
many times in the course of the years, but as far back as 1958 my mother, quite a good cook,
explained to me that it's almost impossible to make a good Cornish pasty at home.
Still, how hard can it be? Now that we have
the World-Wide Web, I can go
looking. And how about that, a genuine recipe from the Cornish Pasty Association. It went
into more detail about the shortcrust pastry than about the filling. But how important can
that be? Now we have deep-frozen shortcrust pastry, and that should do the trick.
Chop everything into cubes, salt and pepper. How big should the cubes be? Went off looking
and came up with these videos:
The first is interesting because it shows exactly what I wanted to see, including weights
and dimensions (1 cm cubes). But the accent of the presenter shows that she comes from far
further north, something emphasized by her closing words “one of the most famous foods to
come out of the South of England”.
The other one is clearly from the south, and it's greatly different from the first. No
cubes, and even the dough doesn't look much like any of the other recipes. I didn't pay
much attention to it this time round, but it might be interesting for the next time.
OK, start with 500 g beef skirt, 450 g potatoes and 250 g swedes. Cut the potatoes and
swedes into 8 mm slices and then cut them with a knife:
Round here it became apparent that I had far too many ingredients. A pasty
(regulation 20 cm/8" diameter) takes about 150 g of filling, and I had 1.1 kg for 4
pasties. Managed to squeeze up to 170 g in some, but it was a tight fit:
It all took longer than I had planned, and they were still quite hot when they came onto the
table: How did they taste? Edible. Yvonne liked hers, and
this attack seems only to have been intended to cool it down more quickly:
But they didn't taste right. I later discovered that I had forgotten the onion, but the
real issue was the pastry. My recollection of Cornish pasties is that they're relatively
soft and unglazed, unlike all photos I've seen. Next time, at the very least, I'll make my
own dough. And, of course, get the recipe right.
It wasn't until much later than I found that the Cornish Pasty Association has another recipe with different quantities. Why?
Finally got round to processing
the frijoles negros that Yana
bought in December. I'm still puzzled: there were about 850 g of them, not a quantity that
you would normally buy in a shop (though it occurred to me later that it could be 2 of the
old Avoirdupois pounds).
After soaking they swelled up to a total of 1.82 kg, a factor of 2.14:1, and more than I
recall from other beans, though I'll have to check that. At least it matches the claim in
the Wikipedia page:
One cup of dried black beans yields approximately 2½ cups of cooked beans.
Cooking took about 4 hours, and the following morning they looked like this:
And the taste? Alright, but I don't see a big difference from red kidney beans. Yvonne says that she prefers the kidney beans, but we have another 17
helpings of these before she has to make up her mind.
Part of the effort of setting up the shelving in the garage is to put things that are too
sensitive for the shed (which can become moist on occasion, and frequently hot), thus
enabling me to move some of the mess out of my office.
It's painful stuff, mainly because I'm not used to being tidy. Today I moved much of the
stuff out on the west side of my computer desks, where random stuff has been piled for the
past 5 years. To my surprise I found no fewer than 8 keyboards (including a Northgate
OmniKey) and 5 computers. One was only a motherboard, one is something ancient that I once
used as teevee while we were still in Wantadilla,
one a ThinkCentre that (I think) is
defective. But then another was complete, and the fifth was only missing a disk.
What are they? The last two both had fanless display cards, so maybe they were old
instances of teevee or tiwi. If the machine that I'm currently (very slowly)
setting up as dereel really proves to be defective, at least I have two backups.
Yes, this is a severe crop, but I've taken far too many macros where the details look like
that. It can't be camera shake (flash), clearly not focus (the head is relatively in the
middle of the focus range). More investigation needed.
And taking the photos of the beans was another issue. My exposure meter told me f/11.6.
Exposed ⅔ stop more (f/11) and got:
Just before these photos were taken, we had four days with particularly cold overnight lows,
apparently breaking decades-long records in some places. In all probability the ground
temperature went below 0° every night, though the weather station (3.5 m above the ground)
measured between 0.5° and 2.2°. Probably it has affected some of the plants, but it's too
early to be sure. Only the daun laksa
(Persicaria odorata) is
looking significantly unhappy, though new shoots are growing in the undergrowth:
The mistreated Epazote seedlings that I
planted at the end of March also look better, though I'm not convinced that they'll make it
through the winter. Here then and now:
The original Hibiscus
rosa-sinensis “Uncle Max” has not taken kindly to being repotted. Strangely, whole
branches have died, while others are doing well:
Other plants are behaving appropriately to the season. For only the second time since
moving here, the Betula pendula
(Birch) are behaving as if it were autumn:
Hopefully it will handle the winter better this year.
Another plant of some concern is
the Robinia pseudoacacia
that we transplanted last month. It had started to grow new leaves, but they're not looking very happy
either.
The Banksia integrifolia has
been one of the success stories. It has grown well and consistently since we planted it,
and it's now 4 or 5 m high, with many cones:
Last month I noted that one
Spathiphyllum plant, in the dining
room, was quite sickly, and moved it to the bathroom. I thought that it had improved, but
in fact there's very little difference to be seen. Here last month and now:
One of the results of the SARS-CoV-2
pandemic was the shortage of certain food ingredients. We still can't get yeast—a good
thing I use mainly sourdough—and for a
while pasta was a problem as well. At one point Yvonne bought some bucatini, which I had
thought looked pretty much
like spaghetti. But today I cooked
them and found that they're much thicker:
The package is marked “Number 6”. I don't know if that relates to the thickness or just a
company internal numbering. When I followed up on the description, I discovered that
bucatini have a buca or hole down the middle. I didn't notice that, but on checking,
yes, there's a tiny hole down the middle. I wonder what good that is.
I use them instead of various Malaysian noodles, and quite possibly the thickness is a
better match for Hokkien or
laksa noodles.
From time to time I've heard people talk about colour rendition of lenses, and I haven't
paid much attention. But today it reached out and grabbed me. While taking photos of the
indoor plants, I found this:
One of the problems with my Cornish
Pasties on on Sunday was that I wasn't happy with the pastry. Another was that I still had a
lot of filling left over.
For shortcrust pastry (rough puff can also be used):
500 g strong bread flour (it is important to use a stronger flour than
normal as you need the extra strength in the gluten to produce
strong pliable pastry)
120 g lard or white shortening
125 g Cornish butter
1 tsp salt
175 ml cold water
Method
1. Rub the two types of fat lightly into flour until it resembles
breadcrumbs.
2. Add water, bring the mixture together and knead until the pastry
becomes elastic. This will take longer than normal pastry but it
gives the pastry the strength that is needed to hold the filling and
retain a good shape. This can also be done in a food mixer.
3. Cover with cling film and leave to rest for 3 hours in the fridge. This
is a very important stage as it is almost impossible to roll and shape
the pastry when fresh.
4. Roll out the pastry and cut into circles approx. 20cm diameter. A
side plate is an ideal size to use as a guide.
That seems straightforward enough, but the devil's in the detail. At the risk of ruining
the recipe, I substituted Australian butter for Cornish butter, but I do appreciate the
(almost complete) use of sane measurement units. How much is a teaspoon of salt? The one I
used held 7 g, but according to my 2% rule, I put 10 g in.
Knead. In a British (if not
Cornish) Kenwood mixer. Sometimes I
think that this thing is out of date, and that other constructions would do better.
Certainly getting “breadcrumbs” was not easy. But finally it was done. Then:
Cover with cling film and leave to rest for 3 hours in the fridge.
Oh. No time for that. I'm reminded of what I wrote 50 years
ago, just after I had moved into my first own domicile:
why can't I face facts and learn to cook a few things that don't take such a long time?
OK, no pasties tonight. Yvonne had a plan for using up the
old pastry, the stuff I didn't want: quiche lorraine. But by 17:30 she hadn't started, and I was getting nervous: it had
to be on the table at 18:30, and it required 45 minutes' baking.
Somehow, with a little assistance on my part, we made it. The baking time was only 40
minutes, not 45, and there's really not much preparation. In addition, it was “just right”,
something about which we hadn't been sure.
Somehow I've spent half the week cooking. It hasn't all been fun.
Today a batch of baked beans were on the
agenda. I put some pork rind in there, but how much? Today, for some reason, I only had
660 g of (dried) beans, and for that I put in 100 g of rind, possibly more than is a good
balance. We'll see.
And then there were the Cornish pasties of
which I made such a mess on Sunday, and which I mistimed yesterday. Today I gave myself plenty of time, and
used most of it.
Firstly, the dough was elastic, unlike the deep frozen dough I used last time. That meant
that the carefully cut 20 cm circles of dough immediately reduced in size to about 12 cm
when I took them off the benchtop. I found it easier (on the last attempt) to pull away the
surrounding dough and fill the circle in place.
How much salt and pepper? They almost never specify quantities, even in recipes that
are otherwise good in their specifications. “1 teaspoon”? I established yesterday that 1
teaspoon was 7 g, and that 10 g would be a better fit. But now they wanted “ Salt &
pepper to taste( 2:1 ratio)” (their punctuation). What does that mean? “2:1 ratio” may not
be my taste. And if it were, what kind of pepper? 2:1 by weight or volume? I decided that
it would probably be white pepper, and by volume. Measuring one teaspoon of pepper gave me
2.6 g, so the 1⅓ teaspoons (10 g) that I had chosen for the salt would correspond to 4 g of
pepper.
From then on, things went downhill. I weighed out my 4 g of pepper and put them in the
mixture, along with (finally, this time) the onions. Started filling the pasties, which was
made more difficult by the flexibility of the dough. Maybe that's what they meant when they
wanted to leave the pastry in the fridge for 3 hours. The implication may be that the dough
should be very cold. As it was, crimping was very difficult, and the pasties didn't look
nearly as good as Sunday's.
And there was enough filling for 6 pasties! I used up nearly all my pastry, and it wasn't
until I started filling the 5th that I realized that I had forgotten the
salt again! And of course I had measured out the pepper on a 1:1 ratio, not a
2:1 ratio, so to make up for no salt I had double the pepper. Why am I having so much
difficulty with this recipe?
Baked them at 160° this time, for 40 minutes. They tasted better because of the onion, but
there was no significant difference in the pastry, and they still don't taste like the
pastry that I recall from my childhood.
So: some time in the future I'll make a third attempt, with frozen pastry. But first we'll
need to eat the 6 pasties that we have left over. I have a feeling that I'll have to eat
most of them myself.
Into town today for a dentist's appointment, almost as planned before the panic. It was
actually for yesterday, but they needed to reschedule. Nothing of interest.
On to Beaumont Tiles to
look—once again—for tiles for the verandah. It's only been 5 years since we laid the slab. But this time I found a couple of tiles that didn't
look too bad:
Then on to Bunnings , this time the one in
Creswick Road, because I was in the area. They've made different and equally complicated
arrangements for access:
The actual entrance is hidden behind the cars on the left of the last view. And I admire
the distance that they require between people (which, of course, nobody observed).
Inside, they were the only place I've seen to actually sell serious face masks:
Only 4 per person? How long do these masks last? But the shelves looked well stocked, and
I didn't see anybody buying them. That's not surprising: during the entire time in town, I
wore my face mask. I don't know why: I was the only one. I went to the barber's, where
Kerry took one look at me and asked me not to shoot: clearly I was a robber with a mask on.
I didn't get the hair cut, though: too many people there.
And at the dentist's? No, even there only Mario, the dentist, had his usual face mask on.
They did ask me to sanitize my hands when I came in. Somehow people aren't taking
this overly seriously, and maybe they're right.
On to OfficeWorks, who had their own
entry restrictions:
When I was a lad I lusted after a Nikon
F, one of the iconic cameras of the 1960s. And today I received the one that I bought
on Saturday.
It's a little the worse for wear, as the photos showed. And as I feared, it came without a
body cap. But to my surprise, an
Olympus Four Thirds system
body cap fits:
It's not a good fit: a bit of a shake and it'll fall out. But it engages and closes the
hole. I wonder if the Olympus people thought of that when they
There's clearly something wrong with the Photomic head, which doesn't fit the way it should,
and I discovered that there's a screw missing on the side of the head:
And the shutter? That's always an issue with older cameras. But it works surprisingly
well, so well that I thought there was something wrong. Compared to
a Spotmatic it's almost instantaneous.
With the Spotmatic you can clearly hear the mirror going up, the shutter releasing and the
mirror going down again. With the Nikon there's just a single sound
And then of course there's the leatherette covering. Is there a way to repair that? Do I
care enough?
tstamp Pac Status SOC VBat PacGrid PacBat PacPV FromPV W1
2020-05-22 12:11:22 2440 Waiting to connect to On-grid 24 258 NULL 1493 1003 0 0
2020-05-22 12:11:23 2787 Waiting to connect to On-grid 24 256 NULL 1768 1019 0 0
And that was all. Normally this kind of message indicates recovery from a grid power
failure, and there are at least 60 such lines, since they come every second, and the
inverter waits a minute before reconnecting. Why were there only two?
Where did the entries between 12:08:25 and 12:11:22 go?
After a bit of thought, it became clear: I was in town, but Petra had had problems with the
floor washer, which had blown multiple circuit breakers. It appears that it had blown the
mind of the inverter as well, so now we know that it takes it about 3 minutes to restart and
come on line, though presumably it had started working nearly a minute before the first
message.
Also yet another indication that I should keep the UPS in front of my
office computers. Both lagoon and teevee lost power.
In my youth I frequently ate kippers for
breakfast. When I moved to Germany I could no longer get
them. But now that I'm in Australia, I
should be able to get them again, right?
Well, sort of. The only ones that I have been able to find so far have been two whole
kippers (two fish) frozen together, an idea that makes you wonder what the maker is
thinking.
But recently Yvonne found some kippers in a can that looked
far too small to hold a whole kipper, made in Poland of all places. Today I finally opened
it:
And how did they taste? Like canned fish, nothing like the aroma of a fresh or thawed
kipper. There were two servings, and I'm hoping I can convince Yvonne to eat the second one.
I postpone a lot of email messages, far too many in fact. Currently there are round 1,170
messages in my postponed folder, the oldest of which is dated 4 October 1999.
There's even this message:
Date: Mon Nov 15 11:28:41 1999
From: Greg Lehey <grog@mojave.sitaranetworks.com>
To: Dennis Ritchie <dmr@alice.att.com>
Subject: Re: Cute Comments (was Re: Bad commenting style)
Alas, it's too late to send it now, but it's interesting to look at the email addresses (for
once, not changed). mojave was, I think, my laptop, which I got from Jerry Dunham,
who worked for Dell laptops at the time.
The system name was apparently the internal project name for the laptop. And I was
in Waltham, Massachusetts
at the time, working for the now apparently defunct Sitara Networks.
And then there was this message, coincidentally also involving Jerry Dunham:
Date: Thu, 14 May 2020 11:40:30 +1000
From: Greg Lehey <groggyhimself@lemis.com>
To: Jerry Dunham <jerry@dunham.org>
Cc: Wes Peters <peters@softweyr.com>
Subject: Re: No joy
Message-ID: <20200514014030.GE1670@eureka.lemis.com>
On Friday, 17 December 1999 at 23:23:05 -0600, Jerry Dunham wrote:
> On Fri, 17 December 1999 at 15:54:34 -0700, Wes Peters wrote:
...
Well, I could answer that one, and I did. And it bounced. Can't find dunham.org.
But only after 5 days. What was wrong there? Clearly dunham.org still exists, or it
would have bounced immediately. What does whois say?
So it's been around for over 22 years, and it's still active—with completely invalid name
server specifications. Both echunga.lemis.com and battunga.lemis.com ceased
to exist thirteen years ago. What does he use it for?
Nothing, it seems. For obvious reasons jerry@dunham.org is no longer active.
Sent him a message to his new email address, and it seems that yes, in principle he still
wants the domain. OK, we can do that. What do I need?
A new zone file. That's pretty bare-bones, since he doesn't even have a web site. Took
the zone file for narrawin.com, changed the names and removed the entry
for www. I also only have one name server under my direct control, so commented
out the other until I can contact the people who run the server.
Update named.conf with an entry for dunham.org, again a modified copy of
the entry for narrawin.com.
Revive battunga and echunga, both pointing to www. I won't check
in this change; once Jerry has updated his domain data they won't be necessary.
Add an entry to my virtual file:
jerry@dunham.org jerry@texas.net
Total time: about 10 minutes. And it worked. It's nice to see something Just Work.
A week ago I went on an abortive search for an ssh that kept trying to connect
to the now-defunct w4.lemis.com. While setting up the mail forwarding for Jerry
Dunham, I finally found the source: it was an xterm that had connected to www.
The display looked something like:
ssh: Could not resolve hostname w4: hostname nor servname provided, or not known
bash: /usr/local/bin/xtset: No such file or directory
Disconnected at Sun 24 May 2020 10:13:53 AEST
^C
So why didn't I find it? The second line is the clue: xtset is a program that sets
the title of an xterm. I had moved it from /usr/local/bin
to /home/local/bin, which requires a hash -r
for bash to recognize it. And
the ssh was running, so I couldn't run hash -r. As a result,
the xterm title showed the previous details (usr, system and directory):
Chris Bahlo came for dinner this evening, in the process delivering the meat for the main
course. Yvonne had understood that it was beef filet, but
what she brought looked like an
oversized rack of lamb. Here after
cooking and carving:
What is it? Off to the web to investigate. It's a rib roast. Or maybe
a standing rib roast.
Or maybe just prime rib, as the US Americans call it. I have eaten it under the last name,
and it tasted very good.
OK, how do I prepare it? It looks like a roast. Somehow it's impractical: ours had three
ribs and weighed 1.44 kg. In general I reckon 180 g per person (120 g for Yvonne) for filet. A little more for fat and bone, but it's still
enough for 6 people. How do you work around the bones?
And how do you roast it? Found a surprising number of recipes, including this one from the Australian Women's Weekly, who should know better:
Preheat oven to 180°C. Weigh beef and calculate cooking time (Cook for 20 minutes per 450g
for medium/15 minutes per 450g for rare.).
20 minutes per 450 g? Why that? Why not 1⅓ minute per 30 g? 64 minutes per 1440 g?
Another ghost of the past: Australia has been using the metric system for 50 years, but
people still can't get rid of
old avoirdupois measurements.
But this cut is complicated. The Women's Weekly recipe wanted 2.5 kg of meat, more than
most. Others wanted “four ribs”, about 1.8 kg. But the thickness of all these cuts is
roughly the same: only the length differs. How does that influence the cooking time?
Certainly a time based only on the weight is not going to work. My guess is that the weight
is relatively unimportant when calculating the cooking time.
Then there's a question of oven temperatures. One recipe wanted to start at 230°, others at
160°. More than elsewhere, I couldn't get a good idea of what to do. In the end I went by
my roast beef cooking times and set it at
180°, estimating 70 minutes for a meat temperature of 53°.
And how about that, I was spot on! The only issue is that 53° might be a little warm. Next
time I'll go for 52°. But if I had followed the Women's Weekly recipe, I would have cooked
for 48 minutes, and it would have been excessively rare. On the other hand, 83 minutes for
a 2.5 kg cut might be correct.
Why that? Simple: I always use the central focus point in the camera to focus, maybe moving
away after establishing focus. But here things were too fast, so I ended up with the lid in
the middle of the image, nothing below, and half a horse above. Time to use the
capabilities of the camera.
Another battery recalibration today, almost exactly what I would have ordered. In the past
I had multiple cycles, and I lost significant amounts of PV energy in the process. As I
wrote 2 months ago, a single recalibration cycle once a month would be acceptable, but no
more.
And that, it seems, is exactly what I have been getting. Today the cycle started after
sunset, charged to 100% and discharged to (only!) 10%:
What's special about that? It's the Bahlo coat of arms, which presumably Chris wears in
full armour. Presumably one of the few mugs with that kind of emblem.
And what does it mean? Chris explains:
Description
Per pale, dexter potenty bendy or and gules, sinister sable with a mullet argent of six
points, pierced by a rondel.
Details
Per pale: split vertically
dexter: left side (when worn on the right)
potenty bendy: potenty vair is an abstraction of a "fur"; bendy indicates it is
diagonal.
or and gules: gold and red (refers to the vair pattern in this case)
sinister: right side
sable: black
mullet argent: silver star
rondel: circle
Sources
The potenty vair in red and gold is originally from a 13th Century miniature, showing a
horse caparison (I have a replica of this). The mullet/star/spur rowel is a common
symbol.
What it signifies (to me):
The vair pattern signifies both the complexity of life, but also the fact that there are
patterns and cycles. The colours are (to me) happiness and prosperity. The upward
diagonal is about improvement. The black half is a stark contrast on purpose, it means
calm, privacy, introspection. The silver star is both a spur rowel, which is a symbol
for horsemanship at a high level, and a star, which is about goals and aspirations.
Years ago we drank a considerable quantity of French wine. There were three kinds, in
decreasing order of quality « Appelation nom d'origine contrôllée », « Appelation
contrôllée », « Vin de pays ».
But now we're in Australia, and there's plenty of local wine. Still we get some, but the
descriptions are different: « Appellation d'origine protégée », « Appellation géographique
protégée ». They're not particularly good wines, and my suspicion is that the difference
between « contrôllé » and « protegé » is significant. I think I've even seen a « Vin de
France », presumably something that they can't quite disown.
How do they compare? Time to check. Appellation d'origine contrôlée doesn't help much beyond confirming that it's still
in use, since it doesn't mention the alternatives. This page looks convincing, mentioning two of my three new names (there's also « vin
de pays », also known as « indication géographique protégée »). But where's « appellation
contrôlée »? Presumably it's now the same as « appellation protégée ». Or is it? Time to
check the French
information, Appellation
d'origine contrôlée. What do I read there?
Not to be confused with appellation d'origine protégée. Read on. It's not a question of
quality but of jurisdiction. In France it's AOC, in the rest of Europe (and thus presumably
anywhere outside France) it's AOP.
None of this has much to do with our experience last night. Chris brought a 2017 Château de
Cathalogne Bordeaux AOP, and I presented a 2018 Loire Valley Sauvignon Blanc (clearly a name
intended for overseas marketing) IGP. We didn't like either much.
When I installed the gate opener, exactly according to the instructions, it didn't
work: the opener was off the end of the drive, and I had to open the gate to get it to
engage. And then it didn't shut correctly any more:
But it didn't stay that way. The gate stayed more and more open. I established that the
mounting of the opener was not very firm, at least partially because of the post on which it
was mounted. But clearly the design is not conducive to exact closure. Still, it should do
better than this:
The position of Limit Switch A was fixed in factory, do not adjust it again. Plug on the
power to running gate opener, use a screwdriver to loosen the screw of Limit Switch B,
slide Limit Switch B to the desired closed position and fix it. Limit setting for Gate 1
is finished now.
NOTE: Always place the magnetic ring between the Limit Switch A and B.
OK, what does that mean? The diagram shows a section through the opener body. In reality
it looks like this:
So how do you adjust the thing? I have to make a number of assumptions:
The limit switches are magnets that are held in place by the screws.
When you loosen the screw, you can slide the magnets back and forward.
The magnetic ring senses the magnet and stops.
Under those circumstances, the position of the limit switch B directly relates to the
extension of the piston that drives the gate. I want it further to the left, but it's
already as far to the left as it will go. And clearly the information about the placement
of the magnetic ring is stupid. What they really mean is “don't slide the limit switch
beyond the magnetic ring, the one that you can't see”. But if my guesses are correct, we
know where the ring is: in the closed position, it's next to limit switch B. So sliding it
to the right violates the NOTE at the end.
Still, that's not what I want to do. Out today with some spanners to tighten up the
mounting of the bracket on the post. To my surprise, I was able to rotate the nuts up to
720°. But it didn't make the mounting noticeably firmer.
OK, adjust the angle bracket again. I couldn't get the old bolt out! After loosening it
and the central bolt, I had to move the bracket towards the closed position. And when the
holes lined up exactly, the gate was also completely shut.
So: the bolts are too small for the hole, and the considerable pressure on the plate had
moved it slightly (2 mm?) towards the “open” position. And given the extreme angles of the
thing, that was enough to cause that level of opening. OK, replace the bolt, tighten as
much as I could justify without shearing the bolt. Open. Stop halfway. Close:
Clearly the force on the plate is too much, and the screw is too small. My guess is that
the inertia of the gate stopping pulled the angle bracket out of position. I need to find a
screw that fits the hole, and all should be well.
One of the issues that Daryl Buchanan identified with the PV inspection last month was lack of documentation. Effective Electrical promised to send it
to me, and today it arrived.
It included some interesting stuff, including an overall circuit diagram, complete with
ballpoint impressions of something that had been filled out on top of it.
Unfortunately, it's incorrect. It shows that the battery is on the AC side of the inverter.
It also shows two isolating switches between the PV panels and the inverter, and it doesn't
show the correct switching for the inverter, in particular the bypass switch and the fact
that the inverter is between the grid and the consumers. This latter must be so obvious
that I'm wondering how they came to this circuit.
It also included documentation for the components, the inverter in both Spanish and English,
and documentation for a number of BYD batteries, not including the one installed. Still,
it's a useful addition, even if they claim that they had left it all already. I'll have to
compare what I have to what they claim.
Finally got round to replacing the wind break trellis that was blown away three weeks ago. It wasn't easy: all sorts of things had grown around it, and I
nearly removed the rest of
the Jasminum polyanthum while
positioning it. The Clematis “Edo
Murasaki” wasn't so lucky. It had already been almost uprooted by the detachment of the
trellis, and today I did the rest:
That's all that was left above the ground. Will it recover? We'll see. It might appear
that clematis are inactive at this time of year, but the “General Sikorski” is at least
planning for spring:
PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
38033 grog 3 20 0 3416M 693M STOP 2 17:01 100.23% firefox
And once again I couldn't kill X, and I had to
reboot. It didn't occur to me until later that I could at least have run ktrace
against the firefox process and see what it was doing. But it's beginning to look
less like a hardware issue and more like a kernel bug triggered by a
misbehaving firefox.
The recent problems with firefox hanging on teevee suggest that there's a software problem somewhere. Time for an
upgrade? Built a new world and kernel today, and then fired up.
The mouse didn't work correctly! It took me a while to realize that the button mapping
wasn't working. It took even longer to realize that I had changed neither moused
nor X. Only the kernel had changed; I still needed to install the new userland.
I have a two-button mouse, but X requires three
buttons. So for ever there have been kludges to simulate the third button. The one I use
on teevee is to press both buttons at once. From the man page:
-3 Emulate the third (middle) button for 2-button mice. It is
emulated by pressing the left and right physical buttons
simultaneously.
But that wasn't working. Why? Spent a lot of time checking. First I found a third mouse,
a Logitech M705, the same as I use on eureka. There I used a different
mapping, -m 2=4:
-m N=M Assign the physical button M to the logical button N. You may
specify as many instances of this option as you like. More than
one physical button may be assigned to a logical button at the
same time. In this case the logical button will be down, if
either of the assigned physical buttons is held down.
That in itself is incorrect. The button that I have assigned reports as button 8, but for
some reason I needed to say 4 to convince moused to perform the mapping.
Looking back, it seems that I have had issues here for
decades.
But somehow that didn't work either. Pressing button 8 and checking with xev, I saw:
ButtonPress event, serial 33, synthetic NO, window 0x5c00001,
root 0x299, subw 0x0, time 69545332, (159,159), root:(184,220),
state 0x0, button 8, same_screen YES
ButtonRelease event, serial 33, synthetic NO, window 0x5c00001,
root 0x299, subw 0x0, time 69545502, (159,159), root:(184,220),
state 0x0, button 8, same_screen YES
It didn't remap! It still claimed button 8. OK, moused can also output debugging
information. Tried that and got:
moused: received char 0x87
moused: received char 0x0
moused: received char 0x0
moused: received char 0x0
moused: received char 0x0
moused: received char 0x0
moused: received char 0x0
moused: received char 0x7e
moused: assembled full packet (len 8) 87,0,0,0,0,0,0,7e
moused: ts: 69422 629804517
moused: : 69401 500254541
moused: flags:00000008 buttons:00000008 obuttons:00000000
moused: activity : buttons 0x00000002 dx 0 dy 0 dz 0
moused: mstate[1]->count:1
moused: button 2 count 1
moused: received char 0x87
...
moused: received char 0x7f
moused: assembled full packet (len 8) 87,0,0,0,0,0,0,7f
moused: ts: 69422 785808495
moused: flags:00000008 buttons:00000000 obuttons:00000008
moused: activity : buttons 0x00000000 dx 0 dy 0 dz 0
moused: mstate[1]->count:1
moused: button 2 count 0
So moused appears to be doing the right thing, but somehow X is ignoring it.
OK, what about the old mouse with the -3 option?
ButtonPress event, serial 36, synthetic NO, window 0x5c00001,
root 0x299, subw 0x0, time 69698388, (153,106), root:(178,167),
state 0x0, button 1, same_screen YES
ButtonPress event, serial 36, synthetic NO, window 0x5c00001,
root 0x299, subw 0x0, time 69698388, (153,106), root:(178,167),
state 0x100, button 3, same_screen YES
ButtonRelease event, serial 36, synthetic NO, window 0x5c00001,
root 0x299, subw 0x0, time 69698480, (153,106), root:(178,167),
state 0x500, button 1, same_screen YES
ButtonRelease event, serial 36, synthetic NO, window 0x5c00001,
root 0x299, subw 0x0, time 69698496, (153,106), root:(178,167),
state 0x400, button 3, same_screen YES
Yes, it's seeing two separate key presses, not the chord. But moused sees:
OK, is this an issue between moused and the kernel? It doesn't look like it, but I
installed the new userland anyway. No change.
OK, time's getting on, and I want to watch TV in the evening, not debug recalcitrant mice.
X sees button 8. What do I need button 2 for? Mainly for the window manager, fvwm2. And I can update the configuration for that:
# W: application window
Mouse 1 W M Lower
Mouse 2 W M Iconify
Mouse 8 W M Iconify Mouse 3 W M Raise
And that worked, but it didn't help with an xterm, of course. And somehow there's a
secondary issue that the mouse doesn't work as reliably as it did; marking text sometimes
aborts, for example. But at least it helped me make it through the night.
But what's going on? Apart from rebooting, the only thing that changed was the kernel. And
of course a reboot, but I do that all the time. Things to consider:
Could it be a random issue with X startup? Could restarting X fix it?
Could it have something to do with the keyboard remapping? Clearly something has
changed there, since I can no longer do it automatically on startup. I could check
before doing the (manual) remapping, next time.
Could it be a random kernel issue? Could rebooting fix it?
Could rebooting the old kernel fix it?
Alternatively, could building a new kernel fix it?
How do moused and X communicate? Clearly X is ignoring moused's carefully
faked keypresses. Could the communication be via the kernel? That would explain its
appearance now.
And just to twist the sword, mountd on eureka lost sight of teevee
again, silently. I couldn't access files on eureka. When I tried to umount
and mounteureka:/, I got:
May 27 21:23:52 eureka mountd[98292]: mount request denied from 192.109.197.158 for /
That's nothing new, and one of the things that I hope will be fixed when I finally get the
new eureka/dereel pair working. Restart mountd and all is well, but of
course that's still more problem solving.
Sometimes I wonder if Linux would do things better. But every time I've tried it in the
past, it has not done as well, and there's a considerable learning curve involved.
It's been over three years since I started eating laksa and similar for breakfast, based on commercially available pastes. They're not
bad, but the pastes are sufficient for many servings, and I have worked out various methods
to portion them.
But it has always been difficult to ensure that each portion gets the same amount of the
ingredients. Why bother? I have always had to add some ingredients on each occasion, so
why not add them all? Cook the laksa paste with coconut milk (for curry laksa) and freeze
it.
That's what I did today, also taking note of the fact that the amount of coconut milk
(originally 150 ml, which I rounded up to 200 ml) was only half a can. More measurements,
made more difficult by the fact that the coconut separates from the water, so at the very
least I needed to remix it.
So: two sachets of and one can of Tean's Gourmet curry laksa and one can of coconut milk. Produces 3.425 kg of laksa
base, enough for 8 portions. Use one, freeze the rest:
Where do I find 7 containers for my laksa base? In the pantry, maybe. The corner with the
containers has been a real mess, and I've been meaning to tidy it up for a long time
(years?) now. OK, drag out all the messy boxes:
The circular plastic ones at rear right are the ones I use most. We get them as packaging
for food from the supermarkets, and thus for free. They last about 10 uses before disintegrating. But
they're too small for a 430 ml laksa
portion. For that I've used larger ones, not in this photo for obvious reasons. And then
there are the rectangular ones at front right, quite usable, but not ideal for laksa because
they're wider than the pan in which I want to thaw them.
And then? Almost the entire left side of this photo is pretty useless. Lots of lids with
no matching container, containers without lids, and most of them are just too small to be of
any use. I'll see if Petra wants them.
And while I was at it, how about sponges and scourers? Some time ago I found hundreds of
them hidden in the cupboards, some clearly ancient. We tidied that up, but it proved to be
only part of the issue. Today I found:
Again, we only use some of these things. The pile in the rear middle are 33 sponge cloths,
enough for the next 3 years or so. At front left are two scourer pads, which
will—exceptionally—need replacing in the not-too-distant future.
And the rest? The yellow soft cloths on the right are potentially something that Petra
uses. But these 7 thick sponge pads (centre right) we don't use, though we might. And I
don't know when we last used a sponge with scouring pad, like the 11⅔ pink ones on the left
(for some reason one is shorter) nor the 21 others spread around the pile. And I don't know
if we'll ever used the metal scouring pads, though it's not beyond the bounds of
possibility.
How long have we had this stuff? The pink pads give a clue:
„Vileda kratzfrei“. Vileda is a German
company, the name a homophone of „Wie Leder“ (“Like leather”). They sell products in
Australia too, but they don't have German text. „Kratzfrei“ means “scratch free”, something
that few Australians would understand. So it seems that we brought these scourers with us
from Germany 23 years ago. But instead of using them,
Yvonne bought the other 21 or so, not to mention some that we
have used in the meantime.
So what's causing the problems with the mice on teevee? The whole thing is such a
pain that I'm not sure that I want to know. Compared the Xorg.0.log with older ones
and found no obvious differences. But when discussing it on IRC, Callum Gibson wondered why
I was still using moused. He also didn't understand my issues. Why did I want to
remap the mouse buttons? Every mouse nowadays has at least three buttons. Button 2 is
incorporated with the scroll wheel.
Ah, right, I had (mercifully) forgotten that. It's been over 17 years since I stopped using it because
it was so ergonomically painful. But at least it's a workaround for the time being. Tried
it again, and confirmed. But Callum says:
<cgibson> oh you don't have any standard 3 button mice (with a wheel)
<groggyhimself> You mean this emetic "push the wheel, hope not to turn it" "button"?
<groggyhimself> That's so ergonomically appalling that I had forgotten about it.
<cgibson> Yes annoying isn't it.
<groggyhimself> I suppose I could use that to work around the problem.
<groggyhimself> Is that what you do?
<cgibson> It's not really a workaround, it's just what everyone uses these days.
<cgibson> Mostly it's ok... depends on the quality of the mouse as to how easily it is to
avoid scrolling while clicking
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.
But that's only part of the question. We carried on:
<cgibson> Dunno, but I am still surprised you're using moused with X.
<cgibson> I didn't even know you could.
<groggyhimself> If I stop moused, nothing works.
<groggyhimself> Well, the mouse doesn't.
<cgibson> Are you using a config file or auto?
<groggyhimself> I'm using a config file.
<groggyhimself> I could certainly try without and without moused, and see what happens.
<cgibson> It sounds like you already have the new xorg with the different input methods
<groggyhimself> It's the latest X.
<cgibson> I have a mouse section in my config which refers to /dev/sysmouse. I suspect X is
moving more and more to automatic device detection though, so I couldn't even say
if that's needed.
* cgibson checks work PC
<groggyhimself> How does the New Order handle multiple mice?
<cgibson> work PC just autodetects and uses devd -> /dev/ums0
<groggyhimself> And /dev/ums1? And ums2?
<cgibson> I guess they all just get added as devices.
<cgibson> yes, I'd expect so
OK. I didn't know about that. If it works, don't fix it. But now it doesn't work, so it's
time to fix it. Got Callum to send me his Xorg.0.log files, and then shelved them:
enough pain for one day.
So now my question list is:
Could it be a random issue with X startup? Could restarting X fix it?
Could it have something to do with the keyboard remapping? Clearly something has
changed there, since I can no longer do it automatically on startup. I could check
before doing the (manual) remapping, next time.
Could it be a random kernel issue? Could rebooting fix it?
Could rebooting the old kernel fix it?
Alternatively, could building a new kernel fix it?
How do moused and X communicate? Clearly X is ignoring moused's carefully
faked keypresses. Could the communication be via the kernel? That would explain its
appearance now.
Could it be an old, worn-out method (moused) to access the mouse?
While checking the (very poor) recording
of Al Jazeera news this afternoon, I
saw an advertisement that interested me. A company called WiFi-X who offered a cheap WiFi device that increased
my Internet connection speed.
Huh? What does WiFi have to do with my
Internet link? Watched the full 5 minute commercial, which didn't explain, though it seems
that my ISP is deliberately throttling my link speed, and this device will miraculously fix
it.
Not just lies, but conspiracy theories too. What kind of idiots would believe that? Too
many, I fear: not nearly enough people understand even basic networking. But why is
YouTube carrying that kind of commercial?
We had decided to cook a coq au vin for dinner tomorrow. Where's the recipe? We don't have one!
Well, we have plenty in our cookbooks. But I write all non-trivial recipes up as a web
page, and we didn't have one.
What did we use in the past? There are a number of candidates, but I decided to check the
web for ideas, finding (and mercifully losing again) things like “easy 1½ hour coq au vin”.
A search for coq au vin recette worked better, bringing this recipe and its
variant from Marmiton. But
some of the details still didn't ring true, like the cayenne pepper in the second one (from
mère Michèle). But then I found this one, which looks excellent not only from the point of view of the composition of
the dish, but also the presentation of the recipe. Today I had not only to marinate the
“cock” (really, of course, a normal chicken), but also prepare a broth out of the bones:
Somehow I couldn't bring myself to analyse Callum Gibson's Xorg.0.logs today.
There's enough pain already. But teevee had other ideas: once again firefox hung and couldn't be stopped:
The RE state is a process that is trying to exit. I had already had an idea when
next in this situation. run ktrace. But before I could, it actually finished
exiting! And then X looped and couldn't be stopped.
OK, what's ktrace for the fox is ktrace for the X:
=== root@teevee (/dev/pts/6) /var/tmp 34 -> ktrace -i -p1111 === root@teevee (/dev/pts/6) /var/tmp 35 -> l -rt total 1
...
-rw------- 1 root wheel 0 29 May 17:53 ktrace.out
Damn! No output to ktrace.out. What does that mean? Looping in kernel code? It
wouldn't happen if it were looping in user code, because then I could shoot it down, and if
it were alternating between the two, ktrace would show something. And on one
occasion, top showed
X waiting on nvidia. Both of these suggest that there's something wrong in
the driver. Time to update?
So: reboot. Before doing that, though, investigate Callum's comments about moused.
Disable it, remove the X configuration file, and see what happens.
Nothing! The display had gone blank, and after I pressed the RESET button the display
showed its usual incorrect estimate of the display resolution presented by the graphics
card. But then nothing at all, apart from a bit of backlight. Repeatable.
What's happened now? Serious hardware issues on the computer? Dammit, I really need a
backup in case one fails, but apart from thinking of it now, it's not going to happen. This
was round 18:00, time for dinner and then relaxing watching TV. Except that we have no TV.
On a hunch, power cycled the display (TV). Then it worked! Well, until the next problem:
I've seen that before. This is actually the photo from last time, but the message was the same.
The loader tripped over some issue with the keyboard. Enter boot and it
booted completely normally.
So why did it only happen now? Simple: the last time I booted this kernel, it booted using
the old loader, which I had doctored to eliminate the test (for serial console) that failed.
But the make installworld installed the defective loader again. I'll fix it
later; I can see a whole series of builds ahead of me,
And X? Fired up happily without a config file,
using the NV driver. But that didn't last long:
So then I got the VESA driver, which didn't like things either:
[ 36.329] (II) VESA(0): <default monitor>: Using hsync range of 31.50-48.00 kHz
[ 36.329] (II) VESA(0): <default monitor>: Using vrefresh range of 50.00-70.00 Hz
[ 36.329] (II) VESA(0): <default monitor>: Using maximum pixel clock of 65.00 MHz
[ 36.329] (WW) VESA(0): Unable to estimate virtual size
[ 36.335] (II) VESA(0): Not using built-in mode "1280x1024" (hsync out of range)
So I ended up with 1024x768. But at least it ran, and that without me
starting moused. OK, replace the config file and see what happens
without moused. It worked normally, with the same issues as before. What mouse
interface do I have?
=== grog@teevee (/dev/pts/10) ~ 1 -> ps aux | grep mouse root 409 0.0 0.0 11040 1328 - Is 18:08 0:01.61 /usr/sbin/moused -3 -p /dev/ums0 -t auto -I /var/run/moused.ums0.pid
root 426 0.0 0.0 11040 1288 - Is 18:08 0:00.06 /usr/sbin/moused -3 -p /dev/ums1 -t auto -I /var/run/moused.ums1.pid
Where did they come from? A bit of brute force searching through /etc showed:
So much for Callum's claims! I wonder if he has checked his own system. But at the very
least it seems that moused is still the only game in town, sadly.
Where am I now? From my questions the other day:
Could it be a random issue with X startup? Could restarting X fix
it?No, it's repeatable.
Could it have something to do with the keyboard remapping? Clearly something has
changed there, since I can no longer do it automatically on startup. I could check
before doing the (manual) remapping, next time.No, tested that.
Could it be a random kernel issue? Could rebooting fix it?No.
Could rebooting the old kernel fix it?
Alternatively, could building a new kernel fix it?
How do moused and X communicate? Clearly X is ignoring moused's carefully
faked keypresses. Could the communication be via the kernel? That would explain its
appearance now.
Could it be an old, worn-out method (moused) to access the
mouse?No, it seems that moused is still the only option.
Round midday I chanced to look at the PV dashboard on my desktop. PV Generation: 750 W!
And that with bright sunshine. What could cause that? It should be closer to 4 kW. From
time to time I've had uneasy feelings that some of the panels aren't working correctly. But
that was extreme. Is the inverter doing something silly again? Off to check.
A series of no less than four grid
power failures. The battery was fully charged (column SOC), so there was
nowhere to put the PV power, so it only delivered what the house needed
(column PacPV).
On with the coq au vin this
afternoon, taking copious photos.
Ingredients
For 6 servings
quantity
ingredient
step
3 kg
cock or chicken
1
750 ml
red wine
1
4
juniper berries
1
1
clove
1
1
sprig of thyme
1
2
bay leaves
1
500 ml
water
2
bouquet garni
2
200 g
smoked pork belly
4
20 g
lard
4
200 g
onions
6
15 g
garlic
6
25 g
salt
8
black pepper
8
4 cloves
garlic
8
300 g
carrots
11
12
pearl onions
11
300 ml
fond de veau
11
50 ml
cognac
11
2 squares
black chocolate
11
300 g
mushrooms
12
Preparation
Day 1
Cut the cock into serving pieces. The general recommendation is to make 10 pieces, and
that's what the photos show, but then they're huge, about 220 g a piece. For a cock of
this size it's better to cut the breast into 6 or even 8 pieces. Keep the back and any
other bones. Add the spices to the wine, put the cock pieces in and mix well. Put in
the refrigerator.
Take the back and other bones and cut into small pieces. Fry in a little butter until
coloured. Add water and bouquet garni and cook, reducing to about half the volume.
Cut the pork belly into small cubes (« lardons ») and fry in a pan until well browned
but not crispy. CET suggests adding butter if necessary. My belly was relatively lean,
and I added lard rather than butter.
Mix salt and pepper with the marinade. Cover the onions with the remaining cock and add
the marinade and the broth from part 2. Add some water to the pan and boil to loosen
the remaining deposits:
Add to the pot. Also crush the remaining garlic, unpeeled, and place them in the pot.
Bring to the boil and simmer gently for 1 hour. Then raise the heat slightly, maybe
removing the lid, to reduce the sauce. After another 45 minutes (a total of 1¾ hours)
remove the pieces of cock and place in the pan with some of the sauce. Place in an oven
at 180° with no fan for 45 minutes. This is intended to make the meat crispier. Here
the result at the end of the 45 minutes:
Well, that was the recipe I came up with after the event. It's close: I somewhat overcooked
the lardons (the recipe did say « bien dorée »). And as usual I had to guess the
quantity of salt (« sel » with no further information), and I chose 30 g, marginally too
much. And I spent 2 hours worrying about whether it would ruin the sauce, but fortunately
it didn't.
The real issue with the dish is, as always, the amount of liquid. I decided that the fond
de veau wasn't really necessary, so I omitted it. Next time I'll put it in, along with the
chocolate, which I hadn't noticed on the ingredient list. Shades
of Mexican cuisine.
I took a lot of photos yesterday, using the Olympus OM-D E-M1 Mark I and my
remote flash trigger. Most came out fine, but a couple looked like the shutter speed had
been set wrong, showing typical focal plane shutter vignetting. Here two images taken with
identical settings a couple of seconds apart:
What causes that? It looks as if the shutter speed had been set wrongly, but they were both
taken at 1/320 s, the documented flash sync speed. And apart from this example, there was
one where the flash didn't fire at all.
Today I tried a few more at higher speeds. Conventional wisdom, backed up by my own experiments, says that the higher the speed,
the shorter the band, like this photo taken with the Olympus E-510 years ago:
This was taken at 1/1000 s. The E-510 had a sync speed of 1/180 s, though it happily synced
(at least in this example) at 1/320 s. The E-M1 should do much better. But what I was was
very different. Here at 1/250 s, 1/320 s, 1/400 s, 1/640 s and 1/800 s:
Beyond 1/800 s there was no change. That's interesting for a number of reasons. Firstly,
in this example the synchronization failed even at the rated 1/320 s. And secondly there's
no obvious relationship between the shutter speed and the vignetting. And in passing, it
seems that the shutter travel on the E-M1 is in the opposite direction to the E-510.
What's causing that? Is the shutter failing? Is the remote trigger triggering at the wrong
time? Sent off a message to the Facebook M43 Tech
Talk group and got few and useless responses that seemed to have ignored the real
issue.
Finally got round to looking at Callum Gibson's X logs. They were interesting. One was close to my logs, and the other was very different.
One loads a driver:
[1564593.546] (II) XINPUT: Adding extended input device "kbdmux" (type: KEYBOARD, id 6)
[1564593.566] (II) config/devd: kbdmux is enabled, ignoring device atkbd0
[1564593.566] (II) config/devd: adding input device (null) (/dev/sysmouse)
[1564593.566] (II) LoadModule: "mouse"
[1564593.566] (II) Loading /usr/local/lib/xorg/modules/input/mouse_drv.so
And the other ignores /dev/sysmouse:
[ 18.027] (**) Option "Device" "/dev/input/event5"
[ 18.027] (**) Option "_source" "server/devd"
[ 18.028] (II) event5 - Logitech USB Optical Mouse, class 0/0, rev 1.10/21.10, addr 4: is tagged by udev as: Mouse
[ 18.028] (II) event5 - Logitech USB Optical Mouse, class 0/0, rev 1.10/21.10, addr 4: device is a pointer
[ 18.028] (II) event5 - Logitech USB Optical Mouse, class 0/0, rev 1.10/21.10, addr 4: device removed
[ 18.028] (**) Option "config_info" "devd:/dev/input/event5"
[ 18.028] (II) XINPUT: Adding extended input device "Logitech USB Optical Mouse, class 0/0, rev 1.10/21.10, addr 4" (type: MOUSE, id 11)
[ 18.028] (**) Option "AccelerationScheme" "none"
[ 18.028] (**) Logitech USB Optical Mouse, class 0/0, rev 1.10/21.10, addr 4: (accel) selected scheme none/0
[ 18.028] (**) Logitech USB Optical Mouse, class 0/0, rev 1.10/21.10, addr 4: (accel) acceleration factor: 2.000
[ 18.028] (**) Logitech USB Optical Mouse, class 0/0, rev 1.10/21.10, addr 4: (accel) acceleration threshold: 4
[ 18.029] (II) event5 - Logitech USB Optical Mouse, class 0/0, rev 1.10/21.10, addr 4: is tagged by udev as: Mouse
[ 18.029] (II) event5 - Logitech USB Optical Mouse, class 0/0, rev 1.10/21.10, addr 4: device is a pointer
Somehow there are too many variables here. Looking at my own logs, they seem to have
changed too. Is the interface via /dev/input/eventx different? How does all
this stuff fit together? In the process of searching, found a man page mousedrv(4) describing the driver. But the driver itself isn't on the system! Looking at my own
(new) Xorg.0.log, I found:
[ 34.648] (II) XINPUT: Adding extended input device "System mouse" (type: MOUSE, id 7)
[ 34.648] (**) Option "AccelerationScheme" "none"
[ 34.648] (**) System mouse: (accel) selected scheme none/0
[ 34.648] (**) System mouse: (accel) acceleration factor: 2.000
[ 34.648] (**) System mouse: (accel) acceleration threshold: 4
[ 34.649] (II) event1 - System mouse: is tagged by udev as: Mouse
[ 34.649] (II) event1 - System mouse: device is a pointer
OK, so maybe it shouldn't be using moused at all? Well, at least it isn't
using /dev/sysmouse. Could that be part of the problem? At least it could explain
why moused sees the button reassignments but X doesn't. But I know what happens if I
don't start moused: no mouse. What happens if I start X next time
without moused running? Still more things to think about. I wish I had an overview.
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.
