Today my satellite connection (apparently) changed from Aussie broadband to SkyMesh. No interruption, which of course also means
no change of external IP address: it's still 121.200.1.204, and it still resolves
to 121-200-1-204.cust.aussiebb.net.. Tried checking the My SkyMesh pages, and was refused login. Sent off the
“Forgot your password” form and was returned a message sent in only HTML,
which promptly landed in my spam folder. It contained an old user name and password, one
that I haven't used for over a year—and it didn't work either. Tried to fill out the
support application form—how I hate web forms—and got a message rejecting
the message because it wasn't sent from my SkyMesh mail address. Hopefully they're not
going to expect me to read WebMail!
Finally called up the support number, 1300 759 637, and spoke to Joel, who confirmed that
they were having trouble transferring the account details. You'd think that they'd have
that sorted out in advance. In any case, it looks as if I'm going to have to do without
access to “My SkyMesh” for a day or two. At least the network connection is no
worse than before.
More playing around with the RSS code today, and changed pubDate to set different
minutes for each item. We'll see if that makes any difference; at the very least it will
cause yet more entries for some people, something that hasn't gone unnoticed. Mitch Davis
contacted me mainly because of the greenhouse, but he also mentioned the number of reposts.
Norbert Wigbels also contacted me with another issue, one that I can't completely fix: the
RSS feed includes images. In the original PHP version of my diary, you can click on them
and get progressively larger versions of the image. That doesn't work in RSS, because the
aggregator stores a static copy of the page. So I do what I can and instead link to the RSS
feed itself. This isn't ideal, because it's a lot larger. To add to the problem, and
something that I can and should do something about, it seems to lose the position in
the page and positions people at the beginning, though most RSS users read from the end.
But that will have to wait for a larger overhaul of the showphoto function.
And then Alistair Hogge told me that he was having severe problems with the feed:
No idea what happened there; the RSS validator was happy with the feed, and it looks right on the right-hand side as
well. The browser was Konqueror, and maybe it has something to do with that.
Yesterday's problem building FreeBSD didn't
seem to be so serious, but today Yvonne told me that she
couldn't start firefox on her machine. She got an incredible number of identical
messages, followed by a bus error:
Fatal error 'Cannot allocate red zone for initial thread' at line 384 in file /usr/src/lib/libthr/thread/thr_init.c (errno = 12)
What went wrong there? My best bet was that it was an attempt to install VLC a few days ago; it seems to have installed new
libraries. Yvonne's version of firefox was also ancient, so set to installing new
versions of that, running into dozens of conflicts along the way. I really should reinstall
the whole system, which is what I had intended with the kernel build, but it looks as if I
first need to debug make. What a pain!
The sun was shining again today, something we haven't had for a few days. Did some thinking
about the critical panorama from the verandah. Unlike the others, I hadn't tried an HDR
version, but today it sounded like a good idea, so tried it out. It's the second of the
images below:
On the face of it, it looks a little better, in particular the part of the garden in the
middle of the image. But exactly there is the problem. I take these photos with flash to
illuminate the foreground, and I need full power, so it takes about 5 seconds to recharge
the flash. And in that time, things move:
Decided to postpone the remainder of the photos until sidereal noon, which improved the
remaining panorama, but made some of the other photos far too hard. Here the same photo
taken last week and this week:
For some time I've had a suspicion that something was wrong with the hoses on my beer kegs.
Today I found out why: Piccola has been biting
them. Chased her away today, but came back and found the external hose well and truly
perforated. Damn! When I replace it, I should put a bit of old garden hose around it.
A few months ago we planted some plants
in boxes in the kitchen, in a place where there's very little light. Initially they did
well, but it didn't last. Only
the Spathiphyllum seemed to like
the light levels. The Kalanchoe lost
all its leaves, and though new ones have come, it's looking anything but happy. Here's a
photo taken 3 months ago, and then today (with the box turned the other way):
Also took another look at
the Haworthia, which has been
recommended as an interior plant. It flowered a while back, and I didn't cut off the flower
stem. A good thing too: it has developed a small bud:
Years ago Yvonne wrote down a recipe for a „Curry-Pirogge“ in her
handwritten recipe book. It came (apparently)
from the German magazine „Stern“, and
from time to time we cooked it. It's been a while, though, so it sounded like time to do it
again. The recipe we have is basically a giant curry
puff, and we decided to use the same filling for it. The results were mixed: it's
supposed to be a straight roll, but we ended up with something too long to fit in the oven,
so it ended up semicircular:
What does „Pirogge“ mean? We
had always thought of some kind of boat, but that's
„Piroge“ (English
“Pirogue”). The English word
for „Pirogge“ is Pierogi.
It's of Eastern European origin, and it seems that it's really a kind of dumpling or pasty,
more like a curry puff than the recipe we had. It's also supposed to be made with
unleavened dough, but that wouldn't work for the enormous thing that we made, so our recipe followed the original recipe and used a
bread dough with yeast.
So, my RSS script changeover is complete—and I am still only getting a single entry
per day from ACM Queue. That's clearly some
limitation of their grabber; others, such as Google Reader, get it right. I'll stop playing with it for now.
So I've more or less come to a halt trying to get lagoon upgraded. I can't build a
system because of the make bug. I can't upgrade ports because everything is so out
of date. And then Yvonne came and told me that she couldn't
start firefox or even
an xterm: my attempts to upgrade the ports had somehow messed up the libraries. What
a pain! And how fortunate that I had a backup of the system. Restored /usr/local
and all was well. But I really must find out what's wrong with make.
We're into the last month of autumn now, and the weather has been unseasonably cool. That
hasn't stopped the weeds growing like wildfire, and it's also clear that it's about time to
do some restructuring work. Did some weeding and considering what to do for the spring. In
the immediate future, as soon as the weather allows, we're going to have to spray the weeds,
and it also looks as if we're going to need a new ride-on mower: this little one I have now
is good for detail work, but it's enough work that it keeps us from mowing the lawn.
Over the last few months the appearance of my letter box has changed significantly. Ten
months ago (first photo) it looked OK, but about 3 months ago I noted that the right edge of
the brim had been bent. It's relatively solid metal, so that shouldn't be easy:
Since then, further damage has occurred. The pole holding the box has been bent back away
from the road, something that's not simple (it's cast in concrete), and the top part has
been deformed:
Who would do that? How? It seems reasonable to expect it to be the postal delivery
contractors (they're not called postmen here for some reason), but it would require quite an
effort, and it doesn't seem to make any sense.
A couple of years ago I started writing a page on frying eggs. I didn't get very far. But
today Sue Blake was discussing the matter on IRC, and I revisited the matter.
The most surprising thing is that we weren't in agreement on what constitutes a good fried
egg. Two years I wrote:
It's surprising how difficult it is to make a perfect fried egg, one with the white closely
surrounding a perfect soft yolk.
It's an egg with a uniform consistency in all parts: the white just solidified enough to be
able to pick it up with a fork, white and creamy; the yolk still liquid, a little thickened,
covered with a light white veil which makes it reflect...
But, surprisingly, this isn't what people on IRC think to be a good egg. The white should
be browned. I've always looked on that as a sign of overcooking, and maybe it's what
encourages obscenities like our William Creek eggs (first
photo) or the ones I got in Canberra five years ago:
Clearly there are two issues here: how long to fry the eggs, and how carefully to serve
them. Both of these ones fail in the care department, but arguably the second ones would
have been acceptable to some people if they had been served with more care. In this
connection it's interesting to notice Americanisms like “frypan”, “sunny
side up” and “easy over” creeping into Australian usage. I had never
heard them until I went to Cupertino
with Tandem.
Madame Saint-Ange recommends baking the eggs in an oven rather than frying them (the usual
French word, «œufs sur le plat», means “eggs on the plate”, so there's no
contradiction here). Did some thinking about that; she recommends small baking dishes,
which were apparently available at the time the book was written in the early 20th century.
But they sound pretty much like the Chinese saucers I used for my steaming experiments 2
years ago, and I suspect we'd have the same problems. So I'll probably give that one a
miss.
Still, Madame Saint-Ange summarizes some of the problems of frying eggs:
if you make them on the stove, the heat only
reaches the egg from underneath, in too partial and also too direct a manner. Only the
bottom of the egg is warmed. The white hardens to a certain thickness and remains slimy
on the surface, while the yolk is barely warm when the white is cooked.
We have always addressed that problem by spooning hot oil over the egg, which cooks it from
above as well. That's not so much the issue as keeping them from spreading out, especially
if they're older. Rings don't seem to work properly, and in any case give a rather
artificial appearance to the result. But it still seems to be the best we have.
The curry pieroge we made on Saturday
didn't quite match the definition
of Pierogi, and Sue Blake had the idea
that it might be closer to a Pirozhki
(пирожки). From the photos and description, that makes sense, but the Russian page shows various pictures,
including ones that look more like Pierogi. At the bottom it also states:
Категория: Пироги
I don't speak Russian, but it's easy enough to transliterate: “Kategoriya:
Pirogi”. There's no Russian link on
the Pierogi page, but if you follow the
Polish link from each page, you end up with Pierogi in each case: it's just an alternate
name, and the English pages are misleading. I'd be more inclined to think that what we
cooked is a completely separate dish, though. I'd consider finding a new name for it,
except that that would muddy the waters further.
Finally the weather was conducive to spraying weeds, and did a fair amount, unfortunately
not getting finished. I suppose we'll have to do a lot of weeding by hand, and keep it up.
I've been keeping a careful eye on my satellite link since the beginning of the month. I've
already established that the external IP address hasn't changed, and I've had three outages
so far this month, but the latency seems to have improved marginally. Finally managed to
log in to My SkyMesh, only to read:
Unfortunately your modem has not yet been transferred, so your usage information is
currently unavailable through My SkyMesh.
Approximately 1,000 modems are transferred each day by IPSTAR and they expect the process
to be complete by May 17, 2010.
It's difficult to understand why this should take so long. Still, there's no indication
that they have a mirror server, let alone one that doesn't get charged towards your data
allocation, so took the opportunity to download the latest Ubuntu DVD image.
Today was supposed to be rainy, but we had bright sunshine, so out with Yvonne to try out my new idea about how to mount the roof of the
greenhouse. In exactly the second where I started to lift the roof section, the rain came
and quickly made it clear it didn't intend to stop. Something doesn't want me to finish
this project.
Two and a half years ago we started on the garden to the
east of the house, with a circular bed with succulents
(aeonium) in it:
Even at the time, we weren't sure that it would stay that way, but we haven't done much with
it. Since this photo we put an Aloe
vera in the middle, and at some later date some kind
of saltbush established itself. The
aeoniums proved to be less attractive than we thought, and the whole thing became pretty
overgrown:
We weren't helped by the weather, which rained off and on, but in the course of the
afternoon removed the aeoniums and saltbush, and finally found our aloe, which had become
seven individual plants:
I've decided I don't like aloes very much. Unlike
agaves, they don't seem to grow bigger,
just more, and they look a mess. Admittedly, these ones aren't a good example, since
they've hardly had any sun in a year or so, but I've seen it with others. We'll probably
take them out too and put in a cycad that's
in a less-than-optimal place.
Planted some of the removed plants in the extreme east, in the area between the existing
birch trees and the south-east paddock, which looks like being the next phase of the garden:
In two weeks I'm doing a presentation of my weather station software for BLUG. So it's high time to prepare the slides. I've been
doing them with groff and my own macros for about 10 years now, but it's been a while
since I did any work with them. Some time ago I started on a framework for making this sort
of thing easier, but I never finished it, and I have long since forgotten many of the
details.
All that made creating the slides anything but easy. It's certainly not an advertisement
for the UNIX Way, at any rate, nor are the kludges I needed to make to get it to work at
all: modify ghostscript/lib/gs_statd.ps to create a display-specific “page
size”, and massage the PostScript output to reposition the images. And for some reason the position on the page had changed, so I
had to change the repositioning. The whole thing is almost too hackish to be practical.
One point to Microsoft? Maybe. But the last time I
tried anything similar from the Microsoft space, it drove me mad. Maybe that's just a lack
of understanding of the tools, but I'm still left with the feeling that I can only do what
the developers thought of: it's not extensible. Mine may be incredibly painful, but at the
end I get what I want.
My attempt to download the Ubuntu DVD failed
sometime during the night with ECONNRESET. Given the reliability of the link, I
would have been surprised if it had not. In the course of the day restarted many more
times. There's a specific pattern to these things: normally the download proceeds at about
100 kB/s, considerably more than it used to from this site—I suspect that Aussie Broadband have upgraded a link
somewhere. But then it stalls. tcpdump shows that it waits a long time, then
returns an ack with window 0 and what appears to be the wrong position in the stream. This
is immediately (< 1 s) followed by an ack for the same position with window 8192. It
seems that the congestion control algorithm is broken. On each occasion I have been able to
stop the transmission, restart, and it continued at full speed. I'll have to keep more of
an eye on this one.
More discussion about the Pierogi vs Pirozhki debate today. I discovered that the
“Категория: Пироги” information I found yesterday was in fact a link, and it led
me to the Пирог (Pirog) page. And
that does have an English link: to Pie. And then Peter Jeremy asked a Russian colleague, who told him that the
information on the Russian page was incorrect, and that the Pirozhki illustrated on the page were in fact Pirogi. Somehow
this whole thing is a can of worms. I suspect that the terminology varies according to
geography.
To add to the confusion, I put the Pirog page through Google Translate, which came up with something half-way understandable translation
which translates “Пирожок” as “patty” (not a word I'm familiar with,
but which the OED tells me is a “little pie
or pasty”) and “Пирог” as “pie”. So I suppose
“pie” is the best translation, and it explains why it's such a difficult term to
pin down.
The current state of play with erecting the greenhouse was that we had four roof sections, each consisting of a cross-member and
two rafters, and we couldn't work out how to mount them. Clearly two of these fitted on
each side of the central arch, but the rafters weren't long enough to reach the gussets at
the top (thoughtfully mounted on two of the sections):
Since then I had taken another look at the parts and identified screw holes about 20 cm from
the bottom of the rafters forming the arch, and decided that they must fit there. The rain
let up enough for us to make another attempt. It didn't need to be very long: yes, there
were holes on the end rafters, but not on the middle one. Clearly that wasn't the way it
was intended either. Somehow the whole thing doesn't seem to fit. And just as we were
giving up, it started to rain again.
Did some more head-scratching and looking at the mounting holes. There's only one place on
all the rafters where the cross-member would fit, but it's close to the top, so the rafters
would have to point down, and the gussets would be at the bottom, where they don't fit. To
fit the things, we'd have to remove the gussets. But it turned out that that's the correct
way to assemble the roof. The gussets go on the other end of the rafters, close to the
cross-members:
Now everything makes sense. There are ventilation flaps along the roof, and they attach to
the same kind of gussets. They're upside down in this photo:
Each flap has two gussets, which fit exactly into the holes at the (real) top of the
rafters. But that leaves a couple of questions unanswered: first, why were the gussets
attached to the wrong end of some of the rafters, and secondly, where do they fit at all? I
had already established that one gusset was missing, but now it's beginning to look as if we
have three too many. Has somebody (David?) already made an attempt to re-erect the
greenhouse, and didn't finish?
Got a surprising amount of feedback about making slides. It seems that there's lots of
stuff out there, some of which is clearly not worth pursuing. ewipe didn't even get
off the ground: its home page is gone. ImPress didn't: the web pages are badly laid out and overflow their boxes.
It's not as bad as some “high profile” web sites, but not good enough for a
presentation package. But then, the last “release” was 1.1 Beta 6, some time in
mid-2000. pylize, on the other hand, doesn't look bad. If I were starting again, I
would probably look more carefully at that one. But now that I have fixed the immediate
problems with my groff kludge, I don't see any reason to change.
By the end of March, I was getting the impression that my network dropout problems were over. That has definitely changed. There have been 12
in the last four days, 7 of them today, and independently of that, my downloads are hanging
all the time. Nevertheless I have managed to miss the smoking gun in my tcpdump
traces.
Last week, when changing the RSS generation functions for my diary, I installed newsfox, and have been doing a bit of playing around
with it. Surprise: it seems relatively useful, particularly for forums. It pulls in feed
updates (automatically if desired), it can show the index in chronological format, and the
standard display is black on white, which makes many feeds a lot easier to read.
There are two kinds of down side. One is the way people format their feeds, which
isn't newsfox's fault. For example, oly-e.de formats its messages as white or light grey on dark grey, and in a mixture
of chronological and reverse chronological sequence within a thread. newsfox fixes
that, but oly-e.de avenges itself by sending only the first 250 characters of each
posting, omitting even the name of the author. To see the rest, you need to click on the
title, bringing you back to the original posting, and thus rendering the use
of newsfox pretty much useless. The dpreview Olympus forum is
even worse: it only provides the header. And then there are things like Yana and Sundance's
travel blog, where it only
shows some of the entries, though the original includes them all. It's not clear whether
this is the fault of the blog site or newsfox.
Then there are things that newsfox could do better. Like so much software, it has no
idea of time zones, so the times I get displayed don't match the original feed. It's also
one of these horrible single-window things with panes (pains?), so it tries to fit too much
into too small a space, with ellipses (…) all over the place:
I've found that you can move the message display to the right, which helps a little if you
make the window full screen. But wouldn't it be nicer to have a group of windows that all
iconify and deiconify at the same time? GIMP goes the multi-window route, but unfortunately you have to iconify and deiconify each window
individually, which makes the idea pretty useless.
Apart from that, I thought that one of the ideas of news readers (which is what this really
is) was to remove entries when they had been read. But I can't find a way to do it
with newsfox. It changes the emphasis, but it leaves the message there, making
effective use of chronological sorting difficult.
I read about your pierogi/pirozhki naming dilemma. The answer is actually quite simple:
The naming differs between Polish and Russian. In Poland, we call a big pastry a "pieróg"
in singular. The little dumplings are called "pierogi", the plural of the previous word.
Russians do it differently. They call the pastry "pirozhki" (both singular and a
diminutive). The little dumplings are called "vareniki", plural again. If you look at the
Wikipedia pages for Polish periogi and Russian vareniki, you will see that they are the
same. Pieróg and pirozhki equally match.
Phone call from Robert at SkyMesh this
morning, in reply to my email. What email? I hadn't sent any, not even a web form. I
suppose he really meant the phone call I made on Saturday, but I hadn't really expected a
call back, certainly not to tell me my new IP address. Still, that's what he did, and he
assured me that it would have reverse DNS (it currently doesn't). He also offered to raise
the priority of my changeover, but also confirmed that they don't have a mirror server, so
I'm happy to stay with Aussie
Broadband for the while. And he couldn't give me any information about reliability;
they don't monitor it. sigh
After yesterday's success with the greenhouse, continued with the rest of the structure,
which went well. Got the remainder of the structure in place (I'll screw most things tight
later), including one door and rail. Now I only have a couple of things over: the flaps at
the top, which need straightening before I can mount them, and a few odds and ends which I
can't place.
As I had noted, there are three gussets, engraved “arch gusset”, but which would
also fit between the rafter and the wall, explaining their position on one of the roof
sections when I started. But there are only three of them, and there's space for at least
eight. Have five been lost? Just about everything else seems to be there, and there are
the little plastic gussets to hold things in place as well. I think I'll attach them where
they seem to contribute most to the structural rigidity.
Apart from that, I have a couple of items that I really can't place. One is a U profile
as long as the greenhouse, which looks as if it should fit on top. But there's no obvious
way to attach it, and the flaps don't look as if they need anything on top. Wouldn't it be
nice to get the instructions, or at least a look at a correctly assembled example?
I'm supposed to have a blood test every three months, followed by a discussion with the
doctor. Things don't quite work out that way. I'm not too worried about the blood tests,
but seeing the doctor is more and more of a problem: the Eureka Medical Centre doesn't do
appointments, and I have to wait up to 2 hours to see the doctor. I can usually plan to do
something else during the wait, but it means I have to go in early to get finished before
lunch, or alternatively buy lunch in town, something I'd rather not do. And the doctors
have wildly varying shifts, so I have to call up in advance to find out when he's there at
all. On one occasion I was given incorrect
information, so I ended up going into town in vain.
As a result, I keep putting off the doctor's visits. My last appointment was at the end of January, so I could have gone in any time after
the beginning of February, but in practice I keep putting it off until I need to have the
next blood test; and even then I find excuses to overrun. But today I ran out of excuses.
I rang up yesterday and waited forever for somebody to answer the phone, and when she did,
she just asked me to hold. Finally I was told he would be there today from 9 to 12, so up
early to head in. When I got there, I saw the longest queue I have ever seen there:
From experience, that would have taken 30 minutes to even register—then up to 2
hours to see the doctor, by which time he would have finished for the day. Possibly I
wouldn't have had to wait that long, but if the registration queue's that long, I wouldn't
bet on it. And it's pay in advance—no refund if you don't see the doctor? So,
furious, I left again. It's a real pain; where do I find another doctor? How do I know how
good he'll be?
Then went looking for a new jacket, but somehow I'm not made for that sort of thing, and I
certainly wasn't in the right frame of mind. I ought to get Yvonne to decide for me. Looked round a few other shops, but after an hour, I was
still furious, and went home without having achieved anything.
In the afternoon, got round to replacing the gas line that Piccola had chewed through. This stuff is a pain! It's extremely stiff, and when
cool it's almost impossible to fit over the fittings. So I've taken to putting the ends in
boiling water and then sticking in a chopstick to widen it a bit. Today, though, the
chopstick had the wrong profile, and I had to push it in almost its whole length to get it
wide enough—and then I couldn't get the bloody thing out again!
Putting that length of tube in hot water was not easy, so decided to warm it a little more
in the oven. Ended up having to tie it in a knot to get it in there, and heated to about
150°. Too much, it seems:
At least I got the chopstick out. It's a good thing the stuff's cheap.
Fitting the line at one end required removing the regulator. That required a spanner, of
course, and I went to get one of the set of 4 adjustable spanners that I bought recently.
It was still in its stiff plastic wrap, and while removing it, managed to cut myself on the
plastic. In a society where (allegedly) people can sue microwave oven manufacturers because
they can kill dogs, I wonder how people can still use such sharp packaging.
Removed the regulator and fitted the line with no more than the usual pain. When I refitted
it, I found that the needle of the high pressure gauge had somehow got itself on the wrong
side of the stop:
How did that happen? I can't find any way. I also can't find a way to fix it. Looks as if
I'll need a new regulator.
Then I remembered that I had intended to put a protective outer tube around the line.
Clearly that needs to be done before putting on the fittings at the end. And, of course, I
had forgotten. Ended up cutting the tube lengthwise and fitting it on like that. That, at
any rate, worked without further problems:
Last month I observed that people can abuse
TinyURL to hide dangerous-looking URLs, and
that they even fooled the Apache team that way.
Maybe that's a trend. Today I found yet another spam message (“Your mailbox is full.
Give us your passwords”) that had found its way past an ever less efficient SpamAssassin. It included a text version:
Date: Thu, 6 May 2010 22:01:05 -0400
From: "Marie Hamilton" <Marie.Hamilton@medvance.edu>
To: <service@webmail.org>
Subject: Your mailbox quota has exceed the storage limit which is 20GB?
Your mailbox quota has exceed the storage limit which is 20GB
as set by your administrator, you are currently running on 20.9GB.
You may not be able to send or receive new mails until you re-validate
your mailbox.
To re-activate your account please click the link and login with the
username and password provided for you below:
http://update-mail.4-all.org/ <http://update-mail.4-all.org/> <http://email-update.4-all.org/>
Nothing unusual in itself. The sender domain has nothing to do with the domain where you're
supposed to surrender your personal details, but that's normal. And doubtless there's a
link with a name that doesn't match the real link.
But wait—this is the text! And though the two link URLs don't match, they're both in
the same domain. Would 4-all.org really
host this kind of scam? Who are they, anyway?
whois tells me that they've been around for a couple of years, in the Netherlands,
and following the real link shows that they offer a service something like TinyURL, except
that they don't replace the URL in the URL window, so you're left thinking that you're still
with (relatively kosher) 4-all.org. You really need to look at the page source to
see who the real culprit is:
I know that my weather station is pretty flaky and
frequently returns invalid information, but lately it's been extreme: dozens of readings
with outside temperature -0.1° and wind speeds 70.8 km/h (always these values in each case),
and occasional rainfalls that would put Noah's flood to shame, up to 70 metres of rain in a
few seconds. Is the thing dying on me?
All this comes from the outside station. It seems to have started since I replaced the batteries last month. Cable problems, maybe? No, the
temperature sensor is in the same housing as the transmitter, and the wind and temperature
errors always happen in exactly the same records. But last month I put in NiMH
rechargeables instead of normal 1.5 V batteries. Is this its way of telling me it doesn't
like them? Put in some 1.5 alkalines and got another error almost immediately, but after
that there was no further problem—yet. I still need to improve error recovery.
A very short power failure at 7:51 this
morning. It's interesting to note that the electronics in the oven can handle short
failures; Chris Yeardley has an oven with the same kind of module, and she confirmed that
hers survived too. But every failure for even a second requires me to reset several clocks.
The network dropouts continue. Every time I look to see
if I have changed IP address; at 12:58 today it finally changed to the new address, which of
course doesn't have reverse DNS, as clearly indicated by the IRC logs that Peter Jeremy kept:
2010-05-08 12:56:19 <-- grO0gle (grog@121-200-1-204.cust.aussiebb.net) has quit (Read error: Connection reset by peer)
2010-05-08 13:05:29 --> grO0gle (grog@180.181.112.227) has joined #bugs
The immediate impression is less than stellar, with a 50% drop in TCP speed:
In the course of the day, had no less than 6 outages. Hopefully that's just cutover
problems and not an indication of what's to come. But it got me looking for alternatives
again. Peter Jeremy tells me that they had 3G connectivity with Optus almost up to our
house when they came here three weeks ago, so maybe I
should investigate that again.
Some months ago I removed some potted
daffodils and tulips, and I've been meaning to plant them almost ever since. Where? We've
now removed everything from the original succulent bed:
But we have other things planned for that, and these bulbs flower in early spring, when
we're not outside much. So it makes sense to plant them where we'll see them from inside,
and the obvious place is to the north of the kitchen, where we normally eat, so planted them
in the gravel between the north bed and the house, in groups of three: daffodil, tulip and
Watsonia, which will hopefully
keep us through most of the spring and into the summer.
We had planned to go riding today, and the weather was ideal. And then I saw:
May 9 07:55:04 lagoon postfix/smtp[1127]: AE04D50815: lost connection with c.mx.mail.yahoo.com[206.190.54.127] while receiving the initial server greeting
Further investigation showed that no mail had gone out since midday yesterday,
coincidentally the time the satellite network service cut over. I was able to
connect to SkyMesh's mail server, and I was
just about to complain that they were blocking ports when I found I could also connect to
www.lemis.com (also known as ozlabs.org). Called up SkyMesh support and spoke to
Ben, who needed some explanation about what was going on (“you can't contact your mail
server?”) and put me on hold a couple of times. Finally he accepted that there was a
problem, and probably in their configuration, but that nothing could be done until tomorrow.
That was fair enough, but I also pointed out the missing reverse DNS:
Greg: “I was told that my IP address would have reverse DNS”.
Ben: “Where did they say the reverse DNS would be?”
But he says he'll look into that too. In the meantime, put www.lemis.com in my
Postfix configuration and confirmed that the
mail queue went away.
All that took about 30 minutes, and I was about to go out and saddle up when it occurred to
me that the test message I sent myself by way of freebsd.org hadn't come back, so back to look at the log messages again:
May 9 11:21:47 dereel postfix/smtp[16073]: E82E4A1019: to=<BBruening@gpd-globalpress.de>, relay=www[203.10.76.45]:25, delay=80239, delays=80234/0.01/3/1.2, dsn=5.7.1, status=bounced (host www[203.10.76.45] said: 554 5.7.1 <BBruening@gpd-globalpress.de>: Relay access denied (in reply to RCPT TO command))
My IP address had changed, and the ozlabs configuration refused the new address. Not
only did I have to fix the config on ozlabs (to which, fortunately, I have access),
but I had to go back and find and manually resend the messages. As it turned out, that
didn't take that long, but I didn't know that in advance, didn't want to hold up
Chris and Yvonne, and was generally not in the right frame of
mind to go riding, so they went without me. Fortunately, the rest went without further
incident, but what a pain!
The good news is that the TCP speed seems to have recovered:
In the afternoon, decided to do something completely different, and off to the Ballarat Bird World to see what they had
there. We appear to have been the only guests. The first thing that we saw were no birds
at all, just lots of interesting mainly native plants:
While we were looking at these birds, Chris, one of the people working at the park, and
asked us if we would mind being part of a video they were making, offering us incentives
such as half-price tickets next time. I wonder why they need that kind of incentive; it's
good enough just to get a free guided tour of the place. Of course we accepted.
Most of it was in the cage with red-tailed black cockatoos (Calyptorhynchus banksii subspecies Macrorhynchus if I understood it
correctly), of which they have two friendly males, Cheeky and Jesse. Cheeky is a shoulder
cockatoo:
And Jesse doesn't like to be picked up, just to be stroked and be allowed to investigate
clothing. It's surprising how gentle they are. They look as if they could bite off a
finger if they wanted, but when I held out my hand, Jesse just knocked on it with the curved
part of his beak. They're surprisingly agile. It's surprising how accurately they can
position the tip of their beak:
On the way home, stopped off at a place
in Napoleons where they have plants on
sale on the roadside, and ended up spending $53, mainly on pots. Got some “Belladonna
lilies”, which I later discovered
were Amaryllis, and three plants which
I have yet to identify:
While in a gardening mood, attended to the
small Salix that Laurel Gordon brought for
us last October, and which she suggested we should use as
a fake Bonsai. It's not looking happy,
but there are a number of buds on the stems:
That's the plant in the foreground; the stuff above is a
volunteer Lobelia. I have had the
suspicion for some time that the roots are rotting, and that it would be a good idea to
re-pot it. Took it out of the pot, and discovered that the roots were in excellent
condition: they were about a metre long, and the whole plant was root-bound. So maybe not
such a good idea as a bonsai after all. Put it in one of the new pots while we decided
where to plant it.
One of the things we decided when we got Piccola was that, though we couldn't bear to see how unhappy she was not being allowed out of the
house, she would stay in at night. We thought that getting her to come home in the evening
would be easy enough, because that's when we feed her, but lately that doesn't seem to be
working. Tonight we spent nearly 30 minutes looking for her, and finally gave up just
before she returned by herself. How do we get her to come back at the right time?
I pretty much have everything in place with the greenhouse now, but there are still a number
of things that puzzle me. Chris came over today to take a look, but she couldn't add much:
At first I thought I had missed something out, but there's nowhere to mount anything on
that surface.
Part of that could conceivably be explained by the assertion that I have mounted the
roof sections the wrong way round after all, but all that seems to fit correctly:
There are three arch gussets left over. It looks as if they might fit between the walls
and the rafters, but there are 8 positions and only 3 gussets. I could believe that one
gusset could get lost, but 5 sounds less probable.
The long one is just shorter than the greenhouse, and seems to be designed to fit
alongside. But there's only one of them, and the only place a single component that
would make sense would be along the top of the roof, and there's no place for it there.
It also has only three screw holes: one at one end, and two at an angle close to the
other end. That doesn't make much sense to me:
While in the garden, finished off the renovation of the succulent bed, using some of our
compost in the process. I've decided to sieve it through a wire mesh, which certainly
improves the appearance:
How do you transplant a Cycad? Carefully,
I suppose, since they're pretty spiky, but I also went to some trouble to get out as many of
the roots as possible. I wonder how successful I was:
Planted it in the middle of the round bed, surrounded by some unidentified ground cover that
we picked up in Warrnambool a
couple of years ago, and with
some Gazanias around the outside. Here
the development of the bed over the last few years:
Yvonne took a lot of photos in the garden while we were doing
the work, and I processed them during the afternoon. One of the steps I do is to run it
through the Ashampoo photo optimizer to
enhance the appearance. It doesn't always improve things, so I always compare the results.
Today I had a particularly large number of spectacular failures:
Particularly the first one is astounding. But what's more surprising: this isn't a
comparison of unoptimized and “optimized” images. Both images of each pair have
been made from the same base image and put through the optimizer. The only difference is
that the first one has been done as an individual image, whereas the second one was done in
a large batch. I wonder what caused that. Does the optimizer try to learn from other
images? If so, it's not doing a very good job. But this is a very old version of the
optimizer. I have a newer version, but it's even more point and grunt, and so I never
started using it. The other issue here is that all the failed images were taken with
Yvonne's Kodak
M1093 IS camera. Is there something about those images that upsets Ashampoo? Or the
combination of those and images from my E-30?
I'm still using firefox version
3, and over the course of time I've noticed how horribly slow it is in rendering large photo
images, such as the “big” version of the images on this page—over 15
seconds for an image on the local web server. Today tried again with Opera, which rendered at acceptable speeds, less
than 2 seconds. Time for firefox 3.5? I suppose so, but it's a real pain trying to
upgrade things. I'd be use Opera, except that it does other things less well, and it
had a habit of leaving no vertical space between images, though it does leave horizontal
space. Is that what the standards decree? It seems unlikely, but then I've seen many
unlikely things in the web area.
Interlaced JPEGs
One workaround for the slow rendering problem might be interlaced JPEG images. Did a bit of
playing around with that, conveniently helped by ImageMagick's man page for convert:
-interlace type type of image interlacing scheme
And that's the only mention of interlace in the man page. To find out what type is,
you have to go to the web page. Tried it out, and it worked, but didn't really seem to improve things
much. Yes, the image comes more quickly, but the total time is no less, and until it has
elapsed, the image is usually in the wrong position.
Yvonne is planning a new garden layout. With what tools?
Currently a sheet of paper. Wouldn't it be nice to have some drawing software to do the
job? Went out looking and found all sorts of stuff, none of which appears to be for
UNIX-like operating systems (if you except MacOS). It looks like sketchup might be an option, though the requirements
suggest that my Apple is too wimpy to run it. Tomorrow, when the downloads are cheap again.
I have a pretty rigorous backup schedule for my machines, making a backup every day: level 0
dump at the beginning of the month, level 1 on Saturdays, and level 2 the other days. But
some machines get left out of this schedule, notably pain (the Microsoft box)
and boskoop (the Apple box), mainly because I hadn't worked out how to do it. In the
case of pain I make complete disk images from time to time in order to avoid any
strangenesses that might come from restoring partial images. I've done the same
with boskoop, mainly out of laziness, but today I decided to add it to the standard
backup schedule.
Does Mac OS X have dump? Yes.
But it doesn't work:
=== root@boskoop (/dev/ttyp2) /Users/grog 4 -> dump -0uf /dump/boskoop-Darwin/root / DUMP: Date of this level 0 dump: Wed May 11 12:16:43 2010
DUMP: Date of last level 0 dump: the epoch
DUMP: Dumping / to /dump/boskoop-Darwin/root
DUMP: bad sblock magic number
DUMP: The ENTIRE dump is aborted.
Why not? The man page doesn't say, not even in the BUGS section, but it
seems that dump only works
on UFS. And the root file
system is clearly Apple's HFS+.
So how do you do a backup of that? People suggested that I should use “Time
Machine”, which sounds to me like a replacement
for cron, but which I'm told is a
backup utility. Why does Apple have to come up with such confusing names? Went looking for
it, and discovered it's not on my system: I'm running Mac OS 10.4 on my old machine, and
“Time Machine” wasn't introduced until 10.5.
So didn't Apple have any backup utility before 10.5? Somebody suggested the
Disk Utility so went looking
for that. Nothing obvious there; it has a restore but not a backup. But again I'm
obviously out-of-date: restore is the new backup. “Mac Help” tells me:
You can use Disk Utility to create a disk image of a single device, folder, or volume. You can use the disk image to
transfer files from one computer to another, or burn the image on a CD or DVD and use it to
restore the contents of another disk. To learn more about disk images, open Disk Utility, in
the /Applications/Utilities folder, and choose Help > Disk Utility Help.
So I did that. And got a page with the headings “Repairing hard drives”,
“Erasing disks”, “Partitioning a disk”, “Using RAID
sets” and “Using disk images”. It seems that the last is the one I'm
looking for, and I'm left with the feeling that the sequence, which is clearly not
alphabetic, is in descending order of the importance that Apple places on the function.
Also, instead of concentrating on what I'm trying to achieve, the documentation looks at it
from the perspective of Apple's baroque implementation. It's also particularly encouraging:
If you created a blank disk image, you may be able to copy files from your hard disk to the
image.
Once again I have a disconnect with the mind set of people who write this stuff. It seems
that a disk image is just another name for an archive, but I had to find another page to
find out how to perform the backup: of course it's a point and click utility, exactly what
you want to run in the middle of the night with cron.
And that really seems to be all that Apple offered in the way of backup before Mac OS 10.5.
I know that the idea of backups is foreign to non-technical users. But how can you take any
vendor seriously when he doesn't even offer a basic backup utility? Yes, it's there now,
though I'm not sure I want to look at it, but Mac OS X 10.0 came out in January 1999, and
10.5 was released on 26 October 2007, nearly 9 years later. I've been
bitterly disappointed by Mac OS X in many ways, but this really takes the cake.
So what did I do? There's still tar, and that works. How
many Apple users even know what that is, and how many of those use it?
As planned, downloaded sketchup and tried
it out on the Apple. I don't know whether it's my background or what, but none of this
stuff seems to be the slightest bit intuitive. Tried out a few things, but it didn't do
what I expected. Clearly any relatively complex program must first be learnt, and there are
videos, which I can download in the morning, but I also wonder if there's some issue with
running the software over VNC: it
seems strange that, when drawing a line, everything seemed to be OK, but the line didn't
appear; at some later time it did. So maybe I should try the Microsoft version first. In
any case, I'm reminded of the discovery I made in the late 1980s that in many cases the
effort required to learn how to use a program isn't worth the trouble.
In the meantime, on the recommendation of many people, I installed xfig, which has a really old-fashioned look about it
(Athena widgets, would you believe?),
but which seems to do what I expect.
When Sue Blake came to the Hackers'
barbecue last month, she bought a lot
of doufu to live on for the weekend. But
then she tried our Indian food and changed her mind, and left the doufu behind. So today we
tried a recipe for braised deep fried
doufu based on one in “Chinese Home-Style Cooking”, published by the Foreign language
press, Beijing, 1990. And it suffers from all the usual problems of cookbooks the world
over. I recognized that some of the proportions just had to be wrong, but I still
didn't get things right. But it didn't taste that bad; we'll have to improve on it next
time round. The only issue is that a dinner with just doufu is a little one-sided.
Int the office this morning to find that my satellite link had been down and up no less than
5 times in the night. And of course we still have no reverse DNS, and no SMTP connectivity
to most of the world. Called up SkyMesh support on 1300 662 331, and after 14 minutes waiting was finally connected to Joel, who
took another 4 minutes to get my details on his screen—“Sorry, the computer's
really slow today”. He confirmed that there were “definitely a few
dropouts”, and put me on hold while he spoke with the “senior blokes”,
who, he said, were confused about what was going on, and promised that they would call back
this afternoon.
In the meantime, did some thinking about the issue, including lots of traceroutes to
find out where the problem might lie. Lots of information; the annoying thing seems to be
that I can't use TCP with traceroute from here, possibly because BST confuses the
issue. But normal traceroute suggests that the problem is quite close to the SkyMesh
nodes, as you'd expect. Here's a traceroute to ozlabs.org, which I can reach with SMTP:
=== grog@dereel (/dev/ttypb) ~/public_html/net 111 -> traceroute -e -p 25 www traceroute to www.lemis.com (203.10.76.45), 64 hops max, 40 byte packets
1 sat-gw (192.168.5.100) 0.337 ms 0.303 ms 0.332 ms
2 192.168.1.9 (192.168.1.9) 795.023 ms 917.802 ms 819.796 ms
3 203.169.16.37 (203.169.16.37) 597.413 ms 696.543 ms 1006.503 ms
4 203.169.16.25 (203.169.16.25) 789.502 ms 619.935 ms 617.839 ms
5 203.169.17.38 (203.169.17.38) 633.189 ms 611.906 ms 625.710 ms
6 as7718.sydney.pipenetworks.com (218.100.2.96) 620.205 ms 620.991 ms 617.750 ms
7 gigabitethernet15-15.core01.gate.transact.net.au (202.55.144.81) 653.353 ms 614.158 ms 627.037 ms
8 static-146-86.transact.net.au (202.55.146.86) 601.113 ms 672.221 ms 1126.751 ms
9 ...
And here's a traceroute to w3.lemis.com, which isn't reachable:
=== grog@dereel (/dev/ttypb) ~/public_html/net 112 -> traceroute -e -p 25 w3 traceroute to w3.lemis.com (208.86.224.149), 64 hops max, 40 byte packets
1 sat-gw (192.168.5.100) 0.348 ms 0.585 ms 0.636 ms
2 192.168.1.9 (192.168.1.9) 643.174 ms 579.203 ms 621.394 ms
3 203.169.16.37 (203.169.16.37) 664.295 ms 651.111 ms 606.632 ms
4 203.169.16.25 (203.169.16.25) 619.895 ms 596.455 ms 629.510 ms
5 203.169.17.38 (203.169.17.38) 652.465 ms 583.803 ms 621.370 ms
6 ge-0-0-3-919.bdr01.syd03.nsw.VOCUS.net.au (114.31.193.18) 1036.638 ms 1385.332 ms 1498.367 ms
7 ge-0-1-2.cor02.syd03.nsw.VOCUS.net.au (114.31.192.66) 1399.970 ms 1329.475 ms 2174.834 ms
8 ge-0-1-4-135.bdr01.sjc01.ca.VOCUS.net.au (114.31.192.73) 775.250 ms 1655.733 ms 1559.417 ms
9 ...
In all cases, the MTAs I could reach are connected to 203.169.17.38 via Pipe Networks, and the ones I can't read are
connected via Vocus. But there's little I
can do about that.
That proved not to be the worst thing, though. A couple of hours later I had another
dropout. And another. And another. By 13:15 there had been a total of 13 for the day, and
we had another 2 in the evening. And round 10:30 the TCP
speed halved:
Spent a lot of time looking at alternatives to the situation. It's beginning to look like
3G again. Wouldn't it be nice if Optus could
build the Dereel phone tower soon? But I
think I might as well invest in an antenna; if only I could be sure that I'd then have
coverage.
And of course, no call back from the SkyMesh people. 28 outages in 5 days, and counting:
Total 28 outages, total time 1951 seconds (00:32:31)
Average time between outages: 15429 seconds (04:17:09)
Average duration: 69 seconds (00:01:09)
Availability: 99.55%
ALDI had a number of things on offer today,
including a cheap GPS
navigation system, and I had other things to do in town, so off early with Yvonne to get there when the shop opened: this kind of thing disappears
quickly. Last time they opened several minutes early,
but today they made up for it by being late. I don't know how many of the systems they had,
but they had hidden them at the cash register; I got the second one, and I suppose there
were more. Also picked up quite a bit of other stuff for the garden: secateurs, Li-Ion leaf
blower (which proves also to suck), Li-Ion hedge trimmer and some bamboo screening.
Then the normal week's food shopping, to Mountain Scenery to order some mulch, and in to
town, where we finally found a jacket at Big
W—in fact, found two, at quite acceptable prices. Then to David Chestnut
looking for a lawn mower—we've decided that we want a ride-on mower after
all—but they didn't have anything available, though it looks as if they might in a
couple of weeks.
Back home, unpacked the toys, most of which required battery charging. The GPS navigator
was at least partially charged, but I've never used one before, so I had to RTFM. In fact,
there were two little manuals, one entitled “Quick start guide”, which started
with useful information: how to get the full user
manual, only on the web, and a recommendation to back up the software to a computer:
How to back up your GPS software
Plug the GPS into your computer, it should show as an external hard drive.
Open the drive and copy the main folder onto your computer as a backup file.
I've already commented on this kind of bad language, and also on the backup mentality of commodity computers. So what is a “backup
file”? Anyway, it didn't get that far. I plugged the navigator into my Apple, and
nothing happened. Investigating the USB bus showed that it was there, but with a strange
identification:
Here's the corresponding FreeBSD log file
message:
May 13 13:41:13 lagoon kernel: ugen0: <Generic Manufacturer (PROTOTYPE--Remember to change idVendor) Generic Serial (PROTOTYPE--Remember to change idVendor), class 0/0, rev 2.00/0.00, addr 2> on uhub3
Clearly that's from the device, not the operating system. But how can that show up as a
disk? There's every indication that this is down-rev firmware. Did the obvious and tried
it with Microsoft (XP), which reported the same messages in a transient little balloon at
bottom right, and then refused to find any new hardware when I tried to do something. So
finally called up the “support” number, 1 300 886 649, which made me wait for
nearly a minute before giving me the option to enter a menu selection (5), and then
disconnected me after another 5 minutes. Tried again, and finally I was connected to Sam (a
woman), who told me I should install ActiveSync 4.5. What's that? Why do I need it? She couldn't answer that. Followed
her instructions, including rebooting the machine (why? Because it's better that way), and
nothing of interest happened. Asked to speak to a technician, but there is only one, and he
doesn't talk to customers. Finally left her with a complaint saying that I believe I've
been given down-rev firmware, and with a request to get a call back. I doubt I'll hear from
them.
In the meantime talked to some Microsoft-savvy people on IRC, and discovered that I hadn't
installed ActiveSync at all, just saved the install program to disk. And Microsoft's search
functions didn't find it because it's called setup.msi, just to make it clear what it
belongs to. Ran that and it installed OK and showed a window which didn't make any sense.
Rebooted and got the same window again; after a bit of investigation, it appeared to be
similar to the “My Computer” window, but apparently showed a directory hierarchy
on the GPS receiver. I was able to copy the files off with a point-and-click combination.
So it seems that ActiveSync is a solution to a problem that didn't exist: instead of
mounting the device as an external disk, it has introduced some proprietary way of making
sure that you have to use only Microsoft's tools to use it. No possibility of using it on
an Apple or a BSD machine. And, once again, the documentation is not just misleading but
plain wrong. You'd think people were trying to make work for everybody, including
themselves.
Then out to play with the device, with mixed results. It's a touch screen device, but given
the size and the intended use, this may be the best option, though normally I hate touch
screens. The map section is seriously lacking; although it found us, it just put us on
“Dirt Road”, and other roads are missing in the map as well. Some of the menus
look quite sensible; others are less obvious. For example, I haven't found a way to use it
as a map, and I haven't worked out how to zoom in and out on some of the map pages. Clearly
I'll have to take some time to get to know this box, not made any easier by the fact that
the manufacturer is obviously so in bed with Microsoft that even the manual is in Microsoft “Word” format. But on the whole I'm pleasantly surprised.
Just by chance we were watching Goldfinger on TV yesterday
and today; that's the film where James Bond's Aston Martin is fitted with a navigator.
Science fiction in those days (1964), but this box is only a fraction of the size of the one
in the film. How times change.
Into the office at about 8:15 this morning, and we had already had four satellite dropouts.
And I still don't have reverse DNS. And I still don't have SMTP connectivity. And the
promised calls back didn't eventuate. Called up SkyMesh support on 1300 662 331, and this time was connected to Kurt after only 5
minutes waiting. He said they'd have to check the satellite stuff, and that SMTP was
blocked by policy, as Ben had told him yesterday. So why didn't they tell me?
The problem is, I have this link under the conditions of the Australian Broadband
Guarantee, one of the conditions of which is that I can't change provider. I chose
Wideband, later Aussie Broadband,
because they didn't do things like that. And one of the conditions of the SkyMesh takeover
of Aussie Broadband customers was that the conditions should be the same. So they have no
right to block any of my traffic. Told him I couldn't accept that, and that they should
enable SMTP by close of business today or I would complain to the Department of Broadband, Communications and the Digital
Economy , the administrators of the guarantee. He promised that Robert (who called
last week, and who proves to be his boss) would call back
in an hour or two. I set a deadline of 13:00, or I would put in my complaint.
All that happened between 9:00 and 13:00 was another measured dropout and several that
didn't register. The problem is the way I do my measurements: there two separate scripts.
One pings five different systems, and counts how many respond. The other pings the other
end of the satellite link. Pings can fail under normal conditions, so I consider it an
outage only if both of these programs get no response from any of these systems. But the 5
pings are sequential, so if we have a short dropout, I might still get a response. Did some
work on my network failure program to add
an -a option, which considers all satellite link ping failures to be a dropout.
The result was very different:
But it's difficult to believe that I really had 32 dropouts yesterday. I don't think that
this option makes much sense.
By 13:00, I still had not heard back, though I had requested a call from Paul Rees, the
managing director of SkyMesh. Called up the sales department, and of course he hadn't been
informed. He was working from home, but called me back within 20 minutes and we had a long
discussion, from which a number of things emerged:
Based on Paul's immediate and patient response, I'm sure that it was the first he had
heard of the issue. While Kurt told me that Robert is the support manager, Paul said
that he was the help desk manager, and that they have other support people. I wonder if
there's an issue with the help desk. It would certainly explain my dissatisfaction with
my previous treatment, and justify my insistence on speaking with Paul.
They do monitor dropouts from their end, based on IPStar's log files; apparently it's a behemoth script
that runs in the early hours of the morning and takes up to 5 days to cycle through all
connections.
The problem that requires the clear channel (which Paul calls “special class of
service”) appears to be a bug in the modem firmware. As we had already gleaned on
IRC, the modems use a form of TDM, and they don't
always use the time slot. While a modem is inactive, another may take over the time
slot, and the first modem has no alternative than to reset, presumably to negotiate a
free slot. That makes sense—the modem firmware really is unbelievably buggy, but
this suggests that it may be inherent in the design. They seem to be “working on
it”, but if they can't even get a web interface that doesn't require
Microsoft “Internet Explorer” to work, I don't have much hope.
Based on my dropouts, he was sure that I was not on a clear channel. Based on my
measurements, I was sure that I was. He said he would check, and that if I was still
getting that many disconnects, it was almost certainly an issue with the modem. I said
that I didn't think so, but if changing the modem would fix the problem, I'd accept it.
I just don't want to pay money for a new modem if the problem is elsewhere, and so far
there's nothing that I can see that points to the modem. He's going to get IPStar to
give a suggestion. It seems that they get on better with IPStar than Aussie Broadband
did.
He later confirmed that yes, I was on a clear channel, and sent me a printout of IPStar's
logs for my modem, conveniently upside-down—I think this is the first time I've seen partially reverse chronological log files.
For today we had:
Session 1273793852 4.3 hours...
*Logon 14/05/2010 09:37 am
Session 1273781882 3.3 hours
*Logoff 14/05/2010 09:35 am <-- Modem powered down (0x00)
*Logon 14/05/2010 06:18 am
Session 1273775830 1.6 hours
*Logoff 14/05/2010 06:15 am <-- No response from modem (0x11)
*Logon 14/05/2010 04:37 am
Session 1273773981 0.5 hours
*Logoff 14/05/2010 04:34 am <-- No response from modem (0x11)
*Logon 14/05/2010 04:06 am
Session 1273729497 12.3 hours
*Logoff 14/05/2010 04:03 am <-- No response from modem (0x11)
*Logon 13/05/2010 03:44 pm
Apart from a spurious claim that the modem was powered down, this tallies quite well with my
own logs:
Start time End time Duration Badness from to
(seconds)
1273773927 1273774011 84 0.081 # 14 May 2010 04:05:27 14 May 2010 04:06:51
1273775783 1273775866 83 2.032 # 14 May 2010 04:36:23 14 May 2010 04:37:46
1273776205 1273776241 36 10.619 # 14 May 2010 04:43:25 14 May 2010 04:44:01
1273781854 1273781909 55 0.641 # 14 May 2010 06:17:34 14 May 2010 06:18:29
1273793875 1273793883 8 0.301 # 14 May 2010 09:37:55 14 May 2010 09:38:03
The only difference was the dropout at 04:43, which IPStar didn't register. Checked the log
files and found:
1273775996 1 5 # Fri May 14 04:39:56 EST 2010
1273776081 1 3 w3 ftp.netbsd.org # Fri May 14 04:41:21 EST 2010
1273776138 0 1 w3 www.auug.org.au ozlabs.org ftp.netbsd.org # Fri May 14 04:42:18 EST 2010
1273776205 0 0 freefall w3 www.auug.org.au ozlabs.org ftp.netbsd.org # Fri May 14 04:43:25 EST 2010
1273776241 0 3 w3 www.auug.org.au # Fri May 14 04:44:01 EST 2010
1273776268 1 3 freefall ftp.netbsd.org # Fri May 14 04:44:28 EST 2010
1273776304 1 2 www.auug.org.au ozlabs.org ftp.netbsd.org # Fri May 14 04:45:04 EST 2010
1273776351 0 2 freefall w3 www.auug.org.au # Fri May 14 04:45:51 EST 2010
1273776378 1 3 ozlabs.org ftp.netbsd.org # Fri May 14 04:46:18 EST 2010
1273776418 1 2 w3 ozlabs.org ftp.netbsd.org # Fri May 14 04:46:58 EST 2010
1273776434 1 4 ozlabs.org # Fri May 14 04:47:14 EST 2010
1273776440 1 5 # Fri May 14 04:47:20 EST 2010
Paul attributed this to “greater sensitivity” of my scripts, but clearly there's
more to it than that. Still, let's fix the big problems first. Another issue that doesn't
look like the modem was a complete dropout for about 80 seconds in the late afternoon:
1273819790 1 5 # Fri May 14 16:49:50 EST 2010
1273819865 1 1 w3 www.auug.org.au ozlabs.org ftp.netbsd.org # Fri May 14 16:51:05 EST 2010
1273819867 1 0 freefall w3 www.auug.org.au ozlabs.org ftp.netbsd.org # Fri May 14 16:51:07 EST 2010
...
1273819929 1 0 freefall w3 www.auug.org.au ozlabs.org ftp.netbsd.org # Fri May 14 16:52:09 EST 2010
1273819941 1 5 # Fri May 14 16:52:21 EST 2010
That's clearly not modem-related. To be observed.
Got a call back from Brett, an engineer who knew what he's talking about, later. Yes,
they'll give me SMTP connectivity, but that will require reconfiguring the routers, and
they'll pull it in some time next week along with some other changes. That sounds
reasonable enough. He also went into more detail about why they think it's a modem issue:
in particular, he thinks it could be a misconfiguration, so on Monday we'll go through a
complete reconfiguration of the modem.
Discussing the GPS navigator on IRC, Daniel O'Connor told me that it appeared that the
software is based on “Windows Mobile”, a kind of Microsoft
“Windows”. That would probably explain why they chose ActiveSync: it is probably part of
the SDK. And it's clearly a solution for
another problem: how Microsoft can ensure that the devices can only connect to a
Microsoft-based computer. That's good for Microsoft, but not for the manufacturer of the
GPS. Still, it makes sense.
Contrary to my expectations, I really did get a call back from Sam about the unit. Based on
the USB ID string, the technician had determined that the unit is defective, and I should
send it back. sigh. What a mess this stuff is! She also confirmed that the unit
can't connect to an Apple, and that it shows up under Microsoft as a “mobile
device” and not an external disk. So clearly it doesn't work as documented. I'm sure
that the unit is as good or bad as any other—we did confirm that the firmware
revisions are the latest ones—so this is just another indication of how much
mess this combination of incompetent programming and incorrect documentation makes.
A couple of days ago Yvonne had problems with her computer:
it kept spontaneously resetting shortly after boot. It sounded like a motherboard problem,
so I grabbed another one, put it in the machine, and the problem didn't recur. So
interesting that I didn't even mention it in this diary.
Today was photo day, and one of the steps is to make a backup of the photo hierarchy. I do
that with rsync to a USB-mounted
external disk. Due to the flakiness of the old (pre release 8) FreeBSD USB stack, I don't do this on my own machine: all
you need to do is to turn the drive off before umounting it, and the system crashes,
requiring hours of fsck and various boot-time tweaks that I still haven't ironed out.
So I do it across the network to lagoon, Yvonne's machine. So far we haven't crashed
the system.
Today was the day, though. Not only did the system crash—in the middle of the
backup—but I couldn't reboot. The boot loader found the second-level boot, but then it
couldn't find anything else on the disk. Took it and put it into my test box, and
discovered that the disk was well and truly trashed. Most of the root directory looked OK,
but some parts were completely wiped:
=== root@swamp (/dev/ttyp0) /lagoon 2 -> ls -l ls: etc: Value too large to be stored in data type
ls: var: Value too large to be stored in data type
total 320045
d--------- 5124 4096 80330752 17592473591808 Aug 12 1980 libexec
d--------- 5124 4096 80330752 17592473591808 Aug 12 1980 tmp
It took a while to realise that this was the same motherboard that had caused me
untold grief in cvr2, and which I had removed
because of that. The symptoms match exactly: somehow the USB bus is broken, and using it
results in overwritten disks. I really need to label these things better.
But how to recover? I only had one functional motherboard left, the dual processor machine
I bought a couple of months ago to
replace the one damaged by Powercorlast November. Clearly it's good enough: it's a 2.8 GHz
dual processor machine with 2 GB memory, and for the first time ever I've been able to build
a usable machine with no plug-in cards at all: everything is on the motherboard. Much
faster than either of the boards she had before.
But what should I put on it? Building a new FreeBSD machine is becoming more and more of an issue. I had planned and started an
upgrade anyway last month, but given it
up because of some weird bug that seems to bite only me. And though building ports from
source seems like a good idea, and in the past I've supported it, it's becoming more and
more fragile in the course of time: I would bet that I wouldn't be able to install all the
software I wanted to without some configuration or compilation issues. And somehow, though
processors are now nearly 1000 times faster than they were when I started using FreeBSD,
building software takes about the same time. As I wrote in 1993:
How long does it take?
It is very difficult to gauge the length of time a port will take to complete. If a port
takes a long time, it's not usually because of the speed of the machine you use: few
packages take more than a few hours to compile on a fast workstation. Even the complete
X11R6 windowing system takes only about 4 hours on a 66 MHz Intel 486 PC.
So I decided to give her my Ubuntu disk, which
has most of the stuff on it already. If she likes it, it'll stay that way. Somehow it's
like the end of an era: using Linux instead of BSD simply because it's easier. Of course,
it's not as simple as that, but it seems to be the first step. Sad, somehow. In my
experience, BSD has always been cleaner and more reliable, but it's getting a bit rough at
the edges.
It's been a while since we made a lamb biriani, so had another crack at it. It's a lot of work, and it kept me busy most of the
afternoon. Yvonne helped a bit and found some weaknesses in
my description. It's always good to have somebody else test your docco for you. Emptied a
surprising number of jars of various ingredients, including 2 jars of cardamoms and two cans
of ghee—in each case almost empty.
Somehow I'm not completely happy with the results. I need to think why. Chris and Yvonne
were both happy with it.
The first part of migrating Yvonne to Ubuntu was relatively simple: I already had a disk with an
installation of 9.10 on it, so all I had to do was add a user. But that immediately caused
problems with NFS, because user and group IDs were different. Presumably as a
result, mutt refused to access her (NFS-mounted) inbox. Spent a while trying to work
out how to change that, and in the end just edited /etc/passwd, /etc/shadow
and /etc/group. That included swapping the groups disk
and mail. disk is used only for /var/lib/dumpdates, it seems,
which is strange in view of the fact that the system doesn't have dump
and restore. Hopefully the change of group won't mess up mail.
FreeBSD does have dump
and restore, and that's how I backed up Yvonne's files. Started with a restore of
the level 0 dump I took at the beginning of the month (on a FreeBSD system, of
course; dump and restore are horribly system-dependent), and after working its
way through 6 GB of compressed dump, restore returned—nothing. Only an empty
directory tree.
Much more investigation revealed that the dump had somehow been aborted in the middle,
possibly by the machine powering down (no, there was no power failure at the time). Went back to the old
disk, which I thought had been readable yesterday, and discovered that the inode
for /home/yvonne was completely trashed. Did that happen since yesterday? It's hard
to believe, since I had mounted the disk read-only. Tried fsck, but it left nothing
of /home/yvonne and very little of interest in /home/lost+found. So back to
the dumps. The level 0 dump of 1 April was OK, and I had a level 1 dump on 30 April and
another on 14 May, so the only data I might have lost would have been between 30 April and 1
May, probably very little. Spent much time restoring that.
But mutt still refused to open the inbox read-write. Did a bit of searching, but
everything pointed towards incorrect permissions. The permissions were right, and there
were no error messages, but it came up with the folder read-only, and refused to change it.
At the very least there should have been a message. Finally gave up and built a debug
version of mutt and went through with gdb. To its credit, the code is
relatively readable—something that's becoming rare nowadays—and after a while I
found it invoking mutt_dotlock, which failed. And no error message. I should fix
that when I have the rest of the system up and running. Running the same command from the
shell worked. And then I re-read the permissions on mutt_dotlock. They should be
like this:
wh is a little bash function that
finds all occurrences of an executable and shows them in PATH order with the inode
number. I added the first link for /usr/local/bin/mutt_dotlock because that's where
FreeBSD expects to find it, and I used the
FreeBSD version for debugging. The problem, of course, was the setgid bit
(the s in -rwxr-sr-x). I had -rwxr-xr-x, so it couldn't lock.
But why doesn't it say anything?
That's not the only thing, though. This version of mutt is newer, and they've
changed the syntax of the alternates command. Previously it was a variable, and
now it's a command, and the syntax itself is different:
-set alternates="yvonne@.*" # Names that raise the + flag
+alternates yvonne@ # Names that raise the + flag
It would have been nice to at least have an explanation in the error message, preferably
pointing to the wiki entry describing the changes.
By the end of the day, we had a number of things working, and in fact not too many which
still needed attention:
For some reason, Yvonne's firefox doesn't understand Emacs key bindings any more.
The trick used to be to have a file ~/.gtkrc-2.0 with the line
gtk-key-theme-name = "Emacs"
But that didn't seem to work for her. Confusingly, it does for me.
The mouse didn't work correctly. That's to be expected: since the sad demise of the
middle button, I've taken to remapping it to the left side key (for operation by the
right thumb). I still need to find out how to do that with Linux.
And then, of course, there were a number of programs to install, including xv,
which will be fun: it's too old for Linux, so I'll have to build it myself.
After discovering the SMTP blockage at SkyMesh, I also discovered that they had forgotten to block some systems, those connected to
Pipe Networks, and I used that to send my
mail. But maybe somebody at SkyMesh is reading this diary; sometime on Thursday afternoon
they blocked that too, and once again I didn't find out until today. What the hell. I once
had a tunnel for sending mail (I forget why; maybe it was because I was sending it from
remote locations where SMTP was blocked), so reinstated that. That'll work until we get
things working properly.
That loop in the middle of the stem looks like the beginnings of a new branch, complete with
the leaves.
Also brought in a flower stem of one of the ginger-like plants that we have in the garden.
I don't know what they're called, but they have a particularly sweet scent, and I thought
they'd be good in the house. But they don't seem to last. These photos were taken a few
hours after I brought them in, and already the flowers are wilting:
The first step was to install a surprising number of packages. By the time I had
finished, I had installed patch, libx11-dev, libc6-dev,
libtiff-dev, zlib-bin, zlibc, libpng3,
libpng12-dev, libpng12-0, zlib1g, zlib1g-dev,
libjpeg62-dev, libtiff4-dev, libtiff4,
libjasper-dev, libxt-dev, xutils-dev, csh,
exif and libimage-exiftool-perl. After that, decided to start with the
FreeBSD port and re-run configure. Tried to compile and got a missing header
file machine/endian.h. There's a good reason it's missing: it in Linux it's
(/usr/include/)endian.h, not (/usr/include/)machine/endian.h.
The latter is a BSDism. Somehow the original Makefile had been completely replaced
(built with imake, no less), and it defined a variable CSRG_BASED. To get
things right, I had to extract and patch the sources, and then build again. And then I got
an error in tiff/. Looking at tiff/Makefile, it looks like another FreeBSD
hack:
# $Header: /usr/people/sam/tiff/libtiff/RCS/Makefile.sun,v 1.46 93/08/25 09:05:53 sam Exp $
# modified for use with XV by jhb 4/26/94
That's Sam Leffler and John Baldwin, both of the FreeBSD project. But it turns out that
that was in the original. The real problem was running ranlib (remember that?), and
also a redefined sys_errlist. Ended up doing a couple of little patches, and then
the thing built. Quite a hack, and certainly not even close to the FreeBSD ports system.
But that was to be expected.
Checked and found yes, there's a cleandir in that directory, and it's marked
executable. So why didn't it run?
=== root@zaphod (/dev/pts/11) /src/FreeBSD/ports/graphics/xv/work/xv-3.10a 109 -> ./cleandir jpeg bash: ./cleandir: /bin/csh: bad interpreter: No such file or directory
So why didn't that message appear from make? It makes things very confusing.
Installing csh (another indication of the age of this code) fixed that.
Another unexpected issue was that the resulting executable didn't handle screen refresh
correctly. When we first tried it, it didn't change images when a new one was selected; we
had to physically move the window before it would refresh. While I was pondering that one,
it “fixed” itself, but in case I have to follow up, it seems that xmon might be what
I'm looking for—another program for which a port exists in FreeBSD, but not in
Ubuntu.
Games
We don't play many games, but years ago I hacked Keith Packard's kklondike, and since
then it seems to have disappeared. Found other card games, of course, and
installed kpat, xpat2 and aisleriot, but the two Klondike games
seem rather pedestrian in comparison: you need to move every card. I'll have to find out
how to get hold of kklondike. It's not in the Ubuntu packages, but maybe it's
available elsewhere.
In the course of my career I have spent more time in aeroplanes than most people. In the
fractionally more than a year from 8 May 2005 to 17 May 2006 I flew to North America twice and to Europe 4 times, including once
round the world—52 individual flights, or just under one per week. But flying is not
what it used to be. Over the decades, terrorists have killed hundreds of innocent people,
and round the world authorities continually ramp up security. In the process, they find
occupations for brainless morons who revel in their ability to make passengers' life a
misery. In the 1980s I used to enjoy travel, certainly helped
by Pan Am's policy of giving free first
class upgrades to their frequent fliers. But gradually I came to dread the start of a new
trip. Once I was under way it was OK, but the dread came again every time. By
this time four years ago I was ready to scream at the
treatment I was getting. And I haven't flown since. At the beginning of the last time I
didn't fly for four years, I was four years old. Another indication of my change in
lifestyle.
But why do I dread flying? It's not just the security thugs who make people's life a
misery; the airlines seem to be doing their part too. Clearly all this mistreatment costs a
lot of money, and they're in trouble. You can understand that, but not that they think that
mistreating their passengers is a way to get more passengers.
The real thing, though, is: yes, the terrorists have won. People are scared of them, and
they've crippled the air industry. But despite everything that has happened, air travel is
an order of magnitude safer than other forms of travel. I won't even start about horses,
but car accidents kill orders of magnitude more people than terrorists ever have. And
nobody pays much attention to ABC's emetic news
bulletin about how yet another driver collided with a tree (how do they aim so well?), or
VicRoads' stupid slogans. Clearly they should take some lessons from the terrorists.
Tomorrow I'm doing a presentation about my weather
station software, and finished the slides.
It's funny how starting these things is the most difficult; refining them, even adding
images, is relatively trivial.
Last month Peter Jeremy brought me
some curry plant suckers, and we
planted them. How are they doing? Hard to say. They're not exactly thriving, but they've
been in the pots for over a month now, and they don't look any the worse for it. So I
suspect things will pick up when spring comes. Here the original plant and then two of my
pots:
Baked some bread and also made some kimchi today. It's interesting that both of them
use lactobacillus. And then I had
an idea: I've been pondering the relatively little lift that my current sourdough starter
gives, and Sue Blake was talking recently about making one from scratch, so took some ripe
kimchi and added flour and water:
A couple of days ago I tried out the ALDI GPS
navigator without first reading the instructions. Clearly this is a case of “If all
else fails, read the manual”. But that makes an assumption:
First I looked at the booklet entitled “Go cruise Navigation System”, 42 pages
in length. It refers to a stylus that wasn't delivered. It tells me how to play MP3s. It
tells me how to play videos. It tells me how to read e-books. It tells me how to view
photos. It doesn't tell me how to navigate.
So I looked at the other booklet, entitled “Quick start guide”. It contains 21
pages, including 6 of EULA containing the startling agreements:
1.1 This Agreement has been entered into by and between Nav N Go Kft. (registered seat: 23
Bérc utca, H-1016 Budapest, Hungary; Company reg.no.: 01-09-891838) as Licensor
(hereinafter: Licensor) and You as the User (hereinafter: User; the User and the Licensor
jointly referred to as: Parties) in subject of the use of the software product specified
in this Agreement.
...
9.5 The parties hereby agree that - depending on the nature of the dispute - either the
Pest Central District Court (Pesti Központi Kerületi Bíróság) or the Metropolitan Court of
Budapest (Fvárosi Bíróság) will have exclusive jurisdiction to rule on any disputes
arising in connection with this Agreement.
The Quick start guide is just that. It contains less information than I have already
gleaned by playing with the document. Clearly the meat of the instructions is in the
Microsoft “Word” formatted manual. So went to look at that. My Microsoft laptop was in the car waiting to go to the
University this evening, so I read it with Apple's “TextEdit”, which, once I
persuaded it to show more than a tiny cutout of the document, showed all sorts of strange
things:
Gave up on that and tried with OpenOffice on Yvonne's computer. With a bit of mouse-pushing that went
quite well, but the results were less than useful:
In fact, all images (including presumably icon images in this example) were missing. The
document was completely useless. Bad marks for “TextEdit” and OpenOffice? I
asked somebody with a Microsoft “Word” to convert it for me. But that didn't
work either: the original document is broken. The strange text in the
“TextEdit” document is apparently error messages. The best we can guess is that
the document refers to images what were in the same directory as the source file, but they
weren't included on the web. So it's completely useless. About the only thing of interest
was the statement:
Using a stylus
You do not need a stylus to use Navigation Software. Tap the buttons and the map with your
fingertips.
Called up the help line again and after 11 minutes was connected with Sam, with whom I had
spoken last week. When she heard my name, we were disconnected. Coincidence? Who knows?
After another 13 minutes I was connected with Jake, who wanted my email address to send the
document. I'm reluctant to do that sort of thing with people from the Microsoft space: most
of the time it ends up as spam, or in a format that is so illegible that I don't want to
read it. But clearly this was a time to make an exception. And, of course, no
documentation came.
My credit card expires at the end of the month. Rather against my intentions, I didn't
change the supplier (it's relatively expensive, gives me Frequent Flyer points, which I
clearly don't need any more, and the security is ridiculous), so the new card expires in
2014. And I've received a number of reminders to update online information, like this one
from eBay:
This is a courtesy reminder that the following credit/debit card on file for your eBay account will soon expire:
...
To update your credit/debit card information:
1. Go to the eBay Home page.
2. Click My eBay at the top of the page, and sign in with your eBay User ID and password.
3. Click the "Seller Account" link (beneath My Account in the left navigation menu).
4. Follow the instructions for updating your payment method.
OK, that made sense, so I followed the instructions. At point 3, there was no "My Account"
link on the left, just an "Account" tab at the top. That had a drop-down menu with "Seller
Account", so I selected that. Where are the instructions? I don't know. All I saw was a
section "Payment methods for Seller Fees", telling me that I have only PayPal as a payment
method. So what is this all about? Yet another case of loose ends and inaccurate
documentation? Interestingly, PayPal hasn't sent me a reminder.
Into town with Chris to talk to BLUG about my
weather station software, and also took the
opportunity to try out the GPS navigator in earnest. It's really annoying that I can't
modify the destination: I give an address or geographical coordinates. Clearly the former
is easier, but it's often not very accurate. How do I move the destination? In Google Maps you just drag the point to where you want it.
But that doesn't seem to work here. Also, the speed indication was missing in the map
display, though I had seen it before. Stopped a couple of times to play with the thing, but
no joy with the speed indication.
What we did see was steam coming out of the bonnet of the car. Opened it up, and the
steam was coming from the vicinity of the cylinder head. Looks like a head gasket, which
would probably mean the end of this 20-year-old car. What timing! Especially as Josh had
contacted me to say that he would be late, and that I should start without him. Called up
Yvonne and got her to come and pick us up, and then on to the
University, arriving only 8 minutes late, but then ran into trouble with the projector,
since only Josh knew how to set it up. Ended up using my own, which I had brought with me
for exactly that eventuality.
Apart from that, another presentation. There were fewer people present than I had expected,
but as usual we went over time, and we had quite a discussion.
It's leading in to the inlet manifold. I wonder what that's for. It could be for heating
it, except that I thought that that was no longer necessary with fuel-injected engines, and
there's no exit pipe. Down to see Paul Sperber at Ballarat Automotive
in Sebastopol to leave the
car for repair, but he did it on the spot. 15 minutes later and $27 poorer, I was off
again. A good thing to: Paul confirmed my suspicions that a new head gasket would cost in
the order of $1000, more than the car is worth.
On the way into Sebastopol, continued playing with the GPS navigator, or more specifically
the navigation software. It does some things
quite well, others not nearly as well. For example, it has misplaced the town
of Enfield by about 3 km. On
the other hand, it knows many of the speed limits along the way, and informs you with an
indication on the screen and also a verbal message if you go more than 1 or 2 km/h over the
limit—but only then. These limits, when it gets them right, are very accurate, within
a few metres of the sign. But wouldn't it be a good idea to have at least the option to
display the limit on the screen all the time, even if you're not exceeding it?
There were also some minor inaccuracies in the road data. The turnoff from the Midland
Highway to Ballarat-Colac road has recently been remodeled to make it more difficult to turn
into (“road safety”, I suppose), and the map data doesn't seem to have picked it
up: when I turned off, I got the message “recalculating”, which apparently means
“you have deviated from my route, and I need to start again to work out how to get you
to your destination”. It did it again
in Napoleons, where I turned off to see
if the roadside plant sales had anything interesting (they didn't). In each case, it also
changed from 3D display mode to 2D display mode, which I can only consider to be a bug.
Still no mail from Tempo Australia, of course, so decided to follow up on the presumed name
of the manufacturer of the software, Nav N Go. That proved much more successful; in conjunction with the
“About” screen on the navigator, which tells me that the software is “GPS
IN-CAR Navigation 8.3.4.135494”, I was able to deduce that it's really Nav N Go iGO 8 (you have
to love these StudLycApS), and
finally found the correct
documentation, a slightly different version of the broken document that I downloaded
yesterday. Here that document, then the correct version:
It's not exactly the same: the name “Nav N Go iGO 8” has been replaced with the
more generic “Navigation Software”, and some features described in the manual
are not present, notably the choice between 3D maps, 2D maps with north at the top, and 2D
maps with travel direction at the top. The device only offers one 2D representation, which
seems to switch top from north to travel direction for reasons I haven't understood yet.
Also, the manual shows a couple of alternative icons, presumably used in different devices.
But it tells me much more about the device, and makes it look better than I had
thought.
Chris tells me that SpamAssassin on
our external mail site is still claiming that all mail is from far in the future. We ran
into this problem at the beginning of the year, and I
fixed it there, but forgot to do it on the external server, where only Chris uses
SpamAssassin. Did that today and understood the issues of the completion codes
of sa-update. I ran it with the -D (debug, really verbose) option, and the
first time it completed with:
For that length of time, the CPU usage is quite reasonable. Maybe the recently spawned
child does most of the work.
On the home front, finally SkyMesh have
unblocked SMTP for me, and they've put reverse mapping on my external gateway address
(sat-gw-ext.lemis.com), and even the dropouts have become less frequent. All fine?
Not quite. Yvonne came in with details of a message that she
had sent, both to me and to a mailing list, and neither person got it. Took a look at the
logs and found:
May 19 12:52:56 dereel postfix/smtpd[36846]: connect from zaphod.lemis.com[192.109.197.131]
May 19 12:52:56 dereel postfix/smtpd[36846]: 6CBD5A1015: client=zaphod.lemis.com[192.109.197.131]
May 19 12:52:56 dereel postfix/cleanup[36849]: 6CBD5A1015: message-id=<20100519025235.GB10754@lemis.com>
May 19 12:52:56 dereel postfix/smtpd[36846]: disconnect from zaphod.lemis.com[192.109.197.131]
May 19 12:52:56 dereel postfix/qmgr[63260]: 6CBD5A1015: from=<yvonne@lemis.com>, size=2490, nrcpt=2 (queue active)
May 19 12:52:56 dereel spamd[308]: spamd: connection from localhost [127.0.0.1] at port 56757
May 19 12:52:56 dereel spamd[308]: spamd: setuid to grog succeeded
May 19 12:52:56 dereel spamd[308]: spamd: processing message <20100519025235.GB10754@lemis.com> for grog:1004
May 19 12:53:04 dereel spamd[308]: spamd: clean message (0.0/3.0) for grog:1004 in 7.6 seconds, 2422 bytes.
May 19 12:53:04 dereel spamd[308]: spamd: result: . 0 - scantime=7.6,size=2422,user=grog,uid=1004,required_score=3.0,rhost=localhost,raddr=127.0.0.1,rport=56757,mid=<20100519025235.GB10754@lemis.com>,bayes=0.007218,autolearn=ham
May 19 12:53:04 dereel postfix/pipe[36851]: 6CBD5A1015: to=<grog@lemis.com>, relay=spamassassin, delay=7.7, delays=0.07/0/0/7.6, dsn=2.0.0, status=sent (delivered via spamassassin service)
May 19 12:53:04 dereel postfix/pipe[36851]: 6CBD5A1015: to=<clickerreiter@yahoogroups.de>, relay=spamassassin, delay=7.7, delays=0.07/0/0/7.6, dsn=2.0.0, status=sent (delivered via spamassassin service)
May 19 12:53:04 dereel postfix/qmgr[63260]: 6CBD5A1015: removed
Delivered via SpamAssassin service? Where? I had to look at the log of a successful
delivery to see the difference:
May 19 08:22:23 dereel postfix/smtpd[98024]: connect from zaphod.lemis.com[192.109.197.131]
May 19 08:22:23 dereel postfix/smtpd[98024]: 59B35A1015: client=zaphod.lemis.com[192.109.197.131]
May 19 08:22:23 dereel postfix/cleanup[98027]: 59B35A1015: message-id=<20100518222205.GP10754@lemis.com>
May 19 08:22:23 dereel postfix/smtpd[98024]: disconnect from zaphod.lemis.com[192.109.197.131]
May 19 08:22:23 dereel postfix/qmgr[63260]: 59B35A1015: from=<yvonne@lemis.com>, size=2736, nrcpt=1 (queue active)
May 19 08:22:23 dereel spamd[67650]: spamd: connection from localhost [127.0.0.1] at port 57034
May 19 08:22:23 dereel spamd[67650]: spamd: handle_user unable to find user: 'GaitedHorse'
May 19 08:22:23 dereel spamd[67650]: spamd: still running as root: user not specified with -u, not found, or set to root, falling back to nobody
May 19 08:22:23 dereel spamd[67650]: spamd: processing message <20100518222205.GP10754@lemis.com> for GaitedHorse:65534
May 19 08:22:29 dereel spamd[67650]: spamd: clean message (0.0/3.0) for GaitedHorse:65534 in 5.7 seconds, 2663 bytes.
May 19 08:22:29 dereel spamd[67650]: spamd: result: . 0 - scantime=5.7,size=2663,user=GaitedHorse,uid=65534, required_score=3.0,rhost=localhost,raddr=127.0.0.1,rport=57034,mid=<20100518222205.GP10754@lemis.com>,autolearn=failed
May 19 08:22:29 dereel postfix/pickup[95241]: 15D3EA1098: uid=65534 from=<yvonne@lemis.com>
May 19 08:22:29 dereel postfix/pipe[98028]: 59B35A1015: to=<GaitedHorse@yahoogroups.com>, relay=spamassassin, delay=5.7, delays=0.01/0/0/5.7, dsn=2.0.0, status=sent (delivered via spamassassin service)
May 19 08:22:29 dereel postfix/qmgr[63260]: 59B35A1015: removed
May 19 08:22:29 dereel postfix/cleanup[98027]: 15D3EA1098: message-id=<20100518222205.GP10754@lemis.com>
May 19 08:22:29 dereel postfix/qmgr[63260]: 15D3EA1098: from=<yvonne@lemis.com>, size=3031, nrcpt=1 (queue active)
May 19 08:22:29 dereel spamd[1119]: prefork: child states: II
May 19 08:22:33 dereel postfix/smtp[98037]: 15D3EA1098: to=<GaitedHorse@yahoogroups.com>, relay=mail[208.86.224.149]:25, delay=4.9, delays=0.01/0/2/2.8, dsn=2.0.0, status=sent (250 2.0.0 Ok: queued as 9EE533BAA5)
May 19 08:22:33 dereel postfix/qmgr[63260]: 15D3EA1098: removed
This message, too, was delivered by the Spamassassin service, but then it got another
connection to send the message on. It seems that this doesn't happen in cases where both I
and an external list are involved. At Stephen Rothwell's suggestion, added a line
to main.cf:
# Only send one recipient at a time to spamassassin.
spamassassin_destination_recipient_limit = 1
That seems to have done the trick. But what a pain!
Finally I don't have so much computer work to do, and the weather was sunny and almost
without wind, ideal weather for
spraying glyphosate in the garden.
Did that, and also planted a few plants, notably
the grevilleas that we have been trying
to propagate from cuttings. These rectangular cross-sections plastic tubes are really not
very good: it's almost impossible to get the soil out in one piece, and two of the three
grevilleas that I took out came with almost no soil at all and very little roots. I wonder
why; they've been in there for a while, and for a time they looked as if they were growing.
It has become a lot more liquid, and it smells
of Lactobacillus, so maybe that
side of things is working, but judging by the lack of bubbles, there doesn't seem to be any
yeast in it. If I leave it open to the air I might get some, but why not help on a little?
Put in a gram or so of baker's yeast to see what happens. What should happen is that
the yeast will die in the acid surroundings, but maybe there's some part of the yeast that
can handle it. This isn't the same as just adding yeast to a sourdough to bake bread of
course, any more than this starter would be kimchi: whatever yeast is left in the sourdough
will be a component of the sourdough after a few generations. In the evening, yes, it had
risen. But that might be the only time.
Occasionally we get a TV recording that is really messed up, and that's what we had at the
beginning of the week. And the next day. And yesterday. Gradually it dawned on me that it
was only with SBS. Stopped MythTV and ran tzap to confirm that yes, the signal was
much weaker than other channels. Called up SBS on 1 800 500 727 and was connected to Alan,
who sounded both like he knew what he was talking about and that he cared. But he didn't
have any other reports of poor signal quality, and their own records didn't show anything, so he
wanted me to find at least two other people having trouble with SBS reception. He asked if
the reception was still bad right now (it had been an hour earlier), but I was running some
comparative runs of tzap and couldn't tell him at the time. He said he'd call back
later.
In the meantime, decided to go to the neighbours and see if they had digital TV. First to
Helen and Robert across the road. Yes, they have digital TV (and Helen was watching it),
but she didn't know if they had SBS or not. They did, of course, and the reception was
perfect.
So: problem must be in our place. Antenna? Connections? Tried reconnecting the latter,
without effect. Then I tried the “Microsoft solution”: reboot. To my immense
surprise, that did the trick. But why? I stopped and restarted the processes, and you
wouldn't think that something like relative signal strength would be a software problem.
Anyway, another one to remember.
And yes, Alan did call back, and I told him the news. He must be the first sensible and
helpful person I've run into at SBS.
Fed yesterday's sourdough again twice today, the way it's supposed to be, and each time got
a fair amount of bubbling. That can't be from the sugar in the original yeast, so it looks
as if some kind of symbiosis has already started. Here's the result before and after the
“feeding” in the evening:
Mail from Zoltán Rajnai of Nav N Go, referring to my comments about their
software. I had considered it a bug that it switches back to 2D view when it
recalculates a route. Zoltán tells me it's not a bug, it's a feature, and how to configure
it and turn it off:
There's a feature called 'overview mode', where the device switches to 2D North-up view
with a pretty low zoom level to give you an overview of the route.
This happens if the distance to the next manuever is above than a configured value.
You can set the configuration for the overview mode in Settings / Navigation settings /
Overview mode. There's an option to turn the feature off completely, to set the distance
of the next manuever where the overview should kick-in, as well as the your desired zoom
level for the overview mode.
It's not quite like that on my navigator, but clearly the people who built it have adapted
the software somewhat. Still, the option's there, and though it doesn't make much sense to
me (yet), the fact that I can turn it off if I don't like it means that I don't have any
cause for complaint. It's also nice to get feedback—a far cry from Tempo Australia, who still haven't replied to
my email.
One of the potato varieties we planted in the spring were Kipfler, which seems to come from
Austria (and not Germany, as many pages claim). They're supposed to be particularly tasty.
But they're double the price of other seed potatoes (a moot point, I suppose, since you only
buy them once), and the shape makes them difficult to process. Today we ate some: after 30
minutes, they were still hard, and in the end we decided they're not worth the trouble. The
other kind (“Dutch cream”, which may come from Holland) is easier in all
respects, and if anything I think it tastes better.
We had a very light one a couple of days ago, but it doesn't seem to have done much harm
beyond the Pandorea
pandorana (Wonga Wonga vine), which is looking a little unhappy already:
Came in to the office this morning to find yet more erroneous values from the weather
station, which was claiming a low temperature of -3276.7° and a dew point temperature of
-3301.2° for a period of 5 minutes between 6:49 and 6:54. Went and changed them
to NULL, as I've been doing with other invalid readings, but in the past invalid
readings involved more than just the temperatures, and this time only the temperatures
looked wrong. Then I noticed the temperatures either side of this time period: 0.0°. And
looking at the reported temperature again, it looked more than familiar: 0x8001.
So this brain-damaged weather station is reporting temperatures in sign/magnitude format
instead of twos-complement, and the temperature was really -0.1°. More source code mods
needed. The incorrect value for the dew point is based on the incorrect temperature.
It's bubbling away nicely, and I can't imagine this is just the bread yeast. It must be
getting sugar from somewhere, and I don't know of
any Saccharomyces that can do that.
Once again, this isn't a comparison of non-optimized and optimized: it's a comparison of
optimized and batch optimized. This happened to two photos, both from a different camera
(the Olympus E-30 for most of them, and the Nikon
“Coolpix” L1 for these two photos only). I wonder if there's a
connection.
Took my verandah panorama with the wrong lens this morning, the Zuiko Digital ED
12-60mm F2.8-4.0 SWD at 12 mm instead of the Zuiko Digital ED
9-18mm F4.0-5.6 at 9 mm. I repeated the shots when I realized my error, but decided
to process the original one as well. The results weren't what I expected:
The individual component photos were taken at the same angles, so the top one (12 mm) is
wider. That's to be expected, but the perspective is completely different. In the top
photo, the Crassula falcata on
the table on the right is directly in a line with
the Agapanthus on the floor, but in
the bottom photo it's to the right. Spent a lot of time puzzling about this one, but I've
come to the conclusion that I must have had the camera just a fraction further to the left
for the first set of photos. It's amazing how small differences make the resultant photo
look completely different.
The autumn sunshine outside gave me the idea to have another go at creating an HDR photo
looking out of the house. Unfortunately, the brilliance of the sun outside suffers as a
result:
Round then, we had to turn around, because Carlotta was getting tired. But on the way back,
Darah stumbled a couple of times, the same way she had last time, and at one point (fortunately just before we got home) she was in enough
pain that I had to dismount. It looks like there is something wrong there; something for
the vet to look at. Hopefully it's not too serious.
Did some half-hearted work in the garden in the afternoon. Tidied up the area round the
bird bath: removed the grass bush that had sprung up there and moved the remains to the
south-east corner of the garden. I don't know if it will survive (or would have if we
hadn't moved it, for that matter): it's not clear whether they're annual or perennial. Also
removed the dead creepers from the Eiffel Tower, where
the pandoreas are still looking happy,
and did some much-needed weeding at the extreme east.
Then more thinking about the greenhouse. I
still don't understand some of the details. We've decided to leave the left-over components
out for the moment, in the hope that they can be attached later if necessary, and
concentrate on getting the glass in. But we still have the question of the roof. We've
already established that things look wrong at the bottom of the surface. In particular,
there's nowhere to seat the glass:
David Yeardley should know, but he is currently in the Persian Gulf and unable to look for
himself—for some reason his company won't allow web access, or allow email of more
than 50 kB, which makes it difficult to send photos. But Chris contacted him, and he
replied that things looked correct, and that we should attach the glass at the bottom
cross-member with some of the metal S-strips that are used for the glass. Found a sheet of
glass (the rest is still at the Yeardley's) and put it in place, held by the
thicker-than-normal bolts we bought. Problem: the cross-member is bent down on the inside,
so the S-strips won't fit. And the thickness of the rafters means that there's an air gap
between the glass and the cross-member. Given that the sides have a rubber seal to keep the
air out, this seems wrong:
I still don't know what's wrong with the greenhouse, but finally got round to doing what I
should have done a long time ago and called up the manufacturer in Dunedin. Spoke to Grant,
who told me that they had had the assembly instructions online (locally, it seems) until
about a month ago, when the machine with the information on it died—implicitly, there
was no backup. Growl. If I had called them up at the beginning, I might now have the
assembly instructions. Sent him an email with more details. Hopefully they can find
something.
Seekh kebab for dinner this evening.
It's a lot of work, and somehow the balance is still wrong. Yvonne and I shared the work, and I decided to mix the final mixture in
the Kenwood Chef mixer. You'd think
it were ideal, but in fact it's pretty useless. After 5 minutes of mixing and manually
folding in the various components, the best we had was:
Off to Melbourne shopping today. This
was also the first time I used the GPS navigator in earnest. The result? Parts of it are
excellent.
I had intended to go to MSY in Brooklyn,
but—surprisingly—we got off so early that we would have arrived in Brooklyn
before they opened, so decided to go
to Springvale first instead.
The navigator wanted to take me to Melbourne
via Ballarat, not my favourite way via Geelong, and I had to tell it to go
via Bannockburn. To my
surprise, it found a better way around Rokewood than I
had been able to find for myself, despite some careful examination of the maps. But in
Bannockburn it got confused when I turned left instead of right after the level crossing,
recalculated, and told me to take a right turn about 300 metres later—but there was no
road. After Bannockburn it forgot all about speed limits, and it didn't know the way I took
from Bannockburn to Lara. Then
things were good again: it does quite a good job in the city, though even there it reports
the speed limits very inaccurately. But it knows about speed cameras and red light
cameras—it proves that Melbourne is full of them, and I wasn't able to recognize many
of them. That warning alone is worth the price of the device.
Got to the Springvale shopping centre without difficulty, and parked in front of the Nan
Yang supermarket. The only problem: we had wanted to go to Balkan Meats and Smallgoods
first. Did our shopping and then entered the correct destination. Off we went, doing a
couple of U turns where I suspect we weren't allowed, and ended up being brought back to a
place 20 metres behind our start. It seems that I had set the destination to Balkan
Meats and Smallgoods, but they appear to have closed down. It would make sense for the
device to tell you when the drive (in this case about 2.5 km) is that much longer than going
by foot.
Then off to Hindustan Imports in Dandenong South, which worked well,
and then back with the intention to continue to the Wursthütte, but on the way decided that we could
make it to the Queen Victoria Market before it
closed, so changed the destination on the fly.
That had mixed results: it took me exactly to the destination, getting through the mess
south of the city much better than I have ever done on my own. But it dumped us about 500 m
from the market: I hadn't entered the address correctly, and I'm still not sure how to do so
without a street address. Finding my way manually was an order of magnitude more pain than
the navigator. I think it's proved its utility, though I still have a lot to learn about
how to use it.
An obvious indication of that was the final destination, MSY, I had put in a route via
Bannockburn, since I had originally intended to go there first—so it wanted to take me
first to Bannockburn and then to Brooklyn. Clearly that's not the way to force a preferred
route.
Once I entered the destination address, it wanted to take me over a toll road, which I
ignored. Finally got there with relatively little difficulty, though it's irritating that
moving from a frontage road to the main road is reported as “turn right, then turn
left”. To be investigated. At least both routes looked an order of magnitude more
sensible than what Google Maps later calculated. The way I went was 10.4 km and estimated 11:28
minutes, the “fast” method on the navigator was 13.3 km and estimated 11:26 (!)
minutes. Google Maps had distances of 11.7 km (21 minutes) and 17.4 km (27 minutes).
Of course, what's in an estimate? I've already established that Google Maps is far too
conservative, but it seems that the navigator is conservative too. On our way home it
estimated an arrival time of 16:22; in fact, we arrived at about 16:05. It kept correcting
the arrival time as we went, of course. I wonder if it's also learning.
MSY is a bit of a pain. The first time I went
there, in Clayton, I had over
an hour's wait. The couple of times I've been in Brooklyn were much faster, and today was
no exception, but in the past they never had exactly what I had chosen. So today I went in
and asked for their cheapest AMD processor, a matching motherboard and 1 GB of memory. All
went very quickly, total cost $173. It's still amazing how cheap these things have become.
When I started in the business, the Control Data 7600 was the measure of high performance: 2 CPUs with a 27.5 ns cycle
time (36 MHz), each with 10 functional units, giving a theoretical maximum throughput of
about 730 MIPS. Its memory maxed out at a little under 9 MB.
The Athlon X2 245 I bought today also has 2 CPUs, each with 4 functional units and a cycle
time of 345 ps (2.9 GHz). Using the same invalid method as for the 7600, it would thus
theoretically be able to perform about 23 GIPS. I only bought 1 GB of memory, or rather
more than 100 times the maximum memory of the 7600.
Both of these sets of values are simplistic and unattainable; if they're equally
unattainable, they can still serve as a crude comparison. It really does seem that the
Athlon is about 20 times as fast as the 7600. That wouldn't be so surprising in itself, but
the fact that I can go into a nondescript shop and buy one almost without thinking about it
still takes some getting used to.
Only yesterday I bought a low-end computer motherboard and memory for $173, and a couple of
weeks ago the ALDI GPS navigator for $89. It
occurs to me that the latter contains detailed maps of all the major cities in Australia, as
well as tolerable maps of the country. If I had just bought the maps I need—say, the
VicRoads
country street directory, Melway,
and Sydway and Adelaide street directories, I would probably
be in for double that. Not to mention the lessened chance of being caught in a speed trap,
which costs over $200.
What about map updates? The navigator conveniently points me at https://www.naviextras.com/, where I can buy lots of maps for prices comparable
to or exceeding the cost of the navigator. What I can't find there is an update for my
current maps. It almost looks as if it will be cheaper to buy a new navigator (with maps)
than the maps by themselves.
Today I got yet another example. I've subscribed to Heise-Verlags new quarterly magazine on digital photography, and as a result received
a “free gift” of a 4 GB SDHC
card, sent airmail from Germany, with Yet Another Way of writing my address:
The card is useful—it's a class 6 card. The ones we have are 8 GB, but they're also
class 2, and I was able to confirm that the new one was considerably faster in Yvonne's Kodak M1093 IS. But
they don't cost much—in fact, less than the 12 € postage.
Yesterday's purchases needed to be stored somewhere, of course. Many things needed to be
put in the pantry, which is not the world's largest: it measures 1.4 × 0.8 metres, and in
the course of the last nearly 3 years, it has become particularly cluttered. Finally bit
the bullet and decided that today was the day to tidy it up:
I've been using the same labels for my jars for nearly 30 years now, and this is different.
The bottom two lines are also written in the handwriting of my first wife. I wonder if the
contents are that old, or whether we've refilled them.
Also found a lot of other stuff; the oldest with use-by date expired on 16 August 1999. And a lot of dried herbs of the kind we have growing in the garden.
Ended up with a big basket of stuff which Yvonne took over to
Chris, much of which ended up in the chicken coop or the rubbish bin.
Having tidied up, I needed photos, of course. How do you take a photo in a cramped area
like that? A panorama, of course.
That wasn't even as easy as I thought. To get the maximum vertical coverage, it makes sense
to set the camera vertically and get the width by taking more individual photos. But that
doesn't work with my rails: the camera is no longer above the axis of the panorama head, in
fact about 20 cm away. In this situation, that would give impossible parallax errors. So I
ended up taking a horizontal panorama, which only got a fraction of the height:
It's clear that I need more hardware, something like a bracket that will tilt the camera
independently of the focusing rail. The lighting also greatly needs improving, so I can see
myself doing a number of experiments. But these photos brought out another detail I hadn't
noticed before: my “ring flash” is in fact a left/right illumination only, as
the reflections on the glass jars clearly show:
Is that a problem? Both kinds are available on the market, and they're for different
things. In this case, if it had been round, I probably wouldn't have had so many shadows.
Maybe I should experiment with other ring flash attachments—but how do you know which
ones have round illumination?
Mail from Peter Ludikovsky today, pointing out that the content on http://www.vinumvm.org/ looks strange. I deregistered the domain a year or two
ago after it became clear that there was no future for Vinum.
It was on the cards that a domain squatter would take it over, and that's what happened.
But no porn or other objectionable content; instead the squatter
(Kevin Dunham, 463 Heather Road, Penticton BC V2A 6N8, Canada; I wonder if he's related to Jerry)
seems to be subtly modifying the pages.
Yvonne is still not very happy about the Klondike Solitaire
games available with Ubuntu, and wants the old
version that Keith Packard wrote decades ago. How difficult can it be? I just need to
compile it, after all. But this system comes without sources, even without header files.
Which ones should I install? In the end, I installed them all—and they ended up in
individual trees in my home directory! What a pain. How do I bend the paths to work at
all? I suppose it's at least partially due to the fact that I have to learn all this stuff
all over again. But why install sources in somebody's home directory? I'm taken back to
the situation when I wrote Porting UNIX Software.
Still no proper reply to my mail to Tempo
Australia, but took another look on the web site. Finally they have the
manual for my GPS navigator online, but only very recently. As firefox half
told me:
So I need more hardware for taking panoramas—maybe. Spent some time investigating
what's on the market, and came up with a set of requirements:
Rotate the camera about the nodal point.
Level the camera independently of the tripod.
Switch from horizontal to vertical orientation, preferably without remounting the
camera.
Rotate the camera in specific increments for equally-spaced images.
Leveling is worth thinking about in more detail. My problem was that the tripod was
frequently not on level ground, so the central column wasn't vertical, and I was looking for
a leveller that would swivel relative to the tripod head below it. But that's not
necessary: if you level the tripod head and then leave it static and rotate the bracket
relative to the head, you've achieved the same thing. Probably for this reason, none of the
hardware I found has a separate swivel function.
I already have the ability to rotate the camera about the nodal point (criterion 1), and
criterion 4, the equally-spaced rotation is more “nice to have” than “must
have”, but the other two are important. I've mentioned the trouble levelling the
camera, and the immediate cause of my search is the vertical orientation, which I just can't
get at the moment. There are devices that seem fulfil all of the requirements. But the
prices blew me away! And they don't really seem to relate to the utility of the device.
It's clear that I've left the domain of commodity equipment.
The simplest ones only fulfil criterion 2. They only rotate on the level. They're barely
better than a pan(orama) and tilt head: the only real advantage is that they have a degree
scale. The cheapest I found was the Lenspen adaptor plate at US $17.
Novoflex seems
to be a name that you'll find everywhere in the panorama game, and it's one you have to pay
for. If $17 is too little to spend, you can spend up to $725 to get something very similar
to the Lenspen, the Universal Pro panorama plate. Compared to the LensPen it's “Pro”, of
course, and undoubtedly better quality. It offers click stops at preset intervals
(criterion 4, 8 steps between 10° and 60°), while with the LensPen you have to read the
angles manually. For another $130 you can get a quick release plate that maintains the
nodal point distances and helps with changeover from horizontal to vertical, though I
wouldn't call it automatic. That's a total of $855, for which you can buy quite a good DSLR
body, but it does fulfil all criteria.
Of course, if you have money to burn, you don't need to stop there. The Novoflex VR-PRO kit costs US $1,245 and fulfils all requirements, though I'm still
not sure how good the change from horizontal to vertical is.
The Vista Panorama bracket seems to
fulfil criterion 3, the ability to switch from horizontal to vertical, but it appears to be
only manual. With a pan head it would also fulfil criterion 1. It doesn't seem to address
the other requirements, which makes its price (US $149, or $119 on eBay) rather high.
The first adaptor I saw was on eBay, and
irritatingly doesn't have a non-eBay page that I can find. The price blew me away: about US
$288. It wasn't until I looked more carefully that I found that this is in fact not a bad
price by comparison. It fulfils all the requirements above, including—it
seems—gimbal-based automatic switching from horizontal to vertical, and 10 swivel
increments between 5° and 60°. About the only thing it doesn't have is a second rotation
plate for the camera when mounted vertically; but you can buy a LensPen for $17.
Then there's another similar product on the market, also eBay. It's difficult to say what it
does, because the description is so poor, and the photos show the thing only folded
together. But it doesn't seem to have the gimbal mounting, and it costs US $366, $78 or so
more than the other, so I think it would lose out.
Then there are things like the GigaPan Epic
100, which not only gives you a mount, but also a computer to process the panorama for
you. I can't see much use in that, but maybe others do. In this price range I'd prefer to
pick the best component for each function, rather than take a grab bag.
A little weeding in the garden. I should do a lot. The rightmost section of the compost
heap has probably reached capacity, and I need to consolidate the leftmost and middle
sections:
While thinking about the panorama hardware, decided to do some experimental 2-D panoramas in
the pantry: 3 stripes, one above the other, each of 7 photos. The results were useful in
that they showed that I hadn't thought things through:
Decided to do it in two steps: first seven vertical panoramas of three photos each, then
stitch them together. But I didn't get that far: the first vertical panorama failed to
stitch. It's not surprising. I tried to add manual control points, but at first I thought
I had chosen the wrong images. I couldn't work out any communality:
The problem is the extreme parallax: you can't just jack up the tripod and take a photo from
the other side. In these images, it looks as if the underside of the shelf at the top is
the same in both cases; in fact, the shelf at the top of the lower image is the shelf in the
middle of the top image. The illusion is so complete that Hugin is completely confused by the other two
images. The control points are all around the shelving and the support:
In summary: don't do that, then. The simple way is just to tilt the camera, but that won't
create the effect I want. Maybe there is no way to do so.
While in an experimentative mood, took some photos of my office with the lights off,
illuminated only by the LEDs in various equipment. This, too, was not a success, not even
the optimized version (on the right):
I've been dragging my feet on the sourdough starter. It's been 3 days since I started the
last step, but maybe that's not such a bad thing. How long does
the Lactobacillus need to adapt
itself to the new material? Once the culture has established itself, it should be a matter
of hours, but maybe it needs longer at the beginning. Anyway, it still looked happy enough
today, and it was still bubbling mildly:
Also ate some “King Oyster” mushrooms that we had bought
in Springvale the other day.
They're nice and fleshy, but I wonder if we could have done better than to put them in
stir-fry noodles.
Another message from eBay today, reminding me to
update my credit card details. This time in HTML, but with just as vague information on how
to do it. Since I pay by PayPal and not credit
card, that's a strange thing. Tried the “Contact Us”, but of course, Real Web
Presence doesn't use email or telephones (in fact, I think they do, or did, but they hide it
well), and all I got was a FAQ page where I could enter my “question”,
“Why are you sending me credit card expiration reminders, when I pay via
PayPal?”. To judge by the input field it was too long for the attention span of the
typical webmaster—only about 65% fitted. And what reply did I get?
Yvonne went shopping today, so I set up the GPS navigator for
her. She came back with the information that it had hung itself up, and she had to power it
down. That's happened before. I must keep an eye on it.
Now that the ideas of building a new house are on ice, it's time to get back to things we
had planned to do here. Not too early: there's stuff lying around here that I've been
putting off for nearly 2 years. Today put in a dustbin under the kitchen sink, which I
found enough to keep me going for the day.
Earlier this week I tried out the GPS navigator going
through Melbourne, and I was quite positively impressed. Today we set off
for Heywood: I have finally
given in to Yvonne's desire to get a new dog, so we set off
to Macklin Shepherds to look at a
German Shepherd Dog. And of
course we used the navigator to get there.
How good was the route it calculated? It certainly found the little roads that I used to
get between Adelaide
and Dereel three years ago, and which I
use whenever I go in that direction. But after that? How can you tell? Like Google Maps, the itinerary doesn't give you any
town names. Unlike Google maps, the overview map is far too low-resolution to recognize
anything. So my VicRoads country street directory isn't so redundant after all: to find the route, I
had to step through the itinerary and compare with the map. Very tedious, and at one point
I got the wrong itinerary because I had somehow managed to not delete a “via”
(an intermediate point) that I had wanted to put in
at Port Fairy, but then
decided against. If I hadn't compared the route, I might not have noticed until we were
close to Port Fairy. The route it calculated proved to be quite reasonable, though,
corresponding pretty much to the one that Google Maps calculated.
On the way, it occurred to us that we should find fuel, preferably at a co-branded
Woolworths/Caltex station, where we get 4¢ per litre discount.
Isn't it nice to have that kind of information available at your fingertips? We also
considered going to an ALDI branch, since they
were sold out of some of the specials that Yvonne wanted to buy yesterday. And of course,
since the navigator came from ALDI, the locations of all their branches were stored in it.
There are others as well, in the category “Shopping”, but there's a separate
category “Aldi Poi” [sic]. It proved there was both a petrol station and
ALDI very close together on our way
in Hamilton, so I entered them
as on-the-way stops. It wasn't until much later that I discovered that that completely
changed our itinerary, adding 16 km.
About 20 km before Hamilton, the navigator froze up. The display remained, but it didn't do
anything, and there was no way to turn it off:
The blue LED at bottom left is the charge LED, and it is only illuminated when the USB cable
(connector on the left, currently without a cable) is connected. Blue means “fully
charged”, and red means “charging”. Under the circumstances when the
photo was taken (a couple of hours later), it shouldn't be illuminated at all, since the USB
cable isn't connected.
Continued to Hamilton on manual and found ALDI and Woolworths, and even got our specials at
ALDI. After some examination of the unit, decided that the little hole above the charge LED
might be a reset hole, so back into ALDI and asked for a paper clip (something I should
carry with me). But no, it didn't help. It did get the attention of the store manager, who
I think is called Kate. She wasn't able to help much, but it wasn't for lack of good will.
She did tell me that they've sold “lots” of them, and they haven't had much
trouble.
So we left the thing to discharge the batteries, which according to the manual are supposed
to last about 3 hours. It didn't quite make that, but they lasted over 2½. Finally it lost
power on the way home shortly
before Woolsthorpe. And yes, it came
back little the worse for wear, though some settings seem to have changed. But it didn't
take us through Woolsthorpe: it took us up Harris Road and then told us to turn right into a
paddock with no sign of a road anywhere. Went on, and it calculated another route along Bromfields Road, which must count as one of the roughest roads
anything has recommended to me:
Fortunately, it was the only unpaved road. But, strangely, the route it calculated back to
Dereel was nothing like the one it calculated on the way there, nor like the one that Google
Maps calculated (which, not surprisingly, is same as in the other direction). It's also not
the same as the one it calculated from Heywood, which didn't make the mistake about going
cross-country. As I was able to confirm later, it happened because the route started
shortly before Woolsthorpe. Interestingly, the route from Woolsthorpe to Mortlake was
actually better than the normal routes.
More minor errors: in Mortlake
it told us to take the wrong exit from a roundabout, and the “time remaining”
logic seems a little dubious, as I had already noted on Tuesday. From about Mortlake it was
clear to me that we would get home at about 14:50, but it continued predicting 15:15 until
shortly before we got home, where it gradually oscillated towards the correct arrival time,
including giving an arrival time of the current minute when we were still over 4 km away.
We finally got there at about 14:51.
So: what do I do now? Clearly either the unit or the software is defective. The extent of
the problem suggests that it's the unit. I could take it back to ALDI and get a refund, no
questions asked. And then buy something else. But that would mean making all the same
discoveries all over again, and if I don't like it as much, tough luck. I can also have it
replaced under warranty—maybe. Clearly there's a lot of room for improvement, and I'm
sure it'll happen over the next few years. But in particular the lack of overview of routes
really irritates me, and it's such an obvious requirement that it's possible that other
manufacturers have done a better job. But it's certainly worth its money, though it's a
pity we still need paper maps. More thinking needed.
Our visit in Heywood was to look at the mother of a puppy that Yvonne had booked in advance:
their puppies are booked out before they're born. But Linda Maclean had just had a refusal,
so we did have the opportunity to pick up a dog (no choice, however). Took a look round,
and as I suspected, Yvonne decided she liked the dog, kennel
name Macklin Preto, who was born on 1 April 2010:
He's a friendly enough thing, and has already taken to following Yvonne around. The cats
behaved predictably, but even by the evening things were already looking better, and
Piccola was showing some interest in him.
Three years ago today we were searching for houses in
the Dereel area, and we first saw the
house we're living in. It wasn't exactly love at first sight, but a whole lot better than
the others we had seen, and within 1½ months we had moved in. I took some photos of the
house at the time, and every year since then I've tried to recreate the garden photos from
the same point of view. Did so again today; clearly things have changed a lot:
When we first saw the house, the daisy bushes had a couple of solitary flowers on them.
Since then, they've bloomed relatively well, presumably because of the irrigation we
installed, though they're getting a bit old now. It's also interesting to note how much the
Callistemons have grown, despite relatively heavy pruning. On the left, the tall
yellow-blooming trees don't seem to have been there at all when we first saw the house. The
one on the right definitely self-seeded. They're almost certainly Senna
aciphylla, and that link suggests that they're weeds.
Here, of course, the big difference is the verandah. If I had taken the original photo from
a little further away, the difference would be even more impressive. The smaller of the two
birch trees (now almost without leaves) has now grown into the left of the photo.
When we first saw the house, we had little idea of how we would use the area. The remainder
of the photos are a good indication of that: they're of areas we hardly use.
