First day of autumn! And it shows. Yesterday we had a top temperature of 32.7°. But then
it started raining (12.0 mm), and today we had a maximum of only 22.5°, still better than
the forecast 19°.
12 mm rain has its own special meaning: more water (round 4 m³) in the water tanks. And the
output of the pump isn't connected. Over to the Stones Road property, and sure enough,
water was trickling out the output:
A couple of days ago I
discovered that there was a discrepancy between a photo file on my photo disk and on a
backup disk. Which was wrong? Today was time to make backups to the other disk, so clearly
it was time to investigate before overwriting the good version. A good thing I did, too:
most of the contents of the file on my primary disk was replaced by binary zeroes, exactly
the scenario that I suggested a couple of days ago.
6617089 is not a number that immediately jumps out and grabs you. But it's 2048 × 3231 + 1.
The 1 is a fencepost obfuscation on the part of cmp(1), so this is really the start
of a 2048 byte block, the default fragment size of FreeBSD's UFS.
Investigation showed that the zeroes continued for many sectors.
So now I have a valid version. But how many other failures have slipped through the cracks?
And why are files full of binary zeroes so common? Is it a FreeBSD issue, or (more likely)
something to do with the way disks work? In any case, time to write a program to search for
large consecutive binary zeroes.
Amusing, yes. But it does beg the question about the use of the punctuation at the right.
It's not until you run into weird bugs that you realize that they're on your side. One of
the issues that (not only) I still have
with Python
Yvonne off with Chris Bahlo
to Camperdown this afternoon
for a seminar. And once again they had issues with the car not running smoothly, though
this time Yvonne tells me that no “Powertrain” indication appeared. Who can repair this
without spending most of his time scratching his head?
Yvonne bought a new bread pan last week. Today was the first
time I used it. I started with the same all-rye starter that I made last week, the one that
over-reacted.
Today was different. First, I needed a number of calculations just to get the quantities
right, but the dough behaved completely differently from last time. Last time it rose so
quickly that the top of the loaf separated from the rest. This time it took over 6 hours to
rise. That doesn't seem to have anything to do with the pan, more with the temperature. I
need to do all this under controlled temperature, something I'll postpone until we move
house.
Call today from Kevin, the support manager of Aussie Broadband, addressing last weekend's outage. He wasn't able
to help; despite the claims on the web site, they really don't have any real support at
weekends, at least not for “residential” customers. Apparently they do for business
customers. Does that makes sense? A two day outage will annoy
any VoIP user, whether business or
residential. But Kevin promised to get somebody from the business team to contact me to
talk about pricing. That's good, because they don't have anything about National Broadband Network for business customers on
their web site. He also promised to bring it up at the next management meeting. Hopefully
they'll reconsider: apart from that, they're very good.
During quite a long discussion, Kevin's frustration with the NBN became evident. They don't
have any support at weekends either! And he's seen the same bureaucracy that I have:
“Remember Telstra 25 years ago? These people
are worse”. And it seems that this silly 30 minute power down is really required by the
NBN, and that they're inclined just to ignore a request if any ı has not been dotted or any
l not crossed. That's a pity: it greatly detracts from the idea of the network, and is fuel
for the advocates of the free-for-all that we had before.
Also discussed the O'Dea's problems with connection to the NBN. They need an antenna mast
to conenct, and that seems to be something that the NBN bureaucrats don't understand.
Aussie has done things like that before, but only with their own
(WiMAX) towers. Still, he's noted the
fact, and maybe Steve will get satisfaction from them.
I've bought a backdrop stand for studio work. Nothing fancy, just a couple of tripods, a
cross-bar and a muslin cloth to hang from the cross bar:
The quality is barely acceptable. One of the clips doesn't close properly, the whole thing
is very flimsy, and the cloth (folded) is very wrinkled.
And how does it work? I don't know yet. The thing is much larger than I expected, and I
can't find anywhere to erect it. If at least the cross bar were easily shortened, things
would be a lot better.
I'm selling my Tektronix 555
oscilloscope, but first, of course, I want to get better photos of it. Spent some
time playing around with my new lighting stand, and finally managed to make it narrow enough
to fit in the space I've chosen for the moment:
But only at first sight. The plug-in looked completely clean when I started, but there's
dust, cobwebs and some stain on it. These photos bring out all the minor details that you
just don't normally see. And, worse, there are shadows! That's not completely surprising
given the nature of the illumination, but it's certainly undesirable. Cleaned the plug-in
and put my mecablitz 58-AF-2 on top for bounce flash off the ceiling. This is the configuration
shown in the photo above.
And the results? Unexposed! DxO Optics “Pro” can
extract some detail out of the shadows, but it's unexposed by at least 8 EV. And that with
the same exposure and the same studio flashes as before. What went wrong? They definitely
fired, so at first I thought it might be a problem with the (wireless) synchronization. But
even after dropping the shutter speed to 1/60s, I couldn't get any image. Wikipedia is your friend:
Before the actual exposure one or more small flashes, called "preflashes", are
emitted. The light returning through the lens is measured and this value is used to
calculate the amount of light necessary for the actual exposure.
So the pre-flash set off the studio flashes, and the mecablitz was too weak to expose at
all. So what do I do? Clearly under these circumstances it would be possible to run the
flash in manual mode at full output, but what difference would it make? I still need to
find a better approach.
BigPond email: We don't need no steenking security
Sent a mail message to Gary Murray today. It didn't go through:
<murraydrill@bigpond.com>: host extmail.bigpond.com[61.9.168.122] said: 552
5.2.0 yrRW1p01Q1sUVRc01rRYpC Suspected spam message rejected. IB704 (in
reply to end of DATA command)
I've seen this before.
BigPond is too stupid to distinguish digital
signatures from spam. So they reject messages on the mere suspicion of spam. Is this in
their users' interests? A good reason for any BigPond user to choose a competent mail
service provider. I'm still amazed how incompetent everything to do with Telstra is.
We've been planning to eat more beans for some time now. Too long, in fact. Yesterday I
cooked 300 g of white beans for use in soup (and for the record they weighed 840 g after
cooking). But their “best before” date was a year ago, and it shows: they're rough and
crumbly. So now I'm looking for something else to do with them, possibly patties like
masala vada.
In the meantime, though: is it worth cooking dried beans? They're not even overly cheap,
about $5 per kg, though of course that goes a long way. But if you buy pre-cooked beans in
cans, they're going to be cooked correctly, and though they're more expensive (how much?),
the overall increase in food prices is barely noticeable.
Spent a considerably amount of time today trying to get better photos of the Tektronix 555 oscilloscope. I failed. Clearly
bounce flash with the mecablitz 58-AF-2 was too weak to make any difference. But how about a mirror below
the unit to reflect flash up? That required lifting the plug-ins a little, so got a thick
paperback and put them on that. After all, I can crop later, right?
Wrong. I suppose I could white out the area with some other software, but the plug-ins
overhang, and the title of the book reflected in the bars:
If anything, today's results looked worse. What next? Ring flash? Normal daylight? The
light tent that I bought years ago and hardly used? Why is this so difficult?
We've sold our shipping container to Chris Bahlo, and we've been planning to empty it for
coming on 9 months now. The main issue was where to put the contents, and for that we
needed a shed.
Well, we have a shed now. My excuses are running out. So today we set to to move the
contents. Much of the content is old magazines, some of which have made their third move
without being read again:
What's left is at the back. On the right is nearly all Chris' stuff, and the cartons in the
front are empty. It's surprising how little space the cartons take up in the new shed:
They should be ready in the middle of next week, and then they'll be back at the end—he
estimates six weeks.
Six weeks! That would put us in the new house at the end of April at the very
earliest. We were hoping early April at the latest. Duncan has been very cagey about
telling us any dates, and of course Mark doesn't have to be right. But he has no reason to
distort things, either. So maybe we should be looking at early May. This thing is taking
forever!
The real issue taking photos of
the Tektronix CA plug-in is the length
of the knobs, up to 25 mm for the ones with the red centre. It's clear that any lighting
from the side will cast a shadow.
So: completely equal lighting from all sides would do it. But how do you do that? I still
have the options of the light tent, which previously has not been overly successful, or the
ring flash. In principle the ring flash would be the solution, but the pane
reflects. Tried it anyway today and got (with a previous attempt on the left):
As expected, the shadows are gone, and there's a reflection, though not as bad as I had
feared. It's interesting how much better the adjusting screws for gain adjustment and DC
balance are. But in comparison, I think it still looks worse than previously.
Where to now? Combine ring flash with something else? I suppose it's time to go and grab
the light tent.
More playing around with the photos of the Tektronix 555 oscilloscope today. First, a combination of ring flash and studio
flash. The studio flashes were set at full power, and I modified the ring flash power from
¼ to 1/16:
The first is a little burnt out in places; the second shows shadows again. It's similar
with the front panel. I can't get rid of the shadows without burning out.
There's another issue as well: perspective. I'm looking at the middle of the plug-in, so I'm
seeing the lower knobs from above and the upper knobs from below. It's enough to be
irritating. This was with a 50 mm lens (24.4° diagonal), which isn't that short. But for
the fun of it I tried it with a long telephoto (effectively 281 mm, or 4.4° diagonal). That
required ambient lighting and an exposure of 5 seconds. Here the attempts from 2 days ago,
yesterday and today:
Is that better? I think so. It's not as brightly coloured as the first flash attempt, but
that may not be wrong. Ended up taking a whole lot more images like that. Now to clean the
dust out of the mainframe and take that too, and I'll call it a week.
I somehow seem to have missed the announcement of the Leica T.
Once again Leica have astounded me. A horrendously expensive camera with nothing to
recommend it, neither image stabilization, speed, ease of use nor choice of lenses. Here a
comparison with the Olympus OM-D E-M1 and Yvonne's Olympus OM-D E-PM2. Some of
these are from Digital
Photography Review, others from the March edition of c't Digitale Fotografie:
Only 7 years later, Leica appear to have forgotten about this model. The link I had
(http://en.leica-camera.com/Photography/Leica-T/Leica-T-Camera-System) no longer works,
and neither does the search function at https://www.en.leica-camera.com/. The Wikipedia link also no longer works.
Leica
E-PM2
E-M1
Body Price (B&H)
$1,850
$270
$1,399
Lens
18-56 f/3.5-5.6
14-42 f/3.5-5.6
12-40 f/2.8
Price
$1,750
bundled
$999
Lenses available
4
139
139
Sensitivity (ISO linear)
125-12,500
200-25,600
200-25,600
Max shutter
1/4000
1/4000
1/8000
Flash sync speed
1/180
1/250
1/320
AF speed (1000 Lux)
0.7s
?
0.26s
Images/s
5
8
10
Yvonne's camera meets or exceeds all the Leica specs, but cost $270 instead of Leica's
$3,600. I don't have a comparable AF speed for the E-PM2, but I'm sure it's considerably
less than 0.7 s. And the lens? Hard to find useful test results, but c't Digitale
Fotografie states that the Vario-Elmar is considerably worse than the 12-40
Zuiko Pro.
There are many other specs, of course, but these are the ones that stand out, and I can't be
bothered to follow the others up. Can the Leica do HDR bracketing? Who
knows? But the E-PM2 can. The tests suggest that the Leica is not the easiest camera to
use.
And the number of lenses available? The 4 for the Leica is pretty clear. Why can't it use
their existing lenses? The 139 for Olympus is conservative; it's the number
of Four Thirds system lenses
available from B&H.
All in all, typical Leica. How do they manage to sell cameras like this at any price, let
alone at the inflated price of this camera?
So: why do I even bother to mention it? They have done one thing right: they haven't broken
the sensitivity scale.
According to ISO 12232:2006, the light sensitivity of a sensor can be measured either logarithmically
(effectively decibels, according to the
previous DIN 4512) or
linearly (according to the older ASA standard). In each case, the steps are decibel steps, representing the tenth
root of 10 (10↑0.1) or a linear factor of approximately 1.258. Here's a comparison, showing
how the ASA values were approximated:
log (°)
Linear
approximation
0
1.000
1.00
1
1.258
1.25
2
1.584
1.60
3
1.995
2.00
4
2.511
2.50
5
3.162
3.20
6
3.981
4.00
7
5.011
5.00
8
6.309
6.40
9
7.943
8.00
10
10.000
10.00
For each 10° increase, multiply the linear version by 10. 20° ISO is 100 linear. 38° is
6,400. 41° is 12,500. And so on.
But that's not what the industry has been doing so far. Most manufacturers have been misled
by the close relationship between 3° and linear 2. Starting (I think) with 40°/10,000,
they're deviating from this scale. Instead they're going by powers of 2 and doubling, so
now they're writing 12,800 instead of 12,500 for 41°. We've even got as far as 58° (400,000
linear), which they're selling as “409,600”.
This is silly. In particular, it tears the linear scale apart. Add 50 to the sensitivities
and you get:
log (°)
Linear
Clever modern
approximation
version
50
100,000
102,400
51
125,000
128,000
52
160,000
160,000
53
200,000
204,800
54
250,000
256,000
55
320,000
320,000
56
400,000
409,600
57
500,000
512,000
58
640,000
640,000
59
800,000
819,200
60
1,000,000
1,000,000
The problem here is that the relationships no longer hold true. Some of the values have
diverged, others haven't. I'm guessing at some of these values, of course, since nobody has
actually shown any except 50°/53°/56°. But clearly it's time to recognize that the scale
has nothing to do with powers of 2, and return to the powers of 10. And that's what Leica
appear to have done. I wonder whether they'll be successful.
In passing, of course, it would make sense to finally depart from the linear scale
altogether: it's becoming more and more unwieldy, while the logarithmic scale still looks
sane. But I don't see that happening any time soon.
It seems that they're normal, but in this case they're excessive. Mark confirms that it's a
pain. This is what some of the not-yet-retouched areas look like:
A couple of days ago we
moved out half our content of the shipping container, and today was the day for the rest.
We managed half. There are still a few cartons in the back, which I'll need to sift through
before taking to Stones Road: I suspect that at least half will stay here pending a garage
sale, and most of that will subsequently be discarded.
It's interesting to note the effort required. Theoretically, picking up heavy boxes and
putting them on a trailer should be about as difficult as taking them off the trailer and
putting them in the shed. In practice, the latter is much easier. It must be the pain of
accepting that I have to throw the stuff out.
On IRC today, Jürgen Lock pointed us at this
comparison of network costs around the world. He's in Germany, but the take-home
message, at least for me, was:
Telstra .... charges some of the highest transit pricing in the world — 20x the benchmark
($200/Mbps).
So why hasn't the National Broadband Network changed that? I had thought that the geography was part of the problem, but the same
article also says:
Given that Australia is one large land mass with relatively concentrated population
centers, it's difficult to justify the pricing based on anything other than Telstra's
market power.
My main reason for disliking Telstra was
their extremely poor support. Maybe the pricing reflects the cost of their inefficiency.
Yvonne and Chris have finally got me to agree to go riding
again. This morning, without fail. So we had breakfast, I did my morning admin stuff, and
they went off looking for pallets to put under the hay in the hayshed.
A little later, a call from Yvonne. Come along and take a look. A box that she could
convert into a tack room:
This one is smaller than what Yvonne needs, but Doug (Braddy) tells her that there are many
more, and he is looking for one 2.2×3.9 metres in size, which would work well. Price? $100,
and $25 for delivery. We'd still need to put a front wall in, but you can't beat that
price.
Doug also runs
the DereelMen's shed, something I've never understood.
Currently it's all in one of his sheds:
Doug did a good sales pitch, and so got new members—Yvonne and Chris! I'll hold off until
I've taken a look.
Also bought a number of planks that had been reworked as raised garden beds. Again, at $10
the prices are really good. Now to work out what to do with them.
Taking the photos of the Tektronix 555
oscilloscope has been surprisingly painful. But enough is enough. Tried again today
with flash and the long telephoto lens, this time concentrating on the mainframe. Moving
the lens further away also allowed me to put the flash units closer together, with good
results. Here my five tries of the CA plug-in:
The first was just with studio flash. The second was only with ring flash, the third with a
combination of ring flash and studio flash, the fourth with ambient light, and the fifth,
like the first, with studio flash only. Yes, the shadows are still there, but they're much
less obtrusive.
Took photos of the entire scope, plug-in by plug-in, and the mainframe with and without
plug-ins.
From time to time while walking the dogs, they go off into the bush. The densest kind here
is Acacia paradoxa, also known as
Kangaroo Thorn. And it breaks off, so we continually have to remove prickly stuff like
this:
It's been nearly 18 months since we had to put down my mare Darah, and since
then I haven't been on a horse. In fact, since Darah was sick, it seems that it has been
nearly 3 years since the last
time I noted in this diary that I had been riding.
I bought a new riding helmet months ago, but I got the feeling that Yvonne and
Chris didn't want me to come with them—it wasn't for the lack of a horse. Then they wanted
to go riding yesterday, but spent the time looking at shipping crates instead. Still,
finally we got round to it today. Chris gave me Smokey,
an American Saddlebred
stallion that we're looking after (and who is for sale!).
Riding a stallion is always interesting because he had to be kept away from mares. So of
course Yvonne and Chris both rode mares (Carlotta and La Tigre respectively). Things went
well enough apart from an encounter with some small boys on bicycles at close quarters,
which worried both Smokey and Carlotta. But somehow it's a completely different feeling
from Darah, perhaps in part because of the saddle. Chris was encouraging: “You'll never
find another horse like Darah”.
What went wrong there? 12 different copies of one image, all mutilated. It's a combination
of these silly “Art Filters” and the ease with which you can change the settings on Yvonne's
Olympus E-PM2 (or,
for that matter, my Olympus OM-D E-M1) while riding on a horse. All you need to do it pick it up
wrongly, and you've set all sorts of nonsense. All the fault of touch screens.
The only good thing about the touch screen is that you can deactivate it. On both cameras,
setup menu J, “Touch Screen Settings” allows you to turn the bloody thing off. Maybe Yvonne
will like the camera better now.
Over to the Stones Road property in the afternoon to let the dogs run. Now that the fence
wire and the horses have been removed, there's no reason not to let them run around. Well,
not much. Zhivago and Leonid immediately headed for the dam and got themselves
soaking wet. But not my Nikolai—he stayed
out.
Well, for a while. I think he must have headed for the water tanks instead, soaked himself
and then rolled himself in the sand:
What a mess! Of course we left them in the dog run until they had dried out. It wasn't
until some time after 20:00 that Yvonne let them in again.
And it was a good half hour later before we noticed that they hadn't come into the lounge
room. Out to the kitchen: door open again, dogs run away.
But they weren't far, and came in almost immediately, panting heavily. Leonid soon
recovered, but Nikolai yelped a bit and carried on panting. Felt him: another yelp or two,
but not related to where I touched him. In the end called up Pene Kirk, and she told me
how to take his pulse (on the inside of the thigh, from the front). My best guess was 150,
and that after 10 minutes!
Over to Pene's, and she checked him out. After 30 minutes he was still panting, though not
as much. I had been worried about heart issues or snake bite. His heart was OK, and as
Pene put it, if it had been a snake, he wouldn't be standing any more. Her concern was that
it could be bloat,
apparently Gastric
dilatation volvulus. And that's serious, and it's known to affect dogs
like Borzois
and greyhounds. But she couldn't hear
any typical symptoms, so she gave him an injection just in case, and we took him home again.
He carried on breathing irregularly for a couple of hours, but seemed to be on the mend.
I'm going to have to read up on bloat.
On the way home picked up an unexpected treat for the dogs: a rabbit had run out in front of
the car and killed itself.
First attempt at bean patties this evening, roughly based on masala vada. Not an unmitigated success: I had cooked
the beans soft, and they were easy to crush, but they needed something to hold the patties
together. Also too much salt. On the other hand, it shows promise, something to do with
beans that are too old to eat whole.
Yvonne dragged me out of bed this morning to tell me that we
were off the net. It took me a while to understand, but in to the office, and sure enough,
we had been off the net for hours. The usual thing: DHCPDISCOVER going out, no
reply. Called up Aussie Broadband support, spoke to Kylie, who relatively quickly connected me to Jerom, who is
(ahem!) 3rd level support.
He confirmed that they were receiving the DHCPDISCOVER and replying correctly with
a DHCPOFFER. But that reply never made it back here. Shades of last month? In any case, it's
not acceptable. Got him to transfer me to Kevin, the manager, who told me that it was part
of a more general outage, that the problem was within the National Broadband Network, with whom a ticket had been
raised, and that hopefully things would soon come back to normal.
They weren't. The outage lasted all day. And it seemed to be limited to customers of
Aussie Broadband (more specifically, the address block 180.150.4.0/24).
That gave me an opportunity to revive my network status page. I had been through three iterations: firstly ADSL,
then satellite, and finally HSPA. And when the NBN arrived, I thought I could stop. It seems that I was wrong.
So I frobbed the old scripts, and to my surprise discovered the relative reliability:
Medium
Number of
Time between
Outage time
Average outage
Maximum outage
Availability
outages
outages
(seconds)
(seconds)
Satellite
1684
15:40:49
9 days, 04:00:01
470
118282
99.17%
NBN
79
4 days, 23:19:02
5 days, 16:33:03
6222
192150
98.55%
Less reliable than satellite! And look at those average outage times! Satellite outages
were mainly short, whereas the average NBN outage is 103 minutes, and the longest
(admittedly not the fault of the NBN) was well over 2 days.
How can this happen in a modern network? Bureaucracy, it seems. Everybody I have spoken to
confirms that they're really hard to deal with, and being a monopoly, they don't need to
justify themselves. Not quite what we were hoping for, and possibly an argument for the
current government's deemphasis of the network.
How many people were affected? Hard to say. CJ called me in the evening to tell me that
his phone (VoIP via Aussie Broadband) wasn't
working. How many other people didn't notice until then, or at all?
While I was grumbling about the network outage, Yvonne, who
had been riding with Helen Miller, came in with the news that Helen had fallen off her horse
and had probably broken her wrist. That required somebody to drive her and her horse back
to Enfield (in fact, just round
the corner from the property in Inglewood Drive that we nearly bought two years ago). So off back with her,
and gave her instructions to go to
the Ballarat Base Hospital
and have it looked at.
She called back in the evening: it looks like she'll need surgery, but first
a cat scan tomorrow.
And I thought Friday the 13th was supposed to be the unlucky day.
I spent some time in Kuching nearly 50
years ago. Looking at the map nowadays shows me how much it has changed. Jalan Kong Ping didn't exist at the time; probably the whole area was virgin jungle.
How times change (not to mention the length of the road, which on the map looks to be only
about 1 km long).
Jesse, the tiler, has just bought himself a beagle puppy, and he was really impressed by our
dogs. Yvonne gave him lots of tips, and he played around
getting them to sit on command:
Time to change my ISP, I'm afraid. Whom should I choose? Internode has the best reputation. They offer a similar product as Aussie Broadband: 300 GB per month for $75
compared to 250 GB per month for $60. The price difference is real: I never use 250 GB, so
the additional 50 GB from Internode are of no interest.
Called them up and got the usual message, that I had a wait of between 5 and 9 minutes
before I could speak to anybody. Accepted the offer of a call back when somebody was
available. And the call back came almost immediately, followed by a 6 minute wait before I
was connected to Alex. He wasn't prepared for my questions about reliability, of course,
but gave me the impression that they did have real techies available over the weekend. Went
through the usual stuff, and was in the process of signing up when he told me that I would
have to reconfigure my “router” with a new user name and password.
At first sight that's obvious. But what user name and password? You don't need that for a
DHCP connection to the National Broadband
Network. Are they using PPPoE on the
NBN? He went and checked. Yes.
He couldn't tell me why, of course. But he also didn't understand my objection. How much
overhead does PPPoE create? It depends on the packet size, but for small packets it can be
as much as 10%. Why should I choose that? Why did Internode choose that? So, at least for
the time being, I backed out of the contract.
It wasn't until some time later that it occurred to me that PPPoE might also have an
advantage: no DHCP problems. Apart from Aussie Broadband, I have also had short-lived NBN
contracts with Exetel and SkyMesh. Exetel uses PPPoE, and SkyMesh uses DHCP.
Exetel had no problems with connection setup or maintenance, while it took SkyMesh
2 days to configure their
DHCP server correctly. And the current problems with Aussie also appear to be related to
DHCP, as was the longest ever outage last month. But by then it was evening.
But not everything is going exactly according to plan. I had already noticed the rough join
between tiles and the skirting board. Duncan tells us that this is normal. I can't recall
ever seeing it before, and next time we're in town, we'll have a look at their display home
to see what that looks like.
There are also a few less-than-perfectly laid tiles:
Shouldn't that row have been between the top two rows? What does it say in the contract?
Nothing. But the plans from the contract have been updated to show the location:
As if that wasn't enough, there's a 3 cm gap below each door. That seems excessive. When
the flooring goes down, it'll drop to 2 cm or maybe a little over, but that doesn't seem
right. Duncan, of course, says “that's the way we always do it”. Another case for the
display home. Back at our home, took a look: yes, there's a 2 cm gap where floors are
carpeted, because of the rail underneath. But where there are wood floors, the gap is about
5 mm, as I would have expected. Until proof of the contrary I'm assuming that they forgot
to adapt the door size to the flooring.
On the positive side, Duncan has finally come out with a likely date for us to move in: four
to five weeks. Mark had suggested more, but as Duncan said, he doesn't know the details.
Time for us to pull our socks up.
Don't the roof colours of house and shed match well? In fact, the shed has a plain
galvanized iron roof, the same colour as the walls. The colour is just a reflection of the
sky.
Network connectivity came back this morning at almost exactly midnight. I had traced the
network since about 8:30 yesterday, but of course by the time I stopped it, we had about
200,000 packets and a 235 MB trace file. All that interested me was the time up to the
restoration of service. How do I do that? With Edwin Groothuis' help discovered the
wireshark “export” function. You can
specify a packet range, in my case 1-8212. And sure enough, it saved a file with just those
packets.
Tried to read it back in again.
The file "/home/grog/public_html/Day/20150313/offnet.trace" isn't a capture file in a format
Wireshark understands.
Huh? It's a plain text file!
No. Time Source Source Port Source Destination Destination Protocol Length Info
1 08:40:46.745833 0.0.0.0 00:50:da:cf:07:35 255.255.255.255 ff:ff:ff:ff:ff:ff DHCP 342 DHCP Discover - Transaction ID 0x76a73872
Frame 1: 342 bytes on wire (2736 bits), 342 bytes captured (2736 bits)
Ethernet II, Src: 180.150.4.128 (00:50:da:cf:07:35), Dst: Broadcast (ff:ff:ff:ff:ff:ff)
Internet Protocol Version 4, Src: 0.0.0.0 (0.0.0.0), Dst: 255.255.255.255 (255.255.255.255)
User Datagram Protocol, Src Port: bootpc (68), Dst Port: bootps (67)
Bootstrap Protocol
How do you save it as a capture file? At least with the old version I'm using (1.6.8), you
don't, though it appears that newer versions might.
More head-scratching. Then Peter Jeremy suggested using tcpdump:
And sure enough, that worked. What did I see? I had started dhclient, which zeroes
out the IP address of the interface, but then reinstated it, so I had
both DHCPDISCOVER
and ARP requests
going down the line. What was coming back? Nothing, you'd expect. But that's not quite
what I saw. My local interface is aussie-gw.lemis.com, and I've reverse mapped the
zone for other end so that I can call it radiation-tower.aussiebb.net:
What's all that? Microsoft stuff, of course, but how did it find its way to me? It
continued at irregular intervals all day long. Most of the IP addresses don't resolve (it's
Microsoft space, after all), but the one that does suggested that it's in Europe, and the
others seem to be scattered round the world. The last recognizable PTRs returned
by traceroute are:
=== grog@eureka (/dev/pts/8) ~/public_html/Day/20150313 19 -> traceroute 167.114.89.109 21 vac3-3-n7.qc.ca (198.27.73.237) 283.340 ms 284.869 ms 277.921 ms
=== grog@eureka (/dev/pts/8) ~/public_html/Day/20150313 20 -> traceroute 14.148.25.51 11 chinatelecom-hk-gz-peer.hkth01.pr.telstraglobal.net (134.159.158.246) 263.846 ms 264.781 ms 259.981 ms
I assume that means Québec and somewhere
in China respectively.
That's interesting, because I thought that I needed DHCP to create a routing a path to the
other end of the National Broadband Network cloud. But it took my dhclient a little longer before it reacted:
00:01:22.433342 IP 0.0.0.0.bootpc > 255.255.255.255.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
00:01:22.540081 IP radiation-tower.aussiebb.net.bootps > aussie-gw.lemis.com.bootpc: BOOTP/DHCP, Reply, length 300
In the meantime, things weren't silent. The NBNS requests exploded, now coming from
port http:
The first of these also appears to be in China, and the second in the USA. But why am I
getting these requests from source port http? And why only since connectivity came
back? They're not really targeting me (and I'm not replying), so presumably they have been
coming all the time, and until the net recovered they weren't getting through.
In principle I was replying, but since I didn't have a MAC address, I couldn't.
But that's what firewalls are for.
Whatever it was, the raises more questions than it solves. I'll do some more thinking.
One thing that hasn't changed is the poor response to DHCP requests:
20:34:15.264515 IP radiation-tower.aussiebb.net.bootps > aussie-gw.lemis.com.bootpc: BOOTP/DHCP, Reply, length 300
21:04:15.040496 IP aussie-gw.lemis.com.bootpc > radiation-tower.aussiebb.net.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
(46 retries omitted) 21:21:56.966434 IP aussie-gw.lemis.com.bootpc > radiation-tower.aussiebb.net.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
21:21:57.030945 IP radiation-tower.aussiebb.net.bootps > aussie-gw.lemis.com.bootpc: BOOTP/DHCP, Reply, length 300
Given that this seems related to the problem, it's high time to find out what's going wrong.
More work on my network status page today, without
making it really pretty. One thing of interest is the “TCP speed” plot, in blue:
This shows the reciprocal of the time it takes to load a small document from the other end
of the world. It's surprisingly constant. But for some reason the value increased round 10
March. Looking at the raw log data shows:
1425966865 0.70 # Tue 10 Mar 2015 16:54:26 EST
1425966926 0.71 # Tue 10 Mar 2015 16:55:27 EST
1425966987 0.72 # Tue 10 Mar 2015 16:56:28 EST
1425967049 0.71 # Tue 10 Mar 2015 16:57:29 EST
1425967110 0.55 # Tue 10 Mar 2015 16:58:30 EST
1425967171 0.57 # Tue 10 Mar 2015 16:59:32 EST
1425967232 0.55 # Tue 10 Mar 2015 17:00:33 EST
1425967293 0.56 # Tue 10 Mar 2015 17:01:34 EST
The second column is the time in seconds for the transaction; there really was a step change
there.
Also kept an eye on the DHCP traffic. Yes, we're still running into timeouts:
16:51:39.101013 IP radiation-tower.aussiebb.net.bootps > aussie-gw.lemis.com.bootpc: BOOTP/DHCP, Reply, length 300
17:20:52.104443 IP aussie-gw.lemis.com.bootpc > radiation-tower.aussiebb.net.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
(45 requests omitted) 17:42:03.150427 IP aussie-gw.lemis.com.bootpc > radiation-tower.aussiebb.net.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
17:42:59.151418 IP aussie-gw.lemis.com.bootpc > 255.255.255.255.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
(9 requests omitted) 17:49:43.624422 IP aussie-gw.lemis.com.bootpc > 255.255.255.255.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
17:50:07.634879 IP 0.0.0.0.bootpc > 255.255.255.255.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
17:50:07.696161 IP radiation-tower.aussiebb.net.bootps > aussie-gw.lemis.com.bootpc: BOOTP/DHCP, Reply, length 300
17:50:07.704029 IP radiation-tower.aussiebb.net.bootps > aussie-gw.lemis.com.bootpc: BOOTP/DHCP, Reply, length 300
17:50:09.532473 IP 0.0.0.0.bootpc > 255.255.255.255.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
17:50:09.576122 IP radiation-tower.aussiebb.net.bootps > aussie-gw.lemis.com.bootpc: BOOTP/DHCP, Reply, length 300
At 17:42:59.151418, after 47 requests over 22 minutes, dhclient gave up on the DHCP
server with which it had been registered and sent a total of 11 requests out on the
broadcast address. Like the before, none was answered. Seven minutes later, at
17:50:07.634879, it gave up with requests, dropped its IP address, and sent out
a DHCPDISCOVER request. As in every case I've seen, the response was
immediate: two responses, in fact, after 61 ms and 69 ms. This looks like exactly
the same scenario I discussed last
month. Given the experience of the last week, though, it's clearly imperative to fix
the problem.
On the down side, compared to last year
the Bromeliads
and Gasteria aren't flowering. Nor is
the Xanthorrhoea
quadrangulata, which looks as if it may be a once-a-lifetime bloomer. Here it was last
year:
Somehow I'm not keeping up with the work I need to do for moving in to the new house. The
news that it could be little more
than a month should spur me on, and today I spent a fair amount of time planning for
things. Jim Lannen can do the final fit for the electrics on Friday, but will I have things
ready for him? Where are the lights? Where is the UPS? I can't even decide which UPS to
buy, nor how long it should run at full load. At least I now have a list of what lamps we
need:
Earthworks invoice from Warrick Pitcher today. Needs to be scanned and sent to CVI. All went well—but where was the image? Tried again,
and again it didn't appear.
This is Epson software under Microsoft. Had it
decided to store the image somewhere else? Spent some time looking for the configuration
menu which specifies where the document should go. Where is it?
Only three choices of location: My Documents, My Pictures, or
anything else. And it was set for My Pictures. Where's that? After some
searching it seems that the correct pathname in Microsoft is c:\Users\grog\Pictures.
No My to be seen (and, surprisingly, no spaces in the path name). But they should be
going to eureka:/Photos/scans, mounted via Samba, and I had some recollection that I had created something like
a symlink to store it. But where was it?
Tried tree-climbing (entering pathnames would be too simple), but discovered that all my
Samba file systems were inaccessible. Why?
My fault, of course. The other day I discovered that half the world was accessing my Samba
server, so I added a firewall rule, which blocked a surprising number of requests:
=== root@eureka (/dev/pts/13) /home/w3/usr/local/etc/namedb 106 -> ipfw show 30 00030 41931 3281850 deny ip from not 192.19.197.0/24 to any dst-port 137
Looking more carefully, though, I had mistyped: 192.19.197.0/24 instead
of 192.109.197.0/24. I thus effectively blocked just about everything. OK,
that can be fixed. But what's wrong with this picture?
The scan program issued no warning or error message whatsoever, although it was unable
to save the image where I had instructed it.
Something—Epson or Microsoft? Who can tell?—appears to have broken the symlink.
My attempt to reconnect the file systems failed without any explanation.
Into town today for a number of house-related things. First to Beaumont tiles to get
copies of the documents we had signed, not helped by the density of the personnel. But as
suspected, yes, we had signed off on the arrangement of the tiles, but not on the
orientation of the decorative frieze. And they confirmed that there was no obvious way. So
JG King should have asked us. Looks
like a case for negotiation.
Then to Bunnings to finally get some lamps
for the house. It's still like pulling teeth. Moved on to Masters, where we found all except for the lamps over
the bathroom mirrors and two of the outside lamps, which are complicated by the way Jim has
done the wiring through the bricks. I'll have to talk to him about what we can put there.
Then along to look for the JG King display home in Lucas. Sign of the times: it
had been sold. Signs pointed further west, so followed them, and sure enough found a couple
of new display homes. We wanted to compare door heights and the transition between skirting
boards and tiling. The transitions in both display houses looked normal, but the gaps under
the doors were still too big:
Still, that's 14 mm, not 30. How big will the gap be when the floor is laid?
The receptionist in the garage recognized us, and thought we had been there before.
Yvonne recognized him too: Bernie Massey, who had only been
with JG King for 3 weeks. But two
years ago (almost to the day) he was working for Harcourts, and had sold us the wrong plot of land in Enfield. Hopefully he'll
make a better fist of it as a house salesman. While discussing the incompetence of JG
King's admin staff, I mentioned that they sent me the wrong plans. It was on the tip of my
tongue to say “so you should fit in well here”.
Back home past Stones Road. Still no sign of any activity this week. And it was necessary:
today was excessively windy, and the outside door that had been wedged in place had pushed
its wedge out, and had been banging in the wind:
The black toner for our Brother HL-3170CDW laser printer is running low. I've had it for 8 months, so it was to be expected.
Took a look on eBay and found the toner cartridge (TN-251) for $25.30, including free
postage. While in town, dropped in at Officeworks. Yes, they had it in stock—for $129!
That's ridiculous. OK, the cartridges on eBay are almost certainly replacements, but they
have the chips, and there can't be that much difference in quality. The printer cost
$249, so this single cartridge costs more than half the purchase price. A complete set of 4
would cost more than double the purchase price. Nobody can tell me that these prices aren't
inflated. And what good does it do for Brother?
Up early this morning to go over to Chris Bahlo's property with her to meet David Grigg of
BREAZE about solar electricity. He didn't
show. While waiting, noted that Simonds don't put tubes on their downpipes:
I suppose that the brickwork can take it, at least in the short term.
In the afternoon Doug Braddy was supposed to bring her one of the shipping crates we were
talking about last week. He
didn't show: his truck had broken down. And for those two appointments she had taken the
day off work.
So finally today was the day to go shopping
in Melbourne. Up early for a start at
8:00, loaded the oscilloscope into the
car, while Yvonne went off to walk the dogs. She didn't get
far: somebody's Clydesdale had
got out and was annoying Smokey. Woke up Chris to deal with the issue, but the Clydesdale
had moved on.
Discussing the matter at breakfast, it seems that the horse belongs to the Yeardleys'
neighbours. They seem to suit each other.
Somehow as a result of this, Yvonne decided not to go to
Melbourne. I'm not happy.
I've been finding that SpamAssassin has flagged more and more legitimate mail as spam lately. The reason is always the same:
2.7 DNS_FROM_AHBL_RHSBL RBL: Envelope sender listed in dnsbl.ahbl.org
Lately, though, I've noticed that even well-known senders are getting flagged like this.
That's serious because I have set my maximum score to 3, so just about anything else will
cause it to be classified as spam. It really hit home when my system classified local mail
as spam. Clearly time to weaken the score. Went looking and found,
in /usr/local/share/spamassassin/50_scores.cf:
score DNS_FROM_AHBL_RHSBL 0 0.306 0 0.231
Huh? That's also where I set the required score. What's going on? Read the man page,
which confirmed the location. Looked for alternate 50_scores.cf files, but didn't
find anything appropriate. Time for ktrace. spamd is
a Perl script, and what I found was:
13255 perl5.14.2 NAMI "/usr/local/bin/spamd"
13255 perl5.14.2 RET open 3
13255 perl5.14.2 CALL read(0x3,0x8018a7000,0x2000) 13255 perl5.14.2 GIO fd 3 read 4096 bytes
#!/usr/local/bin/perl -T -w
...
my $PREFIX = '/usr/local'; # substituted at 'make' time
my $DEF_RULES_DIR = '/usr/local/share/spamassassin'; # substituted at 'make' time
my $LOCAL_RULES_DIR = '/usr/local/etc/mail/spamassassin'; # substituted at 'make' time
my $LOCAL_STATE_DIR = '/var/db/spamassassin'; # substituted at 'make' time
use lib '/usr/local/lib/perl5/site_perl/5.14.2'; # substituted at 'make' time
What's the difference between DEF_RULES_DIR and LOCAL_RULES_DIR? Have
they changed things so that the other .cf files are only read from the latter? And
why doesn't the man page even mention the second directory? And there I found no mention
of DNS_FROM_AHBL_RHSBL at all. It seems to have been built in at install time,
though nothing in the trace gave any hint. So I put the appropriate line
in /usr/local/etc/mail/spamassassin/local.cf, and sure enough, it worked. But why do
I need to run ktrace to find this? And why is the man page wrong? This is an old
version of SpamAssassin, so it's not worth following up, but it's clear that there's yet
another reason to accelerate the system upgrade that has been waiting for over a year now.
But why is everything being classified as spam anyway? Went looking at dnsbl.ahbl.org. Nothing there! http://ahbl.org/ says little, and leads to http://ahbl.org/node, which also says nothing obvious until you read down a couple of (undated) items:
I'm winding down the public DNSbl services of the AHBL.
This means the dnsbl.ahbl.org, ircbl.ahbl.org, rhsbl.ahbl.org lists are all going away, as
is the public lookup/removal tool.
So it's just not there any more, so SpamAssassin fails hard by flagging all mail as
failing the test. What a pain!
Date: Wed, 18 Mar 2015 18:29:15 -0700 (PDT)
From: Google Webmaster Tools Team <wmx-noreply@google.com>
Subject: Fix mobile usability issues found on http://www.lemis.com/
Google systems have tested 3,000 pages from your site and found that 100%
of them have critical mobile usability errors. The errors on these 3,000
pages severely affect how mobile users are able to experience your website.
These pages will not be seen as mobile-friendly by Google Search, and will
therefore be displayed and ranked appropriately for smartphone users.
What's that? There are lots of links, of course. It looks as if I have to
change all my pages in order to be able to display correctly on mobile devices.
Do I want to? I format my pages to fit well on a normal monitor, and one of the reasons I
dislike mobile devices is exactly because they're such a pain to display text on. But it's
worth reading what they had to say.
The main issues (status 3/14/15, whatever that's supposed to mean) are:
Viewport not configured (3,000 pages). It seems that this one is simple. Just add this
line in the appropriate place:
Small font size, also all 3,000 pages. Huh? I deliberately haven't specified any font
size, exactly for this reason. But I followed the link and discovered:
Use a base font size of 16 CSS pixels.
Why? There's a long discussion of pixel scaling, so CSS pixels aren't real pixels. But
why should CSS have to say anything about the base font? Tried setting it just for the
fun of it, and it didn't make any difference on my screen.
Touch elements too close. What does that mean? I don't have any steenking touch
elements. But they mean links, of course:
This rule triggers when PageSpeed Insights detects that certain tap targets
(e.g. buttons, links, or form fields) may be too small or too close together for a
user to easily tap on a touchscreen.
Now things are getting complicated. Yes, I have lots of links on my pages. That's a
feature, not a bug, and moving them further away would make the page unwieldy. On the
other hand, I have difficulty accessing the pages on my Android tablet. I
blame this on the tablet, not my pages. Still, a lot of this depends on the viewport,
though I haven't seen any difference after adding one.
Content not sized to viewport. Well of course not; I didn't have one. It's just
surprising that this only refers to 2,688 of the 3,000 pages. But the recommendations
all seem to apply. As the TL;DR says:
Do not use large fixed width elements.
Content should not rely on a particular viewport width to render well.
Use CSS media queries to apply different styling for small and large screens.
Modulo viewport I'm doing all this already, and the text of my diary scales nicely on
even a very narrow display (though the headers don't).
I copied the list above from the Google web site. It proved to be invalid HTML.
So I should probably consider widening the gap between links, but only for toy displays. I
certainly don't want to mess up the links on a real display. And before I do anything, I'll
see what difference the viewport makes.
Work on the house seems to have slowed to a trickle again. The plumbers were there
yesterday, but I wasn't. I still can't get in to the house, but I can see the kitchen, and
the tap for the sink hasn't been installed. What else is missing? Why weren't they here
today?
On the (vaguely) positive side, the fence round the house has been removed:
This weekend the Golden Plains Arts Trail is taking place. Yvonne tells me that there are 99 artists
in 30 locations around the Shire, but you wouldn't find it (or very much else) on the web
site. Still, she's participating, so
she was up and off to the Dereel Hall
early this morning. I went over later to take a look:
Admittedly, it was only 10:00, but I was the only visitor. Yvonne told me in the evening
that they had had several hundred visitors, but no sales. I'll be interested to see how
tomorrow turns out.
Our failed trip to Melbourne had a consequence: we
didn't have enough to eat. While Yvonne was doing her arts
thing, I went into town and bought some food. Also stuff for the house: coincidentally ALDI
had a “3000 W” generator:
Still, I've seen that before, and was even expecting it: my last ALDI generator (“2200 W”) was rated at 2 kW.
And 2.7 kW is exactly the rating of the UPS I intend to buy, so it sounds like an ideal
choice. Bought the last two—exceptionally, not in case one fails, but because Chris wanted
one too. Also bought some outside lamps and infrared sensors at Bunnings. Now we just need lights for over the
bathroom mirrors, but we're not decided on what they should be. All the lights had power
plugs, not quite what I'm looking for.
I was out of the office most of the morning, but when I got back I found:
Start time End time Duration Badness from to
(seconds)
1426893625 1426895771 2146 0.005 # 21 March 2015 10:20:25 21 March 2015 10:56:11
Another 35 minute outage! What caused it? The trace is confusing:
10:06:31.519432 IP aussie-gw.lemis.com.bootpc > radiation-tower.aussiebb.net.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
10:06:39.520416 IP aussie-gw.lemis.com.bootpc > radiation-tower.aussiebb.net.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
10:06:48.521437 IP aussie-gw.lemis.com.bootpc > radiation-tower.aussiebb.net.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
10:06:48.566402 IP radiation-tower.aussiebb.net.bootps > aussie-gw.lemis.com.bootpc: BOOTP/DHCP, Reply, length 300
10:56:10.524430 IP aussie-gw.lemis.com.bootpc > radiation-tower.aussiebb.net.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
10:56:30.525427 IP aussie-gw.lemis.com.bootpc > radiation-tower.aussiebb.net.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
10:56:54.526425 IP aussie-gw.lemis.com.bootpc > radiation-tower.aussiebb.net.bootps: BOOTP/DHCP, Request from 00:50:da:cf:07:35 (oui Unknown), length 300
10:56:54.570776 IP radiation-tower.aussiebb.net.bootps > aussie-gw.lemis.com.bootpc: BOOTP/DHCP, Reply, length 300
The first three requests are for a lease renewal, which is finally granted in the fourth
packet at 10:06:48.566402. And then there's nothing until 10:56:10.524430. Why not? The
lease duration is 60 minutes, so the client should have sent out a new request at 10:36.
Why didn't it? It seems it did:
Mar 21 10:36:43 eureka dhclient[24789]: send_packet: Host is down
Mar 21 10:37:20 eureka last message repeated 4 times
Mar 21 10:38:56 eureka last message repeated 5 times
Mar 21 10:47:42 eureka last message repeated 12 times
Presumably the ARP
information had expired by the time it tried. So the lack of communication wasn't obviously
related to DHCP this time. And when the
connection was reestablished, a renewal request was enough. Yet another problem? When will
they iron out these issues?
Some months ago, when I was complaining about Facebook, Peter Jeremy (a Google employee) suggested that I use Google Plus instead. I signed up, took a look, and found
it confusing. It also didn't address the issue that I don't like the concept anyway. So I
forgot about it again.
Then today Peter was analysing the display of my web pages on mobile devices. I asked him
for screen shots, and got the reply:
<groggyhimself> Can you send me a screen shot?
<peter> groggyhimself: I've shared them with you.
<groggyhimself> ?
<peter> groggyhimself: Look in Google+.
* groggyhimself wonders what he has done to annoy peter.
So I went looking. I didn't find anything useful, just lots of input from people
I've never heard of about topics that don't interest me. After 10 minutes following all the
links, I couldn't find anything “shared”. Peter couldn't tell me either. Why hadn't
he just sent me the images by email, like God intended? In the end he did, but why didn't
he do that in the first place? Android makes it
difficult, while “sharing” is apparently easier—when it works.
The upshot: I've never heard of any of them, and none are anywhere near Dereel. One is
in St Austell, which I last visited
about 45 years ago, one
in Sebastopol, California,
which I last visited on 12 December 1996, one
in San Francisco, which I last
visited on 10 February 2002, and one
in Berlin, which I last visited in
27 December 1969. What's recent or local about any of that? Somehow
Google's personalization leaves a lot to be desired.
Over to the Stones Road property in the evening to take the dogs for a walk, and bumped into
Garry and Diane Marriott, who are back again on their way
to Darwin. Discussed
a number of things, including a hedge between the properties. Diane has long ago planted
some flowers there, including Hebe
and Salvia microphylla, all
looking very unhappy without enough water, so we'll water them with the bore water. Another
tree, also very unhappy, was making the best of it:
Chris Bahlo is buying my shipping container, and for months we've been trying to clear
things out. Today we finally removed the final cartons—and then replaced one, a box full of
old lever-arch files that I gave to her. Tomorrow the container will be picked up, and
hopefully also the one on David Yeardley's property. He recently sent her an email message
asking her to remove it:
From: David Yeardley <davidgyeardley@gmail.com>
Date: 18 December 2014 at 14:57
Subject: Removal of Items from 276, Rokewood Junction Rd, Dereel
To: Chris Bahlo <cbahlo@narrawin.com>
Since it is soon to be the season of goodwill to all me can you arrange the
have ALL your items removed from my property before Christmas.
And yes, the wording is original.
But since then he's been getting up on his hind legs and claiming that Chris owes him all
sorts of money for unspecified reasons. It's possible that he might try to cause trouble.
Tomorrow could be interesting. I can see her having to call in the police to get her
belongings.
Last time I tried this, the sauce was far too watery. So I reduced the liquid by 50 ml.
The result? Still too watery. The only other liquid in there is cognac, currently
30 ml. It's not clear that it makes much difference to the taste, so next time I'll reduce
it to 5 ml.
As planned earlier this week, I bought some cheap toner for my laser printer. How do you install it? Easy, in
principle. But there's a little problem. Here are the new cartridge on the top (still with
protective cover) and the old one below:
Half the cartridge is missing! Am I supposed to re-use the old one? If so, the least they
could have done was to tell me how to do it.
The other difference showed up when I read the printer instructions. The new cartridge
(“high yield”) is good for 2,500 sheets of paper. According to the printer instructions,
the genuine ones are good for 12,000. That's very close to the price differential I
mentioned. And the genuine cartridges are available for $90 from other suppliers.
On the face of it, I should return the cartridge and buy a genuine one. But will I even get
2,500 more pages out of this printer? It's not very old, but in the 8 months I have had it,
I have printed 460 pages. At that rate, the cheap cartridge will last another 43 months,
and a 12,000 page cartridge would take 17 years to finish. And I don't expect the printer
to last that long.
I later discovered that
I had been reading the wrong instructions. The replacement cartridges are completely
equivalent to the originals, both in capacity and in size: the other half is the drum
unit, which needs to be replaced separately, but not as often.
David Grigg of BREAZE came today to take a
look at Chris' house and give her a quote. I came along for the fun, though I didn't learn
very much that was new.
But David Yeardley, true to form, had ensured that Chris couldn't get at the one on his
property by parking his tractor and other equipment in front of it:
I sent him an email asking him to move it, which of course got no response. After the green
container had been unloaded, Chris and I discussed the likelihood of getting the other one,
and though the driver was prepared to do so, we saw little chance of getting the container,
so we decided to leave it be.
Going past the Yeardley house, however, I decided to go in anyway, though I had agreed with
Chris not to do so. He was there, and opened the door. I asked him to move the tractor.
The fireworks were spectacular. He cursed me continually with a surprisingly limited and
colourless vocabulary, and claimed that I had been told that I was to keep off his property.
Fine and good, but I came with a power of attorney from Chris.
I asked him why he wouldn't move the tractor. He denied not wanting to let her have her
belongings, but required—for the first time—that she take everything else first, and only
then would be allow her to have the container. And that while the transporter was waiting!
I told him that he didn't have the right to impose that kind of requirement, but he knew
better: he's the owner of the property. A couple of going round in circles with “yes, but
you're not the owner of the container”, and he slammed the door in my face so hard that I
thought he would damage it—maybe he has. And his dog tried to dig his way out of sight.
Presumably he has seen this behaviour before.
I was about to go when he came back out again, shouted a few more obscenities, and continued
the “discussion”. This happened 3 or 4 times. Other issues he had was that I had
called him a thief. I had
genuinely forgotten, and he had to remind me of the details, claiming that I had made it up.
Of course, as I said, my opinion of him was so low that the theft didn't rank overly highly,
but that I stood to it.
At some point he did give me something: a folder with hundreds of pages of
spreadsheet printouts, purporting to show everything that she owed him. I only looked at
the first page: 50% of power, phone, etc, from about 2010. Why should she have to pay that
back? Why didn't he ask for it earlier? They had a company together. He's going to have
difficulty claiming that she did all the work for nothing and had to pay half of all costs.
Somehow he seems to have a poor grasp of reality. I recommended that he go to a solicitor
before making this kind of claim, but he said he knew he would never see a cent from her.
Fine, then we can forget the whole thing.
And then he claimed that Chris had sent me. I told him that it wasn't so, but of course he
knew better. As I told him, he has a surprisingly poor understanding of other people. I
left him with the exhortation to behave like a responsible adult, but of course that's
beyond him.
Lee Kuan Yew died this morning. He
was 91, so it wasn't completely unexpected. I'm saddened beyond belief, though I never met
him. But he was one of the great statesmen of the age, one who was always in the news
during my childhood, and one of the few politicians I really admired. The world will be
poorer without him.
Today was the day we finally went shopping
in Melbourne. We've been planning it
for weeks, and like last
Thursday, things didn't go smoothly. We got away on time, but somewhere in Enfield State Forest we came
across a car crawling at about 20 km/h under the limit. Came to one of the few places where
we could overtake, floored the accelerator... and the car didn't accelerate. Instead the
dreaded “Check Powertrain”
occurred, and the car ran as roughly as ever.
Fortunately we weren't far from home, so hobbled back, put the fridge and the Tektronix 555 oscilloscope in the Hyundai—the
latter barely fitted—and off with a delay of about 25 minutes. But what is causing this
problem? Once again it happened after heavy acceleration, and without any warning. I
should get my ELM327 out again and
see if it can find any codes this time.
After changing cars, the journey to Melbourne was uneventful. I had managed to persuade
Yvonne to come to the Footscray market, which has
mainly East Asian foodstuffs, but also a good choice of fish. So what did we buy? Meat.
Then on to deliver the oscilloscope, hampered by lack of telephonic communication, but
finally made it to the Physics Department of the Melbourne University and dropped it with no further issues.
From there to the Victoria Market, where Yvonne
bought her fish, and we also got more meat and sausage. We must have spent $400.
Then Yvonne wanted to view a pet door with a difference: it's a whole panel that fits in beside a sliding door, and despite my
concerns, she had found a place with one on display. Not easy to get to: 280 Canterbury Road, Surrey Hills, along some of the
most congested roads in Melbourne. After spending 10 minutes getting the 600 m to Victoria
Street, gave up and decided to head
to Springvale first. There
we had a reasonable lunch in a little Vietnamese restaurant—the biggest problem was
understanding what the dishes were. East Asian cuisines can be remarkably varied.
Then to the new IKEA complex nearby. I'm
beginning to wonder if our IKEA days aren't over. We didn't find anything we both liked,
and the displays exhibited a particular lack of care and interest. In particular, they were
badly assembled (exactly the impression they don't want to make). Even apart from
that, we found surprisingly little of interest. Left with a number of catalogues, but
without buying anything.
Next door was a Harvey Norman complex,
where Yvonne wanted to see a lounge suite. Lots of walking around: they appear to have
placed the escalators in a position to maximize the distance you have to go. When we
finally found something, it was not suitable.
Then finally back to look at the pet door. Took us the best part of an hour to get there
and back to the freeway, and all they had was a single small model. The adjustments at the
top look a little dubious:
The real issue, though, is the overall width of the unit, something that the manufacturers
carefully omit from their list of dimensions. The
one we saw (“small” size) was 31 cm wide, and the flap was 16 cm wide. Assuming the same 15
cm overhead, an “XLarge” with a 31 cm flap would be 46 cm wide, which doesn't leave much
space in a 70 cm wide opening. So it looks as if the whole thing is a non-starter.
Parking at the Victoria market was made no
easier by the parking meters. They're modern and electronic, of course, with a
low-contrast, reflective LCD display:
Today was overcast, so the display was marginally legible. On a sunny day I would have been
facing into the sun, so things would have been much worse. Even so, to read the display you
need either to be about 1.50 m tall or kneel down in front of it. Standing at a normal
height (well, mine, anyway), all you see is this:
Apart from that, how do you choose your time? You don't. You choose the amount of money
you want to pay, in increments of $0.50, by pushing the + and - buttons.
But they don't work! Oh yes, they do, sort of. I've long complained about power buttons
that you have to hold down for ever to get any result, but these + and -
buttons have no auto-repeat and respond—apparently—to about one press per second. Whoever
designed this appalling piece of equipment?
In passing, it's interesting to compare the time displayed on the panel with the time my
camera reports: 2 minutes difference, to the benefit of the City of Melbourne. And yes, my
camera had the correct time.
Today was the first time I used the new GPS navigator in the city. Of course it tried to
take me through places that didn't exist, but the most surprising thing was what happened
when I tried to find alternative ways from the city
to Surrey Hills: it kept
changing its mind, wanting to take me north of Victoria Street (probably correctly) and
south of Victoria Street (very definitely wrong). And when we changed our minds to go
to Springvale, Victoria
instead, it tried to take me straight through the middle of town instead of the designated
way to the freeway. I wonder if big name navigators are this bad.
Our new car fridge did its job well—most of the time. On a couple of occasions I checked
the temperature, only to see a display “Err 1”. What does that mean? I was able to clear
it by power cycling the fridge. On returning, I found in the manual:
Low voltage? Sounds unlikely. But clearly it's something to keep an eye on. The
interesting one, though, is Error 5: “Temperature of the mould anormal”. I wonder what that
means. Still, at least there's some description.
Registration SAO 263. No invalid sign in the car. Parking in an invalid space, although
there were plenty of others available. What do such people think?
Spent much of the day packing up yesterday's purchases. In particular, we had bought 2 kg
blocks of whitebait, frozen in salt water. How to repack them? I left them to half-thaw
and then divided into 200 g packets. Unpleasant, to say the least; rubber gloves help.
Yvonne had more fun. Today was her normal shopping day, and
she didn't get finished with everything until evening.
Baking bread again today, nothing unusual except
that it was the second time I used the new bread pan. This time my
quantity calculations were alright, but for some reason the bread is much softer; I have a
feeling it was last time too, though I didn't note it. The bread has a larger
cross-section; could it be that it needs longer baking?
It's been a week since I heard from Duncan Jackson about the house, and in that time not
much has happened. Gave him another call today and discovered that they've been discussing
the matter, and they'll replace the tiles and the doors as requested. I suppose it's worth
a wait for that.
Over to the property later to drop a generator, and checked the laundry door, where we had
considered putting yesterday's dog door. It's only 70 cm wide! A 46 cm dog door wouldn't
leave much space for humans—even the “small” door would take up nearly half the width. So
that's one less possibility.
That's supposed to be a map! I've complained about these sites in the past, repeatedly, but things seem to be
getting worse all the time. And yes, it's like that under Microsoft as well, though in each
case resizing produces a recognizable map.
To be fair, there's a new web site which looks a lot better. But why don't they shut this old cruft down and redirect? Going
to the CFA web site doesn't help; it's
another empty map (maybe):
When I got my Olympus
OM-D E-M1, I briefly
tried a photo or two at the highest sensitivity, 25,600 ISO (linear). It
wasn't very good, but I didn't expect it to be. But today I took some serious photos at
that sensitivity. How do you process them? Ran them through DxO Optics “Pro”, and got
something almost usable. Here the original and what DxO made out of it:
Getting that far wasn't easy. In particular, as the original shows, the white balance was
way out, and it took me a lot of experimenting to get it right, not helped by the fact that
the preview wasn't correct either.
After the events of the last few days, today was a bit of an anticlimax. A few mail
messages about the house, started a new Coursera course on Maps and the Geospatial
Revolution—which still has to prove itself—and that was about all.
My last attempt at a fasting
blood test was a disaster, and I've been putting off the retry until I can get confirmation
that I will get priority the next time round. But gradually it's time. I was given a
number to call, so did that—and got a fax machine! These people seem to be completely disorganized.
Despite yesterday's failure
to contact Healthscope, off
to Sebastopol this morning
for a blood test. Got number 7 and found two people waiting in front of me, numbers 1 and
5. This all looked very much like what happened last time. I was told to press a bell and
wait for the pathologist to come out. Nothing happened for 15 minutes, after which the
pathologist opened the door, presumably not as a result of my button press. There was no
patient inside. And it was this same woman who had given me so much pain two months ago,
and had apparently processed numbers 2 and 6. It seems that she's not a temp, she's the
replacement. I asked for priority, as discussed. No, nothing doing. Please wait your
turn. “Would you please confirm with Dr. Majid?”. “No, wait your turn”.
That's enough. We had an agreement, and that this rate I would have been in for another 80
minutes' wait. I got quite angry. Was just storming out of the clinic when I bumped into
Dr. Majid, who had just arrived. He went in to have a discussion, and then we had a
discussion in his surgery. Yes, she's new, and inexperienced. What he didn't say, but what
seems clear, is that she's also excruciatingly slow. So he gave me a pathology
request for Dorevitch, with whom I have
had good turnaround in the past, so I'll
do that next week. It's not really a problem with Healthscope, though, just with this
specific pathologist.
While discussing the matter with Dr. Majid, he said something that sounded more in keeping
with his national origins that typical Australian behaviour. And his origins have been a
puzzle, one that we discussed on IRC. He looks and behaves like
an Iranian, but as Mohamed Ifadir tells
me, his name (Majid Najafi Zeini) is Arabic, specifically referring to the town
of Najaf
in Iraq. So I asked him. Mohamed and I are
both right: he's Iranian with an Arabic name, which annoys him no end, though Najaf is an
important Shia centre.
While in town, got my hair cut. So it wasn't a complete waste of time.
Over with the dogs to the Stones Road property this afternoon, and noted that one of our
water tanks (obviously the one fed from the house) is nearly full. Time for content
balancing.
While we were there, at about 17:30 on a Friday, a bloke turned up with shower screens.
That gave us a chance to take a look around. Indeed, very little of the plumbing has been
done, and the wash basin is still not there. Hopefully they will soon perform a digital
extraction.
Another attempt with the ELM327 clone
today to find out what's wrong with Yvonne's car. Once again it claimed that there was no
error code stored. Tried to speak to Paul Sperber of Ballarat Automotive, but was blocked
by his wife, who went in, discussed something with him, and said that they weren't
interested in doing the work, and that I should go to Ballarat Central Auto
Electrics. Called them, and eventually spoke to Wayne, who confirmed that it would be
difficult to diagnose if the problem wasn't presenting itself, but that they'd try.
Otherwise he suspects coils, which isn't out of the question, but I'd like some certainty
before spending $600 on the off chance.
Talking with Chris Bahlo this evening about the fun day she had had at work today. They have a new coworker, and she
managed to pessimize a web site by putting in overly large images. How large?
300 dpi. What does that mean on a
web page? Should it scale to the monitor resolution? No, it seems that people still can't
get used to the fact that image resolutions are measured in pixels (these images were
apparently 1200×900 or so, clearly too large). Admittedly pixel dimensions aren't ideal
either, but what earthly use are dpi specifications?
The Yeardleys had specified that Yvonne and I were not welcome, so we had nothing to do.
Margaret Swan and a number of people from the Men's shed came along to help. Things went surprisingly smoothly, and by evening they had moved all
except for the container itself:
It seems that Margaret worked wonders on the Yeardleys. When they arrived, they were as
aggressive and unfriendly as ever, but Margaret says she somehow managed to convince them
that that made no sense, and by the end of the day they were decidedly friendly. Let's just
hope it stays that way.
He was a particularly surly bloke, possibly because somebody had talked him in to working on
a Saturday afternoon; told me to get out of the fenced area. That's his prerogative, but
the first time I've experienced it on either site. Still, the house is coming along again.
We still can't guess which will be finished first.
Decades ago we objected to the German proverb “Man muss die Feste feiern, wie sie fallen” (“Celebrate feast days as they occur”),
something that the Germans take very seriously. So we decided to celebrate Easter at
Christmastide, causing serious concern when we went looking for paint for colouring Easter
eggs.
We can do that again. After Christmas, Chris Bahlo had bought a turkey and froze it. Today
we ate it, 3.8 kg, no stuffing, took 125 minutes in the oven at 180°. The breast was at
78°, a little too dry. I'll try 76° next time.
Nikolai has an interesting habit of raising
his ears when he's interested in something. You don't often see that in
a Borzoi, but Ron Frolley tells me that
it's highly desirable. Sometimes, though, I think he cheats:
Into Ballarat this morning for an eye
test. Took Yvonne with me, since last time they put stuff in my eyes
that made it almost impossible to drive. But this time they didn't. And no change to my
prescription. Decided to have some sunglasses made, which will be ready in two weeks.
To Ballarat Central
Auto Electrics with
the Commodore to see if
they could find any diagnostic information from the ECU. Yes, indeed: two coils
failed. And that will cost us round $700! Tony, the bloke who did the test, also has a VZ
Commodore, and has just changed his coils: it seems to be a common problem with them. And
it was with our old VT
Commodore. So why do they make it so difficult to change them, and why are they so
expensive? And why does a coil failure cause these symptoms?
And, of course, it means that my ELM327
clone is useless. What a waste of $5.99!
Over to Stones Road at 7:30 to meet with Duncan Jackson, who had planned to stay only about
10 minutes. In the end, it was over half an hour. Things are looking good on the whole,
though he tells me that they're not prepared to fix this problem, claiming it's normal:
There are still a couple of issues, as shown in the last two images: no tap for the basin in
the master bathroom (centre photo), not even a basin in the other. And Yvonne's Moroccan washbasin in the master bathroom has a non-standard
drain opening, and we need to find out how to connect the drains to it.
They'll put the floors in next week, which will take at least a week, and in that time
they'll also replace the internal doors (8 mm gap from floor), the damaged rear door to the
garage, and install the main garage door. Except for the electrics, that's everything. We
still have hope to move in by the end of April, and now we need to collect documents to
apply for the Certificate of Occupancy.
Jim Lannen around an hour later to do the final fit of the electrics. A couple of
misunderstandings there, hopefully to be resolved. Fortunately
the UPS arrived in the course of the
day—actually delivered to Kleins Road!—so just about everything seems to be in place.
Last week I bought an
after-market toner cartridge for my Brother HL-3170CDW laser printer, and established that it was only part of what I had
expected: only half the unit, and 20% of the capacity. Complained about the former to the
supplier, who sent a description of what to do: press the green button on the left, and the
cartridge detaches from the drum. Press drum onto new cartridge, and you're away. And sure
enough, it works fine.
But what about the differences original and aftermarket cartridges? Checking again, I found
that the instructions I had read were for my oldHL-2700CN printer—and they disagreed with themselves, claiming 10,000 pages in one
place and 12,000 elsewhere. The correct
manual describes the procedure on page 111: exactly as described. The original toner
cartridges also do not include the drum unit, and it also states (at least in one place)
that the black toner is good for 2,500 pages. That also explains why the aftermarket
supplier didn't include any documentation. So the $25 replacement cartridge is completely
equivalent to the $85 original cartridge.
Call from Wayne at Ballarat Central Auto Electrics this afternoon. While replacing the coils on
the Commodore, they
discovered a vacuum hose with a tear in it. Clearly it needs replacement. More
importantly, though, it completely explains the problems that we have had, and it's
consistent with my guesswork two
months ago. Yes, the ECU confirms that we needed new coils, but that wasn't the cause of the symptoms.
And the ECU didn't notice the problem with the hose. Why not? At the very least it should
have registered the “Check Powertrain” display.
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.