Despite the weather, Yvonne decided to mow the lawn today,
which involved me more than I had expected. First the tyres all needed pumping, and I
discovered that I couldn't connect the tyre pump attachment from the new compressor to the
old compressor: the connections were of the same kind, but both the connector on the
compressor and the one on the gauge were male. So I had to use the strange connections of
the old compressor again.
Then Yvonne came in: she had run into something and damaged the lawn mower, which had
promptly stopped working and produced lots of smoke. Further investigation showed that one
of the blade assemblies had fallen down in the housing, and the smoke was coming from the
Yvonne called up CJ, who was a mechanic in another life, and he came over to take a look,
and we decided that he would take it home with him and look at it there. While driving it
up the ramps to his ute, the cutter housing caught on one of the ramps, pushing it aside,
and the mower fell off to one side and on top of my leg, the one that has just healed from
my accident three months ago, pinning me in
place. Fortunately it didn't do much harm, but Yvonne thought I had broken my leg. Clearly
not the way to transport the thing.
While the mower was lying on its side, took the opportunity to take a look at the cutters.
As I had feared, the cast aluminium mounting of the shaft, which the workshop manual calls
“Housing Mandrel Vented”, had sheared off from the mounting holes. Hopefully we
can find a replacement. Yvonne went over to the Yeardleys to borrow their horse float to
transport the mower, but in the end they came over and drove the thing themselves.
While we were at it, gave CJ the new compressor to play around with. Possibly we can get it
fixed relatively cheaply. We're also talking about putting up shade cloth to the north of
the verandah; hopefully we can get that done relatively soon.
In the evening, Yvonne went and dug out some of the mystery flowers that we had seen
two weeks ago. She thought she had
brought about 4; in fact, there were 14 corms. They look like the same kind as the
Watsonia and the Chasmanthe floribunda:
I haven't been doing much in the garden lately, but that changed today. In the morning spent
some time weeding the old vegetable garden, which has become seriously overgrown. That'll
keep me busy for a while, but I've made some inroads. Also took a look at the Melaleuca which has been in bloom for the
last two weeks. It's now coming to an end, and has dropped a remarkable number of stamens
on the ground:
The first of the Gazanias that we
picked up in Stawelltwo months ago is also flowering (last photo
below). Interestingly it doesn't quite match any of the photos I took at the time. The
first two photos are the most likely candidates, but I can't decide which one it's going to
In the afternoon, Yvonne decided that she needed to do
something about the area outside the entrance to the house, which had been severely
overgrown by Euphorbias and
Pelargoniums. She removed a
lot of stuff:
Chris came over to pick up a lot of the cuttings, and we recalled that she wanted some
Arums, so I started digging out the
ones that we wanted to remove from the middle of the east garden. This is what it look
like after she took her 20 or so plants:
The water pouring through the first photo came from the junction between the roof of the
house and the roof of the verandah. The latter terminates over the gutter (second photo),
so either the gutter is blocked again or there's something else I don't understand.
The rain was also accompanied by one of the fastest temperature drops I've seen in a long
time, 10° in 35 minutes:
mysql> select time, outside_temp from observations where date = "2010-12-2" and time > "15:00"; +----------+--------------+
| time | outside_temp |
| 15:00:52 | 26.1 |
| 15:01:52 | 26 |
| 15:35:01 | 16 |
And of course when there's heavy rain, there's usually a power failure. Today there were two, both mercifully
Another effect of the rain was that, once again, the rainwater filter got clogged up, not
once, but twice. We must have a lot of crud in the bottom of the tanks. Time to find a
I'm subscribed to the Oxford English Dictionary's
“Word of the day” service, for which they conveniently don't provide a URL.
It's vaguely interesting, and of course OED is the authority when it comes to the
English language. As you'd expect, there's not a word or punctuation mark out of place.
But that's the book. As we've seen elsewhere, nobody on the web takes accuracy seriously.
Their word of the day stuff was sent in HTML, of course, and in extremely bad HTML at that.
In particular, the phonetic characters were sent not as characters, but as images, which of
course didn't scale:
The text was in UTF-8, but the content type has been set to ISO 8859-1. A simple mistake,
one that I make myself from time to time. But I usually see this kind of error and correct
it before anybody else sees it. This example is the third day that they have been sending
this breakage. It's not a minor issue, and it's not limited to pronunciation; it seriously
1897 Westm. Gaz. 23 June 2/2You drink in the picture.â€¥ This, you involuntarily
cry, â€˜This is the champagne of the century!â€™
It also doesn't fit in the category of errors that don't get detected on Microsoft due to
some bug in their software: “Internet Explorer” renders it in exactly the same
way. How can a big company with a reputation for accuracy allow this sort of thing to
happen? But it's clearly the way of the future. They've removed all the images for the old
pages, so now I can no longer read the pronunciation for the older pages I haven't read yet.
It's been nearly a month since I cancelled
Telstra BigPond's network
“service”, and I have been meaning to change to their business service. I
haven't done so: the problems I had with BigPond have so completely drained me that I've put
up with this appalling SkyMesh service.
Not that it's good: it's getting worse all the time. I have the distinct impression that
they have put somebody else on the same channel as me (cancelled the “Special Class of
Service”), as they call it. The terrible implementation of their satellite protocols
means that any reconnection can potentially disconnect somebody else, and I'm seeing
something similar: when I start a larger data upload, for example syncing my photos to the
external web site, I regularly get disconnected after about 30 seconds. After reconnection,
things can run well for a while, probably because the other station has chosen a different
channel. But it's all so difficult to determine, since the other problems are still there.
How many disconnects do I get? I don't know. I originally wrote my tracking software for
ADSL lines, and there it was reasonably accurate. On the satellite, I regularly get
dropouts that are so short that I don't register them. But so far this month there have
been 29 outages long enough to measure:
Date Outages Duration Availability Date
1291122000 10 1007 98.83% # 1 December 2010
1291208400 8 1030 98.81% # 2 December 2010
1291294800 11 1181 98.63% # 3 December 2010
I should really do something about this.
In passing, it's interesting to note that even with the antenna, I still can't get any
signal at all with the 3G21WB that
they haven't asked me to return. It's possible, of course, that I have received two
non-functional units, but that doesn't sound overly likely.
Yesterday's weather was strange enough to comment on, both the heavy rain in the afternoon
and the sudden plunge in temperature. We had pretty much the same thing today: not as much
rain, but an even steeper plunge, from 30.5° to 16.2° in less than two hours:
Talking on IRC about when we started using UNIX today. In my case, I got my first desktop
UNIX system (Interactive UNIX System V/386) in May 1990, when I was manager of Tandem Computers' European UNIX
Technical Support. When I left Tandem, I switched to BSDI BSD/386, which I got as beta. For the
first few days, I put in beta test reports, as I had been used to do, but it became clear
that that wasn't the way people in BSD land did it, so I stopped. But I still have the
reports, and today I formatted them for the web and put them up as my diary entry for
Then it dawned on me: with horizontal orientation I have the nodal point of the lens on the
rotational axis of the pan head, but when I mount it vertically, the pivot point is 1.5 cm
further back. I make up for that by moving the focusing rail forward 1.5 cm, but that only
works when the lens axis is horizontal. I do this panorama in two rows, and the vertical
rotation is round the wrong horizontal axis, so I get parallax again. The following photos
show the principle, though the camera is mounted the other way round here, 1.5 cm further
Today the weather forecast predicted rain and storms, so of course it was bright sunshine
all day, and once again very warm, with a top of 32.9°. That created more than usual
problems with contrast, so for one of the panoramas I took a full 5 EV sweep (5 exposures
for each view, 1 EV apart). That had one obvious problem: how do I tell my processing
software what to do with these 72 images? I have basically two description files. The
first, makejpeg, is simply a list of source file names and the names they should
The other two values are rotation and gamma correction, something that I no longer do at
this point. That gets put through a script imgconvert that creates the files. In
the case of HDR photos, the names contain the EV offset, as shown above.
Creating that list is enough fun in itself. Under normal circumstances I use an interactive
script that prompts me for the names, but that doesn't work for the house photos. Today I
had a total of 269 images requiring processing. Clearly I need a script to generate
the makejpeg file. That takes another input file, housephoto-notes, with the
names of the individual files and not much else:
#### verandah (36 base images)
# From irrigation distribution to S side of door of 10
# 1 to-verandah-a+0EV
# 1 to-verandah-b+0EV
This describes which photos to convert and which to skip (the lines starting with -). It
also serves as notes to me about what to do. The initial digit is no longer used. I have
written some Emacs macros which will generate a sequence of these names. The problem
is that the sequences are fairly rigid, and the generated text is difficult to read. Today
I needed something like:
I could have done this with another Emacs macro, but it's getting tiring. Instead,
wrote a short PHP script that I had hoped would generate a makejpeg file, but that
required accessing the source files, and I didn't want to wait with the processing, so
instead I got it to generate a housephoto-notes file. The input to this script
(mkmakejpeg.php) is much more concise:
This specifies the base name of the photos, the number of views, and which of the views are
to be used (marked with *). Thus in the first panorama, all images of each view
are used, in the second panorama just the first, second and fifth image, and in the third
the second and third image.
So was it worth it? It's difficult to say. Here's the panorama done with -1 EV, +1 EV and
+3 EV, followed by the one taken with the other two images (0 EV and +2 EV):
It's been like that for three months now. That's absolutely extraordinary. The previous
highest level of fill was on 10 November 2007 after some particularly
heavy rain. That didn't last, and it wasn't nearly as full as on 4 September 2010:
Since then I've changed the view: the location from which I took these images has been under
water for that time, and the panorama looks more interesting. The new viewpoint is on the
rise to the left of the dam in the old photos, and the old viewpoint is at the extreme rear
right of the panorama:
On the first Saturday of each month I take some additional photos that I don't do on the
other weekends. The newest is the ex-“cathedral”, for which we're hoping great things. At the moment, though, I only
have two images; the first was taken four weeks ago, and the second now. They show one
thing: how much the weeds have taken over in that time.
A couple of birches and the Araucaria bidwillii had disappeared below the weeds, and despite the weather I dug
them out. Fortunately they were none the worse for the experience. I'll do the rest when
it isn't so hot.
Today wasn't as hot as yesterday, but still too warm for much work outside, so spent some
time updating my house photo display pages,
which are in a bit of a mess. In particular, I've changed a number of views of the garden
in the course of time. What do I do? Do I destroy continuity and rename the photos, or do
I destroy the view and give the new one the same name as the old? Either way it's a
problem. So in the past I have done it both ways, depending on the view.
The very oldest view of the house is north-east from the verandah in front of the lounge
room, looking at what is now the south side of the verandah. Things have changed greatly in
the last 3 years. Here the first photo, taken on 19 October 2007 and
the one taken yesterday:
The first photo is exactly the same viewpoint as the first of the previous two photos.
Was that the right thing to do? Or should I have renamed it? That, too, has changed beyond
recognition, but the new panorama seemed sufficiently different from the old that I chose to
keep the old one as well, and gave the new one another name. Here the images from
11 September 2010 and 18 September 2010:
That proved not to be the way I wanted to do things, and I modified the panorama to leave
out the left-hand part, which is yet another view, so now the first and last photos in the
series look like this:
Once again, the first photo above is exactly the same viewpoint as the first photo in the
That required a surprising amount of juggling, not without mistakes, and in the end I had to
rebuild the entire house photo series. To my subsequent surprise, it's even bigger than I
thought. I currently have 106,916 files in my photo pages (31,616 MB), of which 105,767
(31,459 MB) are taken on days with house photos, fully 99.5% of the space. Some of the
photos on those days are not house photos, but the proportion is still much larger than I
thought. Rebuilding the photos took all afternoon, and I can see that there's still work to
Looked out the window while showering this morning and saw our Strelitzia reginae almost ready
to flower. Got dressed and out with my camera, but it was too late: it had already started
to open, and within an hour or two it was completely open. Here photos taken over a period
of 100 minutes:
I assume that this is a slightly different cultivar.
Not all things are that positive. Two years ago Diane Saunders gave me a Leucospermum cross between Leucospermum
conocarpodendron and Leucospermum galbrium with the rather silly name “Mardi gras ribbons”, which has been
happily growing and flowering every spring. For some reason, I never took any photos of it,
but here are a couple of excerpts from other photos, the first taken on 19 October 2008 and the other almost exactly this time last year:
My satellite network connection continues to be appallingly bad. Today I had 19 dropouts
that my monitor scripts detected; I suspect that the real number is double that:
Date Outages Duration Availability Date
1291554000 19 1446 98.33% # 6 December 2010
That's only part of the story. Of the 1367 entries in my log files for today, 143 showed
reduced connectivity. That may look like 10.5%, but in fact it's even worse. Normally the
entries are one minute apart, which should have given 1440 entries. Because of timeouts,
the time between entries with reduced connectivity increases. Probably the easiest way to
look at it is to say that there were only 1224 good entries where there should have been
1440, so the real time with reduced connectivity is round 15%.
Almost every time this happens, BST ensures that I
drop all my TCP links. So maybe it would be easier just to disable BST. Tried that and
yes, it worked, at least for the shorter dropouts. But of course it slows
things down, and though according to my scripts the longest outage was “only”
124 seconds, the connectivity during some of these periods was so slow that TCP connections
timed out anyway:
--- www.lemis.com ping statistics ---
98 packets transmitted, 22 packets received, 77.6% packet loss
round-trip min/avg/max/stddev = 653.357/1036.120/1822.073/327.561 ms
It's becoming clear that there are two different kinds of dropout: the messy ones with
reduced connectivity over periods of time up to 15 minutes, and the fairly clear two-minute
variety. While watching the web display for one of the latter, I found the message
“Logging Out”. Why does a modem log out? Clearly it's seeing some kind of
problem, but it's not saying.
Another message to Paul Rees. Maybe they have some other issue which they're not going to
blame on my modem.
Also set to putting up some shade cloth in the area north of the verandah, but ran into
problems: we needed two brackets and two screws, and it really didn't make any sense to work
around them. So in the end we only put in a post, though the geometry of the construction
kept us busy: we have unusual angles in two different directions.
From midnight on 1 December until 18:00 yesterday I had 72 satellite dropouts, an average of
one every 31 minutes. At 17:54 I sent a message to Paul Rees asking him to do something. I
didn't get a reply, and by 18:20 I had had another two dropouts.
But that was all. The connection looks better (none of the flakiness that accompanies the
“messy” dropouts). And no further dropouts. Well, not until 16:13 today. Then
I had one completely atypical dropout: neither short nor messy. 17 minutes, and just as I
was firing up pain to find out what it thought was going on, it came back.
What's that? My guess is that the fix yesterday was to put me back onto a clear channel,
and today they updated firmware, typically without warning. Unfortunately I don't have any
record of what I had before, but this is what I have now:
I'll wait a while to see how things progress. In the past I've seen a similar sudden
improvement after a complaint, though on that occasion Paul said that they hadn't done
anything. Should I believe him?
Tony of Ballarat Water Tank Cleaning along this morning to clean out the water tanks. He
had some trouble starting his “self-priming” water pump: it seems it only
self-primes if it has some residual water in it. He had cleaned it out yesterday, so there
was no water in it at all.
Based on my previous experience, I had expected to
collect a lot of solid gunge from the tank, and had brought along some baskets for the
purpose, but in fact it was all quite finely divided, so we just pumped it out onto the
I don't think I like that very much, but the batteries charged well enough. Now I have
three new batteries, and it'll be a while before I can discharge and recharge them all once,
let alone the twice that I think I need to do to confirm that they're OK.
I was quite puzzled at first, but it seems to be normal: Sarracenias have both flowers and
pitcher traps. The Wikipedia
page has some nice photos that correspond well to my plant. About the only thing is
that it happened so late; according to the page, the flowers bloom in early spring, and last
year the pitchers were well developed in mid-November.
The story of Julian Assange and Wikileaks has been
pretty much the top of the news lately, so when I got a phone call from Andrew Crook of
Crikey. I thought he wanted to ask my
opinion. But no, it seems I had a connection to Julian: at the hearing after his arrest, he
gave his address as 177 Grattan Street, Parkville (Melbourne). In fact, it's in Carlton, and that's
where I lived as a child.
Of course, the time difference makes any connection via the house impossible. We moved out
12 years before Julian was born. But after getting another call from Reid Sexton of
The Age, I went checking. After all, he did
have some computer connections.
To my surprise, it seems that Julian is a BSD person, and that he was quite active on the FreeBSD and (especially) NetBSD mailing lists up to about 10 years ago. I even had
an invitation to meet up in Melbourne on 20 October 2000, but
unfortunately I was in the USA at the time.
He's well-known elsewhere, including in the Linux world, where I heard of nntpcache, which is also in the
FreeBSD Ports Collection and NetBSD pkgsrc, though the collected messages I have
suggest that he was more of an advocate than a coder.
It seems that Julian doesn't have much time for Linux, to judge by some of the comments in
the NetBSD fortune datfiles:
NetBSD - free yourself from all Stallmanist thought!
-- Julian Assange
NetBSD - the cathedral versus the bizarre.
-- Julian Assange
NetBSD - the power to swerve (penguins, worse than cane toads).
-- Julian Assange
I'm really quite surprised that none of us knew of this earlier.
It seems that yesterday's improvement in my satellite connection was only temporary. Today
things were nearly as bad as they have been all month:
=== grog@dereel (/dev/ttypl) ~/public_html 83 -> ping www PING www.lemis.com (220.127.116.11): 56 data bytes
64 bytes from 18.104.22.168: icmp_seq=13 ttl=55 time=852.591 ms
64 bytes from 22.214.171.124: icmp_seq=26 ttl=55 time=867.098 ms
64 bytes from 126.96.36.199: icmp_seq=40 ttl=55 time=2425.866 ms
64 bytes from 188.8.131.52: icmp_seq=41 ttl=55 time=5535.110 ms
--- www.lemis.com ping statistics ---
59 packets transmitted, 4 packets received, 93.2% packet loss
round-trip min/avg/max/stddev = 852.591/2420.166/5535.110/1908.679 ms
That's the entire ping output for a minute. And then the modem logged off. I suppose that
could indicate a link issue. But I still have no response from Paul Rees. This whole thing
makes me so angry that I can't get round to doing anything about it.
More investigation of the sick Leucospermum today, and found a reference to propagation:
in a mix of bark and polystyrene foam! That's the sort of thing you use to propagate
orchids. No wonder the soggy ground made it unhappy. I don't have the bark and polystyrene
foam, but I do have a quantity of Sphagnum moss, and now seems to be the time to plant them, so cut off four stems and
put them in moss:
A root pathogen, Phytophthora, is detrimental to the Proteaceae. Symptoms include the
plant looking wilted and dry, followed by yellowing and death. As these symptoms are only
seen once the fungus has damaged the plant, the best course of action is to remove the
infected plant and burn it.
That doesn't look encouraging. But I've done the work now, so I'll observe what happens.
Another plant that is looking less than happy is the Abutilon that I propagated last year. It
(barely) survived being planted in the ground some months ago (I think slugs were getting at
it), but it picked up noticeably when I put it back in a pot in the greenhouse. But now the
leaves are curling up:
The moist weather continues, though it's not as bad as had been forecast. But the result is
not just lots of weeds, but lots of growth. The stairs to the southern entrance of the
house (which we don't use) had completely disappeared under Gazanias and Pelargoniums. Spent some time cutting
them free again today, and ended up with about a wheelbarrow full of cuttings. Somehow I
should be able to do something with them. If I put the Gazanias on the compost heap,
they'll probably take root.
Off to Geelong today to have my
teeth looked at. All OK, but what a long way to go.
On the way home, picked up some local mussels, looked for a cheap petrol station, and off to
Bunnings to buy some hardware, all in the
greater Geelong area. It took me over an hour to get to Bunnings, not made any easier by
the way my GPS navigator changed its mind in mid-travel.
Grattan Street, Carlton: Spiritual home of Wikileaks?
The press people who contacted me yesterday wrote their stories. Crikey preferred to make it available to only
registered users (though this is free). That's somewhat at odds with the fact that they
published a photo of mine without permission (though admittedly with attribution). The
article contains a copy of this photo (and not a link to it, which I would have
In any case, the article mentions me, and even gets my job description (“A retired
Linux computer hacker”), though it claimed I had lived there since the early 1950s (I
didn't arrive there until September 1957).
The Age also had an article with their own photo taken from almost exactly the same viewpoint:
I'm pleased to note that mine looks letter. The article had the interesting title
“Leaks began in student terrace house”. That's an interesting concept; it could
even be true. Wouldn't that be amusing? For me, Grattan Street saw the arising of another
concept, at least for 9-year-old Greg: I started learning about chemistry and electronics,
and those short 17 months defined much of what I would do in later life.
The report also included part of our discussion. I had gone to some trouble to explain to
the reporter, Reid Sexton, what a hacker is, but it seems
the concept was too complicated for him. He went on to write:
He [Greg Lehey] said he had never met Mr Assange
despite their both having backgrounds in computing, but said he would have liked to.
''I wish I had of - I would have loved to have some involvement,'' he said.
''I think he's done some things well.
''I think he's going to do more [about] changing the way we talk about sharing information
than anybody else. But I think he's doing it too far.''
That's a fair paraphrase of what I said, except for the choice of words. “I wish I
had of” indeed. I wonder if he misheard (I'm sure they record these conversations;
they'd be crazy not to). I can't work out what he could have misheard like that, though.
And “I think he's doing it too far” was definitely “I think he's doing it
But that's my opinion, all right. Sometimes I think that this current situation may change
the world even more than online commerce. We're living in interesting times.
Still no reply from Paul Rees about my satellite problems, so sent a reminder. In response,
got a bounce of a mail delivery report:
<email@example.com>: delivery via spamfilter: delivered via spamfilter service
And indeed, so it did, and SpamAssassin had a field day with his message, including the
bounced copy. And he replied in extreme detail, about everything except what I was asking
for. The funniest, though, was this point, which I have reformatted the way he should have
answered in the first place:
> I would still like to hear a technical justification for this "cable
The practical justification is that if you don't do it, IPSTAR will
not progress the trouble ticket any further and we can't help you any
> The last time a techie told me is was to "let the electrons drain
> out", and that report has caused much mirth amongst my networking
> colleagues. It also makes me greatly doubt the technical competence
> of anybody who comes up with such a ridiculous idea.
While I have no basis for a judgement on the technical competence of
you or your networking colleagues, static electricity (still involving
electrons) can build up in the cables (essentially long thin
capacitors) and in the capacitors in the modem due to wind passing
over the plastic (or metal) dish and it can upset the delicate balance
of the modem.
IPSTAR claims that static electricity can cause problems and they say
that if you disconnect the cables the capacitors in the modem
discharge over half an hour or so. When the cables are reconnected
the static goes. That explanation came to IPSTAR from the
manufacturers of the ODU and the designers of the modem and it's been
proven to work. They have qualifications to back up their claims. We
have had customers who refused to disconnect the cables when asked to
do so, or said they did when they didn't, because they didn't believe
that it made any difference. Unfortunately IPSTAR can detect when the
cables are disconnected and simply refuse to progress the ticket until
What unmitigated nonsense! It's difficult to know where to start to tear it apart, but
“In the capacitors it can upset the delicate balance of the modem”. I can
just see the poor thing crying its eyes out. In fact, a static charge doesn't have any
effect on the line if it's isolated by a capacitor. If there's also a resistor across
the inputs, as most devices have, it can't build up in the first place. And this modem
has a resistance of about 440Ω across each of the cables, apparently combined with an
electrolytic capacitor, so there's no way for any charge to build up.
The quickest way to remove a static charge in a cable is to short-circuit the two
conductors. It's immediate.
The best way to maintain a static charge is to ensure the maximum resistance between the
two conductors. In the context of the modem, this would mean disconnecting the cables
from the modem so that the modem input circuitry can't drain them. Disconnecting the
cables is worse than useless against such a problem.
“It's been proven to work”. Yes, I've seen their “proofs”.
Even now, Paul is claiming that IPStar have “proven” that the modem is
“IPStar can detect when the cables are disconnected”. With a divining rod,
no doubt. If the modem is powered down, they can't detect anything at all.
I am completely amazed at this stupidity. But these are people with
“qualifications”. Maybe bought from spammers?
About the only thing that Paul said that was of use is that I'm coming up to 3 years with
the service. After that, I can sign up for the Australian Broadband
Guarantee again. That would at the very least replace the modem for free, but at the
same time I could try a different satellite. But what choice do I have in rotten apples?
Woke up in the middle of the night with a horrible recognition: I had bought 3 joist hangers
for the shade cloth frame, as CJ had told me. But since then we had changed our plan, and
we needed 4.
CJ came early, with Sue, and while getting the tools together I found a joist hanger of the
correct width, though a little longer. That was OK: we cut the edges off to the appropriate
size and we had everything in place.
We finally got the joists up, but how do we attach the shade cloth? I'm particularly
concerned about it tearing, and it's not clear what we can use, though I'm tending towards
nailing some kind of strip on it. One way or another, we don't have the parts, so I'll have
to think about it more.
It appears to be two flowers in one: the old part has lifted up and is visibly tatty, and
the new part looks the same as the old one. That probably explains the discrepancy I
noticed between my flower and the photo in the Wikipedia
article: on on Monday:
House photo day today, in a bit of a hurry before bad weather arrived. Finally got round to
drilling a hole in the correct place on my L bracket so that the verandah photos would join
properly. Did it work? I don't know. As a result of the new position, I can no longer
tilt the camera as high as before: the body touches the edge of the bracket. The result,
comparing last week and this week:
I can now tilt further down (previously the edge of the tripod got in the way), but I can't
tilt up as much. And the issues were with the area at the top which now no longer appears.
But even so there are a couple of minor jaggies at the top right, which suggests that I
still don't have it right. More playing around to do.
People have been saying for some time that modern CPUs have more power than people need.
But certainly multimedia has proven that wrong. Today I measured some of the times I needed
for processing my house photos:
That's over 5 hours of CPU time, and it's not even everything I did. Without a
multiprocessor system it would have taken much longer. No wonder I don't get much else done
Chris Yeardley over for dinner, as on most Saturdays, and we had lots of things to bake, so
we did some in our old toaster oven. In the middle of it all, heard UPSs screaming in the
computer room, and discovered that the circuit breaker had thrown. It worked OK when I
turned it on again, and it wasn't until later that we discovered that the toaster oven
wasn't heating. They're on the same circuit, so it seems that one of the elements had burnt
out, taking the circuit breaker with it. And that when we had been talking about replacing
it anyway couple of days ago, since ALDI had
them on special. Hopefully they haven't sold out yet.
One of the biggest problems we have with foodstuffs in Australia is the lack of cold-smoked
(or even hot-smoked) food, and I've been meaning to try doing it myself some time. So when
a couple of weeks ago ALDI had a smoke oven
(their description) on offer, we bought one. It proved to be damaged, but I (finally) found
the well-hidden instructions, which told me to use mesquite or hickory chips, and definitely
Yvonne exchanged the oven, but where do you get mesquite or
hickory here? My guess is that you don't, but there must be something else you can get.
I've tried the library catalogues, but it's a pain to work through the titles, since most of
them relate to smoking nicotine, not food. As a result, though, I didn't get round to
unpacking the new oven until today. At least it's not dented.
Off with Yvonne and Nemo to the German Shepherd Dog Club of Victoria training and subsequent Christmas dinner. Yvonne has been going to training regularly,
but I stopped going months ago, and only came today because of the Christmas dinner.
Dropped Yvonne at the training ground and back to ALDI to pick up a new toaster oven. Then on to Gays looking for wood chips for the smoker and a sharpening stone for the garden secateurs.
Found some red gum chips, but are they suitable? They're intended as mulch, of course.
I'll need to read up on that one. The stone was non-trivial too—the sales assistant
who helped me didn't know what I needed, but I got what I think I need.
I wonder where the money comes from for this sort of thing. We had to pay $3 per person,
but that can't cover it, and it can't come from membership fees. I wonder if the sponsors
might have something to do with it.
In next month's Gardening Australia magazine there was a catalogue of interesting-looking bulbs from
people I have never heard of, Lambley Nursery. The phone number looked familiar, so I had checked and found they're just north of
Ballarat. After lunch, set off
to take a look.
It's really quite depressing: it's not just a nursery, also a very nice water-wise garden,
which looks so much better than what we have achieved that I don't know how we could catch
up. Took a number of photos to ponder about:
Then back to Ballarat and Formosa Gardens, where Yvonne decided she didn't like the
Leucospermums very much, so we didn't buy any. She did like the Alstroemerias, though,
so we bought one of those (Alstroemeria stapricamil “Camilla”), and also a
Passiflora edulis (passion fruit) “Black pearl”, which they say is suited to greenhouse culture:
For once in a while, got round to planting nearly everything we bought yesterday; the only
exception was the passion fruit, which will need more preparation. Put the Alstroemeria in front of the verandah,
behind where the pond will one day be:
Yvonne wasn't to be outdone: her office window looks out to
the south, and onto an overgrown bed. Today she tried to remove most of the content. She
got about a third out, overfilling the wheelbarrow. And then it occurred to us that the
compost heap is full too. I'll have to move some of the oldest compost to start a new bay
before we continue.
I've been exchanging mail with Paul Rees of SkyMesh for some days now. He keeps claiming that they have done things that they
didn't, such as a complete check of the modem configuration. They intended to do
something like that months ago, but didn't. Finally he agreed to do “Level 1
troubleshooting” “again”, ignoring my statement that it hadn't been done
yet. And today I got a call from William at SkyMesh to look at my problems. He asked if I
had “little silver boxes” about the size of a matchbox on the cables (these
proved to be attenuators); I didn't. Then he switched me to a secondary beam. And that was
that. I should observe the connection for a few days and get back to him if there were
still problems. I didn't need to wait that long; in the following 4 hours I got 6
dropouts. In other words, no noticeable change.
Mail message from Paul Rees this morning. He has found a solution for the satellite
problems, at least from his perspective: he has terminated my satellite contract. I don't know whether to be happy or sad; at least
it means that I'm going to have to do something if I want continued access to the Internet.
It also annoyed me by making incorrect claims,
that I wanted to move to another provider.
Called up the Department of
“Broadband”, Communications and the Digital Economy and spoke to Sarah,
who was rather surprised by their actions, but said that there was nothing that the DBCDE
could do about it, and that the TIO was
responsible for that sort of thing. She did confirm, however, that I would probably be
eligible for new equipment when my current 3 years are up (on 20 December 2010), and she also came up with the interesting information that Aussie Broadband has a WiMAX service that is supposed to cover my
area. Called them up and spoke to Carly, who told me that they had towers at Mount Warrenheip and Mount Emu. They went and did a check and
came to the conclusion that due to the lie of the land and the distance, I would be unlikely
to get coverage. Damn. The prices and speeds really look comparable to ADSL.
Returned to investigate Telstra
Business wireless connection, and discovered a little condition on all tariffs:
Customers with 13 digit account numbers are not eligible for this offer.
This seems to be a particularly stupid way of saying “Only applicable to registered
businesses”. My private phone bill does indeed have a 13 digit account number. Took
a look at the online order form, and one of the required fields is “Telstra Account
Number”. What kind of nonsense is that? They also require a company name and
ABN, which I
suppose makes sense, but this sounds like “if you're not a customer, you can't become
a customer”. I suppose I could register the account in the name of LEMIS, but I don't really feel comfortable with that.
It was particularly warm today, and there was almost no wind, ideal weather for spraying
weeds. I didn't really want to: the straps on the spray unit backpack are particularly
uncomfortable. But this kind of weather is so seldom that I didn't really have a choice.
Also did a little playing around with the weed burner I bought last month. It might be good
for stuff like grass growing between other plants, where I can't spray, and where it's
difficult to pull them out. The grass seemed to react well; we'll see.
The issue with this kind of shrub is that little information is available on how to prune
it. It's pretty bare inside, so I only cut it back a little, but it should be enough to
free up the Strelitzias.
I still haven't accepted SkyMesh's notice
of termination, but it's clear I need to do something, especially as the connection is so
terrible. Yesterday I had 22 detected disconnections, the worst ever, and in the first two
weeks of the month it has was 172. And that's only part of the story; in the course of
today I had 144 IRC reconnects, each indicating some issue with the connection.
While looking at the phone tower
page, found a link to Scott's How to get wireless broadband in out the sticks, where he describes pretty much what
I had been planning to do. But he's done it, he has found the correct hardware, and he had
proven that it works here. Did some investigation: as he says, Virgin Mobile offers 8
GB traffic (measured in both directions) for $39. I'm currently paying SkyMesh $69 for 6 GB
download only. For $59, I could get 12 GB from Virgin Mobile, clearly more than I get from
SkyMesh even factoring in the not insubstantial uplink traffic.
What about performance? Scott's page shows the result of a test with speedtest.net, showing adequate speed. In fact, the
downlink speed is slower than my satellite connection (first image), and the latency isn't
spectacular, but it's much better:
But Virgin? Don't they have a terrible reputation? According to their page, APC magazine gave them the “Editor's choice”,
so they can't be that bad. But it was enough reason to go looking, and found that
Internode also offer 3G service using the
same wireless network but their own backhaul. And for $39.95, I get 9 GB instead of 8.
They also confirm that I can get the service, and that it's 2100 MHz (important because of
That's pretty much a no-brainer. Internode may not be as good as they were in the
good old days, when they were a small operation in
York St, but they're miles better than anything else I know. Started to sign up, not
helped much by typical stupid web programming:
Called up the phone sales people, but the consultant said that due to “security”
reasons they required payment by credit card. She couldn't tell me (or didn't want to
understand) why this requirement isn't stated up front: effectively it means that the
monthly charge isn't $39.95, it's $40.94. Internode may be good technically, but this isn't
the first time I've seen the bean counters behave like other companies. Still, there's no
choice, so I signed up anyway and sent a message to Simon expressing my concerns.
The first of our 6 Strelitzia reginae flowers is gradually fading, and none of the others have opened
yet. But they're developing—just completely differently from the first one. The one
next to it is already larger, but it hasn't opened yet:
It looks as if the flowers are fertile, and there's the prospect of more popping up. And we
wouldn't mind having another tree or two. But when should we transplant it? Yvonne suggested as soon as possible, before the roots sank too far into
the ground. She transplanted it into a pot, which in retrospect proved to be a mistake.
It wasn't a seedling at all, but a sucker from the parent plant, and it was joined by an
underground root nearly 1 cm thick. After severing that, the plant looks really unhappy:
Message from Paul Rees in the evening: IPStar have looked at the problem (finally!) and are tweaking some of the link parameters. The
immediate result was an increase in EsN0 from about 8.5 to 9.7. Maybe that will make the
A few days ago Greg Perry sent Theo de Raadt a message claiming that the US FBI had inserted back doors into the OpenBSD crypto framework (OCF). He was bound by an NDA
not to say anything for 10 years, which have now elapsed.
Huh? Can you believe this stuff? The message was intended as a private message to Theo,
but Theo chose (probably wisely) to publish it. Normally
that's not a nice thing to do, but if the software has been compromised, I tend to agree
that it's a valid reaction.
But is it true? So many things speak against it:
If any work of that nature had happened, they would have kept it as secret as possible.
Greg claims that he wasn't directly involved, so he wouldn't have been informed in the
Keeping it as secret as possible also means “forever”. No 10 year limit
here. My guess is that it would run contrary to US law (not that this is a reason to
believe it didn't happen), so they wouldn't want it ever to become public.
The OpenBSD project is renowned for its thoroughness. Theo states that none of the
crypto work was done in the USA, nor by US citizens. They examine all commits
thoroughly, and I'm sure they'd have paid particular attention to such sensitive issues.
One of the people that Greg claims did the work is Jason White, who denies any
involvement. That in itself doesn't mean anything, of course, but he gives details that
are easily checked, and I'm sure the OpenBSD people have already done so. He wasn't
involved in likely areas. He also gives timelines which suggest that Greg Perry had
already left the project when OCF was being developed.
So I don't really believe that there is a back door. There's no way to be sure, of course.
If they find one, I'm wrong. If they don't, that doesn't make me right.
I had hoped that yesterday's increase in EsN0 of the satellite link would improve the
service quality. If anything, they had the opposite effect: on Tuesday I had 22 dropouts,
the highest number ever. Today it was 27:
Date Outages Duration Availability Date
1291726800 9 1310 98.48% # 8 December 2010
1291813200 7 1206 98.60% # 9 December 2010
1291899600 12 1412 98.37% # 10 December 2010
1291986000 18 2363 97.27% # 11 December 2010
1292072400 12 1221 98.59% # 12 December 2010
1292158800 17 1624 98.12% # 13 December 2010
1292245200 22 2121 97.55% # 14 December 2010
1292331600 16 1722 98.01% # 15 December 2010
1292418000 27 3527 95.92% # 16 December 2010
Is that statistically relevant? The availability hasn't been below 97% in months (the
previous lowest was 97.27% on 11 December), and today it was only 95.92%, one of the lowest
of all times. If so, it would point to configuration problems, and maybe there's a way to
solve them. Chris Yeardley is also complaining loudly about her satellite service (from
IPStar directly), as did the Nottles across the
road (with active8me, but also IPStar
satellite). They wanted to monitor things for 48 hours, and maybe they'll change some
things during that time, so I won't complain yet.
My upcoming 3G connection is another issue. I told Callum Gibson about the fact that the
local tower would have a 2100 MHz connection, but he didn't believe me, and went to check
for himself on the Optus web site. Somehow he managed to mistype the street number, so we ended up with
two different statements. For our house, the PNSA 03ROKEWOOD has 2100 MHz, and
down the road it's 900 MHz:
It's difficult to guess where 03ROKEWOOD is, but in all probability it's the one in
Corindhap, assuming that it is
already in service. That would place it about 10 km south of our house. 24 Kleins Road is
almost exactly in the other direction, 250 metres away. Even if the tower handles both
frequencies, you'd expect that information to be the same for both addresses.
Called up Internode, who told me that
they couldn't give me any details about the tower, and not to worry. I explained that the
modem I had only did 2100 MHz, but that didn't seem to sink in. Spoke to the supervisor,
who was a little more understanding and went through and confirmed that he got the same
results, but suggested I should talk to Optus, since it was their information.
Did that, and was told that the software was probably correct and that I shouldn't be so
negative in assuming a bug. In any case, if I signed up with Optus I'd get a free modem
which can handle both frequencies, and there's a 14 day money-back guarantee, so what's the
Finally sent a message to Greg, the project manager who visited us two years ago. He replied saying that yes, no problem
with the info, but he's out of the office until Monday. Hopefully that will clarify things.
Yvonne finally located some wood for the smoke oven:
surprisingly unevenly chopped chunks of Hickory imported from the USA, the princely sum of $12 for 1.8 kg of it—or, as
the volume-obsessed Americans write, 360 cu. in. How can you measure something so chunky by
volume? It seems, though, that I only need one or two pieces of wood per operation, so we
can survive that.
Put the thing together without too much difficulty—the instructions are surprisingly
good, hampered only by being out of date. The quality is poor even for ALDI standards:
Woken up early this morning first by the sounds of horses galloping around and then round
6:15 by one of our neighbours, whose name I forget (might be John). One of their fillies
had not only broken out of their paddock, but into ours. That explained the
galloping. He couldn't stay, had to go to work, and I promised to get it out for him when
we got up.
That didn't happen, though. The horse broke out again, and we couldn't find it. Yvonne assumes that it went back home.
As if that wasn't enough, got a call from Nele Koemle, who is agisting some of Chris
Yeardley's horses, and one of them injured itself. She, too, had to leave for work, and so
Yvonne and Chris set off for Leslie to attend to it. Fortunately it wasn't badly hurt.
And while I was doing something completely different, found a cat in the garden:
Chris Yeardley mentioned recently that she, too, would like to change to 3G for networking,
and that the Optus web
site gave her the frequency 900 MHz. This really puzzles me. Was the Optus
consultant correct when he said that the application is clever enough to give different
results for each address? The Nottles' house was number 50, so tried that. And yes, I got
yet another result:
So, like here, they are supposed to have 2100 MHz coverage. But unlike anywhere else, only
one of the three services is available. Potentially that could even be correct: if the
tower is where I think it is, we have a fairly clear view, while the Nottles have trees
immediately to that side of the house. But the effort required to collect that sort of
information would be enormous. We'll see what the truth is when I hear back from Greg.
The IPStar link continues to be terrible.
I beat even yesterday's record 27 outages: today it was 30. But there's something funny
about it: my TCP connections don't always drop. In the past any modem reboot would
automatically drop all TCP connections because of the BST protocol they wrap around it. Now
it doesn't. This suggests to me that we're looking at some different problem. Anyway, if
we can get the 3G connection working reliably, it'll all be uninteresting.
House photos again today, and this time managed to mount my camera on the panorama bracket
so that I could get more of the roof of the verandah in the verandah panorama. Result:
worse than before. Here two weeks ago, and then today:
Today's panorama was clearly worse than two weeks ago. Spent some time investigating with
manual control point settings, but again didn't get much of an improvement. One thing that
became evident is that Hugin didn't
create many control points between the upper and the lower layer of the panorama. In the
process, found yet another source of potential parallax, so maybe I should rethink the whole
I'm gradually getting some information about smoking food. One of the things most people
agree on is that most smoking is done at temperatures between 110° and 120°, though in some
cases you can go as low as 60°. But I can't get any of these temperatures with the oven:
the flame is far too hot even on the “low” setting:
There's a vent at the back, but it doesn't seem to make any difference in the temperature.
So: what to do? There must be a way to reduce the flame. Most gas burners have an
adjusting screw for the “low” position, but this primitive burner doesn't.
About the only alternative I can see is to control the flow of gas to the oven on the high
pressure side. Fortunately I have an adjustable regulator, so maybe that's an option. But
of course the fittings are all wrong, so there's not much I can do there either for the
moment. I'll have to go into town on Monday and get the fittings then.
The petunias I planted yesterday showed one thing: they had been in desperate need of
planting into something bigger. The punnets are completely full of roots. We've been
planning to put them in hanging baskets—for some reason, probably snails, they don't
do well in the ground—but we didn't have enough baskets free. But then we have two
pots with wild strawberries in them, and they're looking less than happy:
So: big reshuffle. I transplanted the strawberries from one of the pots into the vegetable
garden, in the process also planting some Nicola potatoes that Yvonne had bought at ALDI, and which
she prefers to the ones we can normally get, and also some basil, two kinds of lettuce and
dill. The other pot will stay where it is until we're sure that the ones in the vegetable
garden will do better. Then I was finally able to plant the petunias. We now have three
baskets waiting for the plants to grow a bit:
The snails are a real problem this year, almost certainly because of the continuing wet
weather, and the snail bait I put out seems all to get eaten quickly. There must be
hundreds of them out there somewhere.
Chris Yeardley had a birthday a couple of days ago, so for tonight we cooked a paella valenciana. One of the things that
most paella recipes get wrong are the quantities of saffron. Most English language recipes
require “a pinch or two” or and one just writes “saffron”.
“The Joy of Cooking states “a small quantity, or up to two teaspoons if you are Spanish”. The
only one I found with a specific quantity was Bonniers Kokbok, which specifies 0.5 g saffron for 3 dl of rice—however much
that might be. But the quantity sounds reasonable.
It's an important consideration. Here in Australia, as in other countries I know, saffron
is sold in quantities of 0.1 g, usually in packs of 4 for about $4, but I've seen
packs of 0.1 g go for $11. My recipe
(larger than Bonniers) requires 1 g, which could cost up to $110 if you buy the expensive
The problem, though: I think we didn't have enough saffron. Was it just the quality of the
saffron we were using this time, or do we need even more?
We were celebrating Chris' birthday, but somehow the discussion took on a particularly
morbid tone. Part of it was definitely the fact that Sue Blake is in hospital dying, and is
unlikely to come out alive. We've been watching her decline for over a year, since she
told us that she had
On IRC today read a message from Juha Kupiainen suggesting that Sue Blake died yesterday.
But nothing certain. He called the hospital and was told to contact the relatives, but we
weren't able to do that by the evening. Things don't look good, though.
The satellite problems continue unabated. Today I had a new record, 32 disconnects. But as
I've been noting, there's something different about recently: it happens at specific times,
notably when I'm uploading data. Once a day (theoretically) I run rsync against my web photos, currently about
115,000 files. Today I had to run it 4 times before I got beyond this point:
building file list ...
112000 files to consider
The number counts up, but on each occasion it stopped at 112,000 files. Running
a ping -A showed that the connection effectively died, then about 20 seconds
later the modem noticed it and logged off:
The interesting thing about this image is that it still thinks that it's transferring at 110
kb/s. And I've seen that before: something in high speed
uploads causes the link to drop. I even have this in my sync scripts:
# Uncomment this if we run into problems with IPStar
Time to uncomment and try again.
While looking at that, also saw another issue that looks
... there is a possible fault with their ground stations which is causing services to
randomly disconnect thus forcing a \221log off\222 on an active connection.
Still, when I finally get a 3G connection, all this should be history.
One of the “features” that Olympus introduced some time ago were the so-called
“art filters”, a kind of in-camera processing of the images as soon as they are
taken. I've always thought the idea silly, and I've never taken any photos with the
feature, but a number of people on forums are really excited by the results, so today I
tried them out on my E-30. They don't make things easy: there's almost no documentation. Anyway, there
are 6 different filters, entitled “Pop art”, “soft focus”,
“pale&light color”, “light tone”, “grainy film” and
“pin hole”. Here are the results taking a photo of the garden. A normal image
looks like this:
The art filters make this out of them (unretouched). The first image is in fact the
identical same photo as above; the art filters are applied to the JPEG image only, and the photo above comes from
the raw image.
But how did it get such an exact registration? The buildings in the background show that it
was taken from almost exactly the same angle. I had thought I had taken it from my hotel
room, but the continuation of the film show that it was taken from one of the lifts in the
Pan Pacific hotel; thus the
horizontal but not vertical alignment.
This spring and early summer has been amazingly cool, and today was a good example. The
maximum temperature was only 17°, but it rained much of the day, and the daytime
temperatures dropped under 10°:
Pruned the Begonia that Yvonne bought at the end of October.
They say you can propagate them easily, so put some of the canes in a pot of water to see
what happens. Yvonne did the same with some of the leaves.
One of the other Strelitzia
reginae looks like it's going to bloom soon. But the shape is very different. Is
this a double flower? The other flower had nothing resembling the right-hand part of the
More playing around with the smoker today. Discovered that if I filled the water pan with
boiling water, it would reduce the temperature in the oven to about 115° to 120°. That's
still too warm, but good enough for some experiments. Tried smoking some beef and chicken
breast. After 1½ hours the temperature in the chicken had risen to 60°, but in the
following ½ hour it only rose to 62°. For some reason I decided to aim for 76°, so in the
end removed the water, turned up the gas and let the temperature go up to 180°, which
Certainly something that can be improved on, but for a first attempt certainly acceptable.
The temperature in the chicken had risen to 78°, which was definitely too high. Next time I
do a breast by itself, I'll aim for 70°. The beef came out at 72°, which was quite
Mail this morning from Mary Davies, a friend of Sue Blake. Sue died in hospital at 1:00 on
What can I say? We've known that this was coming for well over a year, after she explained
her blog posting of
10 September 2009 on IRC. It was sad to see her decline, but also somehow
a privilege to experience the process. And now it's over. Should we be particularly sad
right now? I've been through this before when my father
died. It's not the moment of death that is the issue. I've updated her blog, which she wanted to live on, and Mary is going
to send me some stuff to put up there. And it's interesting to note on the side that she
beat the odds by 2,352 hours.
Mail today from Greg, the Optus project manager, addressing my concerns about the accuracy
of the coverage information on the Optus web site. Despite
what the Optus telephone consultant said, I was right to be suspicious:
The proposed site in Rokewood is not built yet, and won't be built until next year. It
will be located at Crn of Dark Lane and Reservoir Rd (I believe also known as Halls Hill
Rd) 1.6km north of Rokewood. The site will run on the 900MHz band (UMTS/3G).
And the site told me that it was serviceable and running at 2100 MHz:
He's enquiring about what went wrong, and will get back to me. One possibility is that the
PNSA is only a reference point, and that maybe I do have coverage at 2100 MHz, though the
nearest site is some distance away.
What do I do now? The USIM from Internode should have arrived today, but they decided to require a personal signature for it (for a
$10 USIM!), so I can't pick it up until tomorrow. But it looks as if I'm going to have to
buy new hardware, possibly what Scott Weston has done. At least we know that that works. But he has bought a router
that requires a separate modem, and that makes it very expensive, well over $200 for the
Europe seems to be in the grip of a particularly cold winter. The news gives reports of
heavy snowfall in the UK, France, Benelux, Germany and... Victoria. But it's summer
here! Admittedly, snow fell only in “Alpine” areas (a strange term for a state
whose highest point is 1,986
m), but it's indicative of an unusually cool summer. Today was a little better than
yesterday, but we're still getting above-average rain and well below-average temperatures.
At least we haven't seen any snow around here.
Into town today, mainly to pick up the USIM card from the post office. Also dropped into Whiteheads to get some timber for
mounting the shade cloth, and to Formosa Gardens, where in the end I only bought fertilizer.
It's fairly clear in advance that the Optus towers in the area will only deliver 900 MHz, no matter what their coverage tool says, so
while in town checked at OfficeWorks to
see what kind of 3G routers they had.
Nothing with modem, as I had almost expected, but lots of buzzwords: “Do you want 54 G
or 150 N or 300 N?” What does that mean? It seems the numbers were Mb/s, clearly
nothing to do with 3G wireless. Then it dawned on me: 802.11g and 802.11n. The sales person had never heard
of 802.11, of course; they've hidden it behind this horrible buzzword Wi-Fi. And it took me some time to find the
terms 802.11g and 802.11n on the package of one of the routers.
They didn't have any modems at all, so that killed the idea of buying one just in case and
only opening it if I needed it. On to the Optus shop in Bridge Mall, where they couldn't
sell me a router, just a USB stick. I suspect the salesperson had never heard the word
He suggested I went to Dick Smith, where
allegedly they had that sort of thing. All I found were USB modems and a sales person
talking at length to a pretty customer trying to choose a cover for her mobile phone, so off
without any further enlightenment.
That's the card itself in front, mounted in a credit-card sized frame. Does any compatible
equipment still use that format? There was also a single sheet of paper with configuration
details. No stupid installation software like Telstra's abomination. In any case,
took the card, put it in the router, edited APN, user name and password, and pressed
the Connect button:
21-12-2010 13:48:43 > * gr0Ogle returns with a shiny new USIM.
21-12-2010 13:49:11 > gr0Ogle: Oh.
21-12-2010 13:49:15 > gr0Ogle: Two bars signal.
21-12-2010 13:52:34 > gr0Ogle: Dec 21 13:52:02 pong-gw chat: send (^M)
21-12-2010 13:52:34 > gr0Ogle: Dec 21 13:52:02 pong-gw pppd: Connect: ppp0 <--> /dev/ttyUSB0
21-12-2010 13:52:34 > gr0Ogle: Dec 21 13:52:11 pong-gw pppd: Could not determine remote IP address: defaulting to 10.64.64.64
21-12-2010 13:52:34 > gr0Ogle: Dec 21 13:52:11 pong-gw pppd: local IP address 184.108.40.206
21-12-2010 13:52:38 > gr0Ogle: Dec 21 13:52:11 pong-gw pppd: remote IP address 10.64.64.64
21-12-2010 13:52:41 > gr0Ogle: Dec 21 13:52:11 pong-gw pppd: primary DNS address 220.127.116.11
21-12-2010 13:52:44 > gr0Ogle: Dec 21 13:52:11 pong-gw pppd: secondary DNS address 18.104.22.168
21-12-2010 13:52:58 > gr0Ogle: Amazing.
The whole thing took less than 4 minutes, including commentary on IRC. And I was connected!
It would be nice to say that that was the end of my problems. In fact, it was only the
start. As we had all expected, the connection was at 900 MHz, and all I got was GPRS. That gave me ping times in the
same order of magnitude as satellite—as long as I didn't try to transfer any
significant amount of data. When I did, I got things like:
21-12-2010 14:03:19 > gr0Ogle: 64 bytes from 22.214.171.124: icmp_seq=19 ttl=57 time=8608.959 ms
21-12-2010 14:03:19 > gr0Ogle: 64 bytes from 126.96.36.199: icmp_seq=20 ttl=57 time=7631.864 ms
OK, I had it connected to an 850 MHz antenna. That would help at 900 MHz, but probably not
much at 2100 MHz. Tried replacing the external antenna with the toy antenna, with the
result that I didn't get any signal at all, neither to the east of the house (which I
suspect is the direction of the 2100 MHz tower) nor to the south. So that's out. I need
Spent the rest of the afternoon looking for hardware. Why is it so complicated? First
there's the issue of compatibility between the modem and the router, a problem you don't
have with an integrated modem. Ran into an amazing number of problems finding good
I couldn't find any summary of what's available. Took a look at the Broadband Choice pages on Whirlpool and found lots of
pages, all almost completely incomplete, and no overview at all. That's completely
useless for finding hardware.
Took a look on eBay and found a number of devices tagged at “3G routers”,
all with one thing in common: no 3G connectivity.
It seems that there are about 3 or 4 potential candidates in the range $50 to $100.
TL-MR3220 seems to be the cheapest, and the specs are completely adequate for
what I need. There's also a model TL-MR3420 with 802.11n 300 Mb/s instead of 150
Mb/s. But the only review I could find about it was very negative, and the unit had
been sent back. It seems that the connection was very unreliable. Not the sort of
thing I want to worry about for a few dollars.
The ASUS RT-N13U, which includes a printer server. The documentation is
available online, nicely prepared, and incomplete: although the specs include 3G
modem support, they don't say how to use it. The support was added in revision B of
the machine, so hopefully it's just out of date. There are also some strangenesses:
it only has one USB port, for example, though they talk of connecting more than one
device. No mention of having to buy a hub for it. But the printer support is only
for printers from Canon, HP and Epson, none of which I have, so the printer port is
not of great interest.
The D-Link DIR-412
Wireless N 150 3G Mobile Broadband Router. It's closer to what I need, just
an Ethernet port, the obligatory 802.11n network and a USB slot for the modem. But
it's actually more expensive than the other two. It's also slower than the ASUS
device, though that's really not of interest to me.
The router is only half the question, of course. Which modem? Again, Whirlpool was
useless. I had heard that the Huawei E1762 was a good choice, but the Whirlpool
page contains almost no information.
The manufacturer's web site was no better. It gives a list of 32 devices, 20 per
page, starting with the oldest. That's certainly a case where chronological order
doesn't make any sense. All the current models are on the second page. Did I say all?
No, the E1762 isn't there at all. The links from Whirlpool are also invalid.
With some effort managed to establish that it's considerably faster than the other
model, the E160, which doesn't do HSUPA, and that there doesn't seem to be much else available.
Finally decided to go with the ASUS router and the E1762. The E1762 may not be listed on
the Huawei site, but it is listed as compatible on the ASUS site. Bought both on
eBay, and was greeted by another surprise in checkout: the listed postage of $9.95 had
suddenly become $12. Sent a message to the seller, who responded quickly with an amended
bill, conveniently requiring me to re-enter all the details that normally get filled out
automatically. Still, it's done: I have ordered a modem and router. What about an antenna?
That'll have to wait until tomorrow: it was 18:00 before I finished this stuff.
Finally the weather's getting more summer-like, and this morning CJ came over to clean the
gutters (since my little accident five years ago,
Yvonne doesn't want me to climb ladders). We had also
intended to finally put up the shade cloth, and the weather was right, but somehow I was too
tied up with the networking stuff, so in the end CJ went off to make some hay.
So now I have a 3G modem and a router on their way. Yesterday I had received a mail message
from the seller of the modem, asking if I wanted a (free) USIM, and of course I replied
“yes please”. Today I got a phone call: the reply hadn't arrived. Why is
eBay such a pain? Anyway, it was nice of him
to go to the trouble to call me.
But I still need an antenna. Which? Scott Weston used a 22 dBi Yagi which cost $148.90. Or did it? I had more or less decided to buy one
when I discovered that I would need to pay at least $20.90 just for a cable (only 3 m long
and with the wrong connector) and another $16.50 for an adaptor to connect to the modem.
The 10 m cable I would need would cost $30.80—a total of $47.30 on top of the price
just for a cable!
That's not the only game in town. RFShop have a 11 dBi Yagi complete with 10 m of (correct) cable for “$74.00”, which
proves to be a price without tax, not a thing we do in Australia. So the real price is
$81.40. But that's only $34.10 more than the price of the cable alone from City Technology.
Clearly an 11 dBi antenna is not nearly as good as a 22 dBi antenna. This one has only 5
directors, while the 22 dBi one has 13. Clearly the latter is more expensive. But is it
necessary? I'm getting reception now with a rod antenna, and the little Yagi will almost
certainly be enough. If not, I can buy just the 22 dBi antenna, use the existing cable, and
be only $34 out of pocket. So I ordered one.
Where should I mount the antenna? The current position of the rod antenna isn't ideal:
It occurred to me that I have a second satellite dish on the top of the house, for some Pay
TV thing that the previous owners used. I never had any intention of using it, and it's
still there only because I saw no reason to remove it. But it has the ideal mount for a
Yagi antenna, and it's even on the correct side of the house. It's the one on the right
here, seen from the south:
As if that wasn't enough, that's my office to the left. All I need is a hole for the cable
under the eaves and I can go almost directly into the office. There's a small externally
accessible cupboard in between, but that's not a problem. Took a look inside and found a
cable without any markings, but which could be RG-58, coming out of the ceiling and going
down through the floor. At first I thought it was from the dish, but that cable runs
elsewhere. I wonder what it is.
Even with GPRS, of course, I have
one great advantage over satellite: reliability. And one of the most frustrating things
I've had has been the continual IRC reconnects. I'm running a proxy on the external mail
server, so I don't actually lose anything with the dropouts, but I've been having up to 90
reconnects a day, and the client shows me status windows, not data windows, so I have to
change windows every time. Now, though, I can set the route to the mail server via
GPRS—speed isn't important—and leave the rest to go via satellite.
How well does it work? As expected, no dropouts any more. What I didn't expect, though, is
that I didn't get any satellite dropouts either. The last one for the day was at 05:50:39.
That could suggest that the problem was with data to the mail server, but that looks
unlikely. I pull mail down every 15 minutes, and it takes about 1 minute on average. This
last one finished at 05:45:38, well before the dropout. So I suppose it's only a
coincidence. And the lack of reporting of the problems is only a partial indication: I had
several dropouts on an ssh connection during the course of the day.
My hardware is starting to arrive: this morning I found a Huawei E1762 USB modem in the
letter box, clearly less voluminously packed than the USIM card the other day. There's not
much I can do with it until I have the antenna, which is unlikely to arrive this week, so
I'll have to wait for it at least until Wednesday. But that doesn't mean I can't play with
Documentation? What's that? There's a 16 page booklet with precious little information, a
diagram showing the orifices on one page and a description of them over the page. But it's
another device with an ISO 9660 CD-ROM image on it, and even Apple complained with many
instances of this in the log:
SAM Multimedia: READ or WRITE failed, ASC = 0x20, ASCQ = 0x00
It proved that the image contained 15 MB of files. I don't know what they are, but
none of them are documentation. I know from others that this thing has AT-style
modem commands. But where's the documentation? It seems that you have to reverse engineer
things to get any information at all.
I installed the software on boskoop, my Apple. Not surprisingly, I couldn't get it
to work. I had a new USIM card, but I don't want to activate it now, and it looks as if I
don't have any signal anyway. But why are these toy “Applications” so
primitive? Modern routers have web servers that present most of the information you want,
not necessarily as well as I'd like, but an order of magnitude better than these toys.
More investigation of the satellite connection. It really is better than it has been, but
not as good as I had thought. My scripts only register an outage when I can't access any of
the 5 systems in my list. And one of them is w3.lemis.com, which I'm accessing via
GPRS. Thus no outage reports at all after I set up the link.
But even when I corrected the information, there were fewer outages than before:
Date Outages Duration Availability Date
1292677200 32 3056 96.46% # 19 December 2010
1292763600 23 1598 98.15% # 20 December 2010
1292850000 16 1628 98.12% # 21 December 2010
1292936400 7 459 99.47% # 22 December 2010
1293022800 7 626 99.28% # 23 December 2010
Why? Who cares. Let's get this 3G thing flying, and forget the satellite.
Mail from Scott Weston, who had been travelling. Most of my questions have been overtaken
by events, but he did manage to point me at a web
site with a graphical representation of the towers in the area. It's particularly
difficult to interface to, but I was able to establish where the “local” towers
are. The closest one with 2100 MHz is apparently in Ballarat. More likely towers are located
in Cressy (presumably
the one Scott uses) and Willowvale, both in a direction accessible from my mast. For places
like Ballarat I'd need to raise it considerably.
Somehow this networking stuff is keeping me preoccupied, and I haven't been doing much
else—not even photography. The garden is in almost its usual state of neglect, but I
did get round to attacking some weeds, and it's even visible.
Which mobile tower gives best connectivity? Spent some time looking at the Spench web site looking for local ones, and far too
much time trying to enter them on a Google map, not helped by the fact that the thing always looked for the closest road.
To be finished.
Note in the letterbox this morning that I had a packet to be picked up at the post office.
That could only be the ASUS RT-N13U
router. Was it worth it? On the one hand, I really needed the antenna to be able to
do anything useful. On the other hand, today was Christmas Eve, and the way the feast falls
this year, we'll have Monday and Tuesday off, and I won't be able to pick anything up until
Wednesday. So I decided to pick it up anyway.
At the post office, asked on the off chance if the antenna had arrived. To my immense
surprise, it had. So it was really worthwhile: now I have all the equipment I need for the
On the way home, for the fun of it, went to take a look at the towers in Smythesdale and Willowvale. The first
was pretty much what I expected, but my GPS navigator had difficulty getting me to the
second one, and unfortunately I hadn't taken my long telephoto lens with me:
As near as I can tell, the Willowvale tower is built on top of a high voltage pylon. It
interesting that you don't hear much in the way of complaints about pylons (any more,
anyway), though they're much more obvious and have real dangers associated with them.
Back home and took a look at what I had. The router left a mixed impression. On the one
hand, it states on the box:
That's certainly an improvement on what Telstra do. It also includes an afterthoughtsheet of paper with
instructions on how to set the router up as a 3G router. It requires first installing on a
Microsoft box. I don't know if that's just to ensure that you can localize any problems
that may occur, but I've heard rumours that the Microsoft installation reconfigures the
stick, and that you can't use them on FreeBSD without first installing on Microsoft.
On the other hand, it comes with a “Quick Start Guide” that really contains
almost no information, not even the default IP of the box. It describes the switch on the
bottom, with positions Router, Repeater and AP, stating to make a
choice, but not describing what the positions mean. Fortunately, I've already established
that there's a good manual, but who can use the quick start guide?
The antenna came with two cables, not the one I
had expected: a normal 10 m connector with female connectors at each end, and an adaptor for
the USB stick. The female connectors are clearly of the same kind as the ones on my whip
antenna, which has a male connector, so tried to join them together. No joy: the length of
the connector is different for some reason, and though the threads are compatible, I
couldn't get them close enough together. Still, for playing around with, found a solution
to keep them together, and also supply continuity for the shield:
Installing the software from the USB stick was surprising: it was “already
there”. I can only assume that the Telstra software I installed a couple of months ago also installed software for
the Huawei, though that has nothing to do with the equipment I got. Removed the stuff and
reinstalled it, despite a flurry of popups from bottom right on the screen telling me that
it had found all sorts of strange hardware, and that it wouldn't work. Looking at the
“Control Panel” showed indeed that there were mobile devices, CD-ROMs, disks and
modems. And the detailed install dialogue showed that something like documentation must
have got installed as well. I sometimes wish I understood this Microsoft environment
Finally it was installed, and I tried running it. It asked for an SP Code, as you would
expect, but that was all. No user name, no password. But what I did get was a couple of
Where did they come from? The status light showed that there was no signal, so they
couldn't have arrived out of thin air. Were they stored there last time I tried to connect?
Are they maybe there on the device? Certainly the date of the older one is anything but
But what does this have to do with setting up connections? Was I connected? How do I find
out? The status page doesn't say anything about that; the information is encoded in some
silly undocumented icon that I've never seen before, but which after some research proved to
say “no, not connected”.
Every time I work with Microsoft-based software, I'm baffled by the non-sequiturs I find.
And how do you set the user name and password? Why does it try to connect before it's
configured? Why is this software so completely different from the software that the same
device provides for Apple?
Finally found something in the “Instructions”: you select the menu
sequence Tools, Options, then Profile Management and Edit.
There you can enter your user name and password, which it promptly swallows and pretends you
have never entered anything. This image was taken after setting user name and password:
But after that it connects, and the LED colours on the device showed that it was running
with UMTS, which it calls WCDMA.
So even with this primitive antenna connection, I was able to connect to the net. It looks
as if I'll have no trouble at all with the real antenna.
Today the Geelong Regional
Libraries mobile library came along and brought me a couple of books about smoking.
As I imagined, hot smoking is done between 60° and 110°; even the lowest temperature of this
ALDI smoker is closer to 120°, so clearly
it'll have to go back. Cold smoking is done below 37°. I still need to find out how to do
that, but one of the books offered some suggestions.
The Alstroemeria that we
bought two weeks ago didn't make it back home unscathed.
A flower stem broke off, and we put it in a vase. It's gradually past its use-by date, but
that's been 12 days, and at least one flower still looks happy:
It's Christmas again, and somehow we're less and less interested every year. This year we
(mainly Yvonne) decided against turkey, and we did a big
chicken instead, filled with the same
Somehow things didn't want to work. I had started making sourdough in the morning, so the
mixer bowl was in use, and I ended up putting the stuffing through the mincer. Managed to
drop an egg shell in the mixer, and had various other problems with what should be a simple
dish, and finished in quite a bad mood. Yvonne didn't fare
much better: somehow the cream got spilt in the fridge. Not quite what we planned.
The chicken also took longer than expected. My cooking times called for 88° in the breast, which should happen after about 100
minutes for a 2 kg chicken. But by that time we were only barely at 82°. Checked and
discovered that the all-important upper leg joints looked good. Clearly the filling was
lowering the temperature of the breast in relationship to the legs, a good thing. We should
fill chickens more often.
It's still not perfect, but it's very difficult to see any of the jaggies.
What's the difference? I think it's because I levelled the pan head as well as the ball
head before taking the photos. Clearly the pan head rotates around a vertical axis, but
that doesn't stop me from tilting the pan head forwards or backwards, and that brings the
entrance pupil away from the axis. This time it stayed in the right place.
Into the office this morning to discover that the NetComm router had lost the connection to Internode. No messages, good signal, just no network connection. Rebooted the
machine and it worked again.
High time to install the antenna and the router. The antenna was easy enough to
install; what I hadn't reckoned with was the difficulty of getting the antenna cable through
the walls, since there are two parts to the wall with a cavity in between, and it was
difficult to aim for the second hole.
Apart from the multiple ^M characters, which Microsoft users don't see, what's this
nonsense with one default gateway per interface? How does that work? My experience was
that it doesn't.
Then continued with the ASUS RT-N13U
router. What a catastrophe! I've complained about stupid hardware in the past, but I
think this takes the cake. I ran into one problem and irritation after another:
First I tried to set the IP address of the device to my network. But that's not allowed:
For reasons that make no sense at all, it insists on an RFC 1918 IP address. That's almost
enough reason to return it by itself.
Well, the thing runs Linux, so maybe there was a back door to the config information.
Just look for files that have changed recently and look at what's inside. No ssh
connection, just telnet. I can live with that, but it doesn't sound very
professional or up-to-date. It proves to be another BusyBox implementation. But the utilities!
No times! Why not? Even the NetComm
software has that. And there's no -R option to ls:
# ls -alR / ls: illegal option -- R
BusyBox v1.12.1 (2009-12-30 16:46:58 CST) multi-call binary
Usage: ls [-1AaCdilnsTwxk] [filenames...]
No find, almost no tools. Went looking in likely places, and found a
directory /var/state/parport/svr_statue. On IRC we weren't able to work out
whether they had meant /var/state/parport/svr_status
Gave up on that for the while and tried setting the system log to log remotely. I
couldn't enter the host name. The input field was too short:
Clearly it wanted IP addresses. But why? Finally set the address and rebooted the
box—another Microsoftism—and got no less than 10 LOG_EMERG
messages. Why do these router desingers so love LOG_EMERG?
“Emergency! The router has booted!
After reboot, connected to the modem again, and got another silly message:
Gave up on any kind of sensible configuration and turned my attention to the sheet of
paper telling me how to configure the modem. It was really very brief: “Insert
modem and select Advanced Setting, USB Application and HSDPA:
Enable HSDPA, select the modem type, ISP, APN, dial number (what's that for? Anyway,
Internode supplied one), user name and password. Did that, and discovered that, despite
documentation to the contrary, my modem wasn't listed. Selected the E176, which is
hopefully similar, pressed Apply, and once again it took an eternity before
returning—with a number of the parameters reset to the default values:
Clicking on the links just took me around in circles. The instructions suggested a
firmware upgrade if there are problems, so went to the ASUS site, which cleverly ensures
that you can't refer to their content, and checked. Before getting any information, you
need to select the operating system. But BusyBox isn't on the list. Selected Microsoft
“WinXP” and found no less than 15 firmware versions. Mine is 188.8.131.52
Beta software without warning in a device of this nature? They should be ashamed of
themselves. But I repeat myself.
Downloaded the firmware and tried the instructions to install it. It hung.
Gave up for the night; this stupid “easy install” router had kept me busy all
afternoon. It looks like the others on IRC were right in asking why I wanted one in the
first place. If I had known what I was up against, I would never have bought it, but now I
want to see if I can get it to work. Others on Whirlpool have reported success with exactly this combination. Is this really what networking is coming to?