I have completed weeding most of the area in the south-west of the eastern bed that I've
been working on, and already new grass shoots are popping up between the mulch. So I
sprayed the with glyphosate the other
day, and now I'll have to wait for it to die off. In the meantime worked on other areas of
the garden, including to the south of the verandah, where the grass is intermingled with
various bulbs which are popping up. What a pain! Spent nearly an hour doing a couple of
square metres. If I ever get on top of this, I must really ensure that it doesn't take off
again.
After months of inactivity, the Friends of the
Ballarat Botanical Gardens have come back to life. Today I got a number of mail
messages from the mail address that I'm trying to close down: it's in the TransACT domain ncable.net.au, so not only
does it not reflect our domain, but it also ties us to TransACT. The more I look at that,
the less sense it makes.
One of the messages looked very dubious:
From fbg@ncable.net.au Tue May 1 11:28:06 2012
Received: from 203.208.114.27
(SquirrelMail authenticated user fbg@ncable.net.au)
by webmail.ncable.net.au with HTTP;
Tue, 1 May 2012 11:28:06 +1000 (EST)
Message-ID: <49325.203.208.114.27.1335835686.squirrel@webmail.ncable.net.au>
Date: Tue, 1 May 2012 11:28:06 +1000 (EST)
Subject: Email update request
From: "Friends of Ballarat Botanic Gardens" <fbg@ncable.net.au>
To: groggyhimself@fbbg.org.au
User-Agent: SquirrelMail/1.4.9a
---------------------------- Original Message ----------------------------
Subject: e-mail verifications
From: Admin@ncable.net.au
Date: Sat, April 28, 2012 9:41 pm
--------------------------------------------------------------------------
Webnews From ncable.net.au Customer Services This Message
from ncable.net.au We want to alert you that we are deleting
all old email account to create more space for the new
account,To prevent your email account from being closed,
Please click on the link below or copy and paste the URL into your browser:
Doesn't that look familiar, the old phishing trick? But look at those headers. Those are
the complete headers, and they show the message originating from our own (shared) mail
address. Is one of us spamming our own mail address? How do I catch them?
Spent several hours turning this over in my head, and then I found another one (this one
somewhat shortened):
Received: from 203.208.114.27
(SquirrelMail authenticated user fbg@ncable.net.au)
by webmail.ncable.net.au with HTTP;
Tue, 1 May 2012 11:30:32 +1000 (EST)
Message-ID: <49345.203.208.114.27.1335835832.squirrel@webmail.ncable.net.au>
Date: Tue, 1 May 2012 11:30:32 +1000 (EST)
Subject: Gmail information
From: "Friends of Ballarat Botanic Gardens" <fbg@ncable.net.au>
---------------------------- Original Message ----------------------------
Subject: Your Gmail address, ballaratbg@gmail.com, has been created
From: "Gmail Team" <mail-noreply@google.com>
Date: Tue, April 24, 2012 3:50 pm
--------------------------------------------------------------------------
Congratulations on creating your brand new Gmail address,
ballaratbg@gmail.com.
This one might look dubious, but it's valid, and it really comes from Google. It relates to
the email address I set up last week. But the headers show that it comes
from ncable.net.au!
On further investigation, it seems that this is a forwarded message. Genevieve was working
on the computer today, and it seems that she cleaned up the mailbox, forwarded the messages
to those people she thought most appropriate, and deleted the originals—thus hiding the
origin of the messages. My best bet is that the first message is malware (dubious subject,
and why should TransACT want me to go to a Google site for things like that?), while the
second is clearly legitimate.
But that's an impossible state of affairs! How can you just strip off the headers when you
forward something? Google mail doesn't do that. Maybe SquirrelMail doesn't have to either, but I don't see any
knobs to tune it—including the all-important auto-forward function. I think it's really
time to get rid of the mail address altogether. An auto-respond function would be good
there, but it doesn't offer that either.
It's been nearly 2 years since I bought my Sanyo PLV-Z700 TV
projector. On the whole it has worked well, but today the “replace lamp” LED lit up.
There's nothing obviously wrong with the lamp. The lamp lifetime is given as 2000 hours in
full power mode and 3000 hours in economy mode—I think. There's no mention of any specs in
the 220 page instruction manual (20 pages each for 11 languages), and the web site is particularly light on specs. On investigation I discovered that it has just passed the
2000 hour limit, 2006 in fact. And for some reason it was running at full power, though I
had set it to run in “economy mode”. So it looks to me very much as if the indicator is set
to come on after a specified number of hours rather than based on the real condition of the
bulb. After resetting the bulb counter, the indicator went off again.
On the other hand, is it a good idea to continue until the lamp grows dim? There's a real
danger of explosion, and though I don't think I'm anywhere close to that, is it worth the
risk? The good news is that projector bulbs have come down significantly in price since
last time I had to buy one. Then I paid $545. Now I can
get a replacement for as little as $118, though that one doesn't have a housing. Presumably
it's intended to fit in the old lamp housing, but they don't say how to do that. The
cheapest complete lamp (from the same eBay seller) is only $139, so I think it makes sense
to buy that one.
Pizza again for dinner today. We've been
working forever on getting the base to rise properly, and today we made significant
progress. In the electric pizza ovens, bake the base alone for 5 minutes from both sides,
then put the topping on and bake for another 12 minutes, again from both sides. That's
actually faster than the way we have been doing it so far. Today the bases rose so much
during the first bake that one of them fouled the upper heating element. Next time round
I'll reduce the amount of dough by a third.
Yesterday's experience made it clear that we
should migrate email for the Friends of the Ballarat
Botanical Gardens from TransACT to
gmail as soon as possible. The first step, of
course, is to forward the messages from TransACT until we can wean people from sending them
there in the first place—a problem not made any simpler by the fact that messages continue
to be sent from that address without a Reply-To: header.
But how? I couldn't find anything. So I tried calling TransACT technical support at 13 30
61. “Optus regrets that the number you have
dialled has been disconnected”. Repeatedly. So I finally found the normal, non-toll-free
number in Canberra, 02 6229 8000 (thank God for VoIP!). Called them, explained to the
receptionist that there was a problem with getting through on the toll-free number. “Never
mind, I can connect you”. No indication that she thought that that was a problem for
TransACT. “It's probably just a problem with the phone company”. No recognition that it
might be a problem for TransACT. She promised to report it and put me in a wait loop that
took forever. After 20 minutes somebody answered and told me that they had clearly made a
mistake, the technician should have been on site yesterday. Clearly not my case, and he put
me back in the wait loop. Another 10 minutes and I was connected to Tahlia, who was
reluctant to spell her name, and who told me I had come in on the wrong number, and that I
should have called the Victorian number. I was about to complain that it was the only
number I could get, but she did connect me, to Ruby, after a total of about 45 minutes.
I first asked Ruby how to set up SquirrelMail to not strip the headers off a message when forwarding. She didn't understand. “Where did
you get SquirrelMail from?”. She thought that it was something installed on my machine. It
seems that they don't call it by its name, just “Webmail”. When I finally explained it to
her, she didn't know what a header was. “The standard SMTP headers”. “Where is this on
your screen?”. “You understand SMTP, right?”. “Yes, I've heard of it”. So I asked to be
connected to somebody who understood the question, and she told me she would get somebody to
call me back.
Finally Sean called back and first wanted me to give him access to the mail account before
he could tell me anything. He appeared to be asking for passwords over the phone. How do I
know who he is? Not very encouraging from a technical support manager. But he finally
confirmed that no, there's no way that you can configure SquirrelMail to keep the headers
when forwarding messages. Indeed, he has been a technical support manager for some years
now, and he doesn't know of any program that keeps the headers when forwarding. And I
didn't know any other that strips them.
In any case, he was more help with my second question. I could configure SquirrelMail to
automatically forward to another address. I asked him to step me though the necessary mouse
clicks, and once again he wanted to get into my account to do so. In the end, it proved
that I couldn't do it from my interface, but that he could, something that Raoul Dixon later
confirmed. And yes, it worked. And didn't store the message in the local inbox any more.
That's a little faster a transition than I had planned, but I suppose it will work.
Later I checked: yes, it seems that most Microsoft space MUAs strip headers when forwarding
messages. I had never noticed before.
Got round to ordering a new lamp for my Sanyo PLV-Z700, choosing
the one with the housing for $139. Also did some investigation of the projector itself: it
seems that it doesn't save the lamp intensity setting in its configuration, and every time I
turn it on again, it goes back to full intensity. Ugh.
A visit today from Pauline Olszanka, a freelance reporter for The Telegraph, wanting to know how to find my doctor
friends whom I mentioned last week. It
seems that Ben Leach has written an article damning them, to be published in this weekend's
“Sunday Telegraph”, and she wanted to hear the other viewpoint, something my friend doesn't
want to do—incorrectly, in my opinion. There's not much I can do without permission, but
it's interesting to see how much interest the matter has generated.
It's interesting to note that Ben has also called her in the middle of the night. In
addition, an editor of the “Sunday Telegraph” also did so. I had assumed that Ben's action
was due to stupidity, but this suggests
that Hanlon's razor does not apply
here: it now looks like downright arrogance and lack of consideration.
Into the office this morning to discover my mouse limping. It moved relatively smoothly
over the root window, but it kept “sticking” when moving over other windows. I don't
understand the details of the mouse processing in X in any detail, but at the very least it needs to send messages to the window, as a quick
play with xev indicates:
MotionNotify event, serial 27, synthetic NO, window 0x6a00004,
root 0x501, subw 0x0, time 2584229172, (128,68), root:(133,89),
state 0x0, is_hint 0, same_screen YES
Looking at my X server, it had been running for a while, and had hit nearly a gigabyte of
memory:
USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
root 54016 0.0 4.9 994904 153136 v0 S 24Apr12 86:15.86 /usr/local/bin/X :0 -auth /home/grog/.serverauth.539
So it seemed a good idea to restart it and see what happened. What happened was that I had
no mouse at all. Going back to the vty where I started it, there was some spurious message
from hald, which unfortunately didn't get saved—something about invalid context, I
think. So tried restarting moused, but that didn't help. And restarting hald
just didn't work if I started it by name: it just stopped with return code 0 and no
messages.
Spent a bit of time messing around in the dark (without X) and finally found that I had
been here before, though not under these
circumstances. To restart everything, it should be sufficient to enter:
More playing around with the Friends of the Ballarat
Botanical Gardens email today. Further investigation of the issue of headers in
forwarded mail messages show that I was just plain wrong, and Sean is right: just about no
MUA, not even mutt, preserves headers in forwarded messages. I was really thinking
of bouncing, not forwarding, and that's something that Microsoft-space MUAs don't seem to
understand.
In the process, discovered things about gmail that were less than pleasant: apart from an inability to configure many things (like date
formats, for example), I can't find any way to edit outgoing messages. Yes, it opens up a
form on the web browser, but unlike most forms, I can't redirect this one to a real editor.
Normally I use “It's All Text” to redirect the form to an emacsclient, but for some reason
gmail turns off this function—probably to implement its own mini-editor. That makes it
quite painful, especially since the editor doesn't understand Emacs key bindings.
The weather is getting cooler. Earlier this week we had an overnight low of 2.3°, and we've
had temperatures (barely) below 0° in May in previous years. Time to move some things to
the greenhouse, notably
the Hibiscus rosa-sinensis
that suffered so badly 2 years ago and also the Mandevilla, which is still in full bloom. I
have my concerns about the hibiscus, which is particularly prone to attack by white flies,
of which the greenhouse is full, but the alternative is worse.
Also moved a number of plants from the shade area, notably the rose cuttings I made
a month ago. Despite the time of year,
many of them have taken well. At the time I put half of them in the shade and half in the
greenhouse. Both have struck, but clearly the ones in the greenhouse are doing better
(first photo):
The Salvia (I think the cultivar is
“limelight”) is flowering, though not as colourfully as I've seen on the web (second
image). Hopefully we have the correct plant.
The Mirabilis jalapa that I
planted from seed two weeks ago have now all
germinated, and even the one attacked by fungus has recovered. The yellow ones were ready
for transplantation, so did that and put three of them in the greenhouse, where the
temperatures are clearly lower than in the house. The other three can stay in the bathroom;
I'll see how they go.
What haven't (yet) germinated are
the Kniphofia seeds that I planted at
the same time. Went out searching on Google (“germination” is your friend) and found some
stuff on the GardenWeb “growing
from seed” forum, including the suggestion that they might like to be stored at a low
temperature for a month before they want to germinate. Took some of them and put them in a
plastic bag with some moist paper, then put the bag in the fridge. They'll come out at the
beginning of next month. The greenhouse is obviously the place for the ones I have already
planted.
Signing up for GardenWeb was interesting: I already appear to have a login, but I've lost
the details, so I signed up again. More clever password security ideas: you can use any
special character you like, as long
as it's black:
I took the photos of the rose cuttings with the Zuiko Digital ED 50mm F2.0
Macro lens, the Mecablitz 58 AF-1 O
digital flash set to TTL measurement and the ring flash attachment. Once again I had
exposure issues, and the images were far too dark. I don't understand why: the TTL
measurement should compensate for the light falloff due to the ring flash. But what I got,
without any further processing, looks like this:
That was taken at f/11 and 24°/200 ISO. To get good exposure at that distance (about 30 cm)
I would need a guide number of about 3.3. In fact, the unit has a guide number of 58, and
the ring flash couldn't possibly have reduced the output by that much. Further experiments
suggest that the only criterion for the exposure is the aperture. At 30°/800 ISO and f/8 I
got a histogram that was 1 EV brighter, but still underexposed:
If it had really been a limitation of the flash unit, it could have been up to 3 EV
brighter. So it can't really be that. What about flash exposure compensation? I moved
back to 24°/200 ISO and set 2 EV more flash exposure. The first image, taken at f/11,
looked pretty much the same as the very first image taken without flash compensation, and I
got an image 1 EV brighter than the original image at f/8 and 2 EV brighter at f/5.6:
The whole thing suggests that the exposure is just not working. ISO speed makes no
difference, exposure compensation makes no difference. It's almost worth thinking about
using manual flash, but that requires lots of calculation.
Into the office this morning to find us off the Net again, since nearly 9 hours.
The ppp process was running, signal strength
was normal, no messages in /var/log/ppp.log. But a ping gave me ping:
sendto: No buffer space available. Restarted PPP, and things worked again.
Somehow there are too many things that can cause link interruptions. When I started keeping records, it was to monitor ADSL line quality, and I had
good metrics from the modem to help me. Now the problems can be this horrible flaky Huawei
1762 USB toy, the link itself, or the Optus network behind it. Once it gets to Internode, I have
not (yet) had any further problems.
But hopefully I won't have to put up with this much longer. The three weeks' period for
lodging an appeal with
the VCAT
against the radiation tower expired last week, and so far I have not heard of any such appeal. If that's the case, and
Peter of Daly International is
right, we could have the National Broadband
Network service up within 2 months. Here's hoping.
Yesterday I had to admit I was wrong:
most MUAs discard most headers
when forwarding email, even mutt. Well,
maybe.
The real issue is what is meant by “forwarding”. There are three different approaches:
Simply forward a message the way
an MTA would do, putting
in a Resent-From: header to show what has happened. In this case, of course,
the headers are intact.
Create a message containing the quoted text of the old message. This is only marginally
different from a reply, and though there's nothing to stop
the MUA from including the
headers, typically they don't do so, presumably because there's normally no way to
suppress the display of the uninteresting ones.
Create a new message and add the old one as
a MIME attachment. In this case, the
original message is completely unchanged, with all headers exactly as when received.
This is what SpamAssassin does,
for example. Callum Gibson showed me today that you can get mutt to do it this
way too, by setting the variable mime_forward to yes.
It's been three months since I started
looking for a new keyboard. Surprisingly, my current one, now pushing 23 years old, has
recovered somewhat and now no longer bounces as much as it did. In the meantime I've been
looking for a cheap Sun Type 7 keyboard, which looks relatively similar and has a USB
connector:
In particular, there are 10 keys in 2 columns to the left of the main keyboard. Yes, I
know, they have special functions
under Solaris, but in
the end they only generate scan codes, so I can modify a key map to get them to
generate F1 to F10. Similarly—I most sincerely hope!—I can remap things to
generate Ctrl and Meta in the correct places. The only concern there is that
some broken keyboards have a hardware CapsLock key that can't be remapped, but I
haven't heard of that from Sun.
The problem I had three months ago was that the only keyboards I could find were in the USA,
and the postage would have been about double the price of the keyboard. So I put in a
search on eBay Australia, and finally one
showed up, postage only half the price of the keyboard.
But what do I need to connect it to my computer? I looked at a lot of web pages back in
February, but I forgot to write them down. I won't make that mistake again. After a bit of
searching, discovered that at least the Type 6 has its own keyboard description in X, and it
seems that the Type 6 and Type 7 have the same scan codes, so there's very little to do.
Still, I found a number of interesting pages, coincidentally several of them from FreeBSD users, though the information is not
FreeBSD-specific. There's some interesting general detail on this page, though the most
important is probably the link to this description of setting up XKB. One detail on that page is wrong, though: it
refers to {XROOT}/lib/X11/xkb. Nowadays ${XROOT} expands
to /usr/local, so this would be /usr/local/lib/X11/xkb, but in fact
it's /usr/local/share/X11/xkb. Still, useful stuff. And it indicates that the
keyboard should work, so I bought it.
Did some investigation of the white fly I have in the greenhouse, and discovered that it's
so common in greenhouses that it's frequently called “Greenhouse White Fly”.
The Wikipedia page is currently pretty
rough, but this page from the South Australian
Research and Development Institute looks helpful. The species is
presumably Trialeurodes
vaporariorum, and they can be controlled in multiple ways: with sprays (surprisingly,
nobody mentioned the pyrethrum that I
have been using, though it seems effective),
with Encarsia formosa, a small
parasitic wasp that is also very sensitive to the insecticides, and with traps, which can be
as simple as fly strips. As they say, spraying is difficult because they hide in their
thousands on the underside of the leaves, so I think I'll try this last approach first.
House photo day today, and for once things went relatively smoothly. I've more or less
decided not to use
multiple-exposure HDR
any more for these photos; even the slightest movement makes things look fuzzy. But how
should I expose things? There's a difference of up to 3 EV between the illumination of the
individual components of some panoramas, and I've already seen cases where the brightened
shadows look washed-out as a result. Today I tried exposures ⅔ EV and 1 EV apart for a
couple of panoramas (first one in each pair is exposed less, with
mouseover alternation):
Which is better? Difficult to say. The contrast was relatively mild today, but what
differentiation there was in the sky is gone. On the other hand, the detail of the garden
looks better. But that's just normal lighting, and it has nothing to do with HDR.
While I was at it, went back to some of the images I took 3½
years ago. They looked terrible! I knew that at the time—I had been using ufraw without the appropriate profiles, and
I had later improved things a little—but it really hit me in the face today. Ran the raw
images through DxO
Optics "Pro" and compared them. It's like night and day:
It's not just the gradation: somehow the old ones look unsharp. The difference in size is
due to ufraw not trimming the edges of the image the way other software does. But it
looks like I have a lot of reprocessing to do.
Another attempt at pommes soufflées today. Cut the potatoes in 3 mm slices as planned, and fried them the first time in fat at
150°.
Did the second frying in the “small” friteuse, the one we normally use for things which tend
to dirty the fat. We use the fat that has already served a term for frying clean things in
the “big” friteuse, and by the time we throw it out it tends to smoke noticeably. I had put
that down to the age of the fat, but today I measured the temperature. I had already noticed that the thermostat of this
friteuse is generous in its heating, but I wasn't expecting the 215° that I measured when I
set it to 190°. No wonder it smoked!
In any case, after cooling down to about 195°, I was marginally more successful than last
time. But some of the potatoes started to inflate in the first bath, and it was still only
a minority of slices that blew up in the second. So it seems that the first bath, at 150°,
is still too warm. I'll try 135° next time. And the fact that they didn't stay inflated
suggests that the 3 mm is too thin. So next time I'll try with 4 mm.
None of this takes the kind of potato into account, of course. Maybe it's just not possible
to get good results with the varieties we get here in Australia.
Once upon a time a computer was something expensive. Now we're throwing out computers that
could still run rings round a CDC
7600, the supercomputer of my youth. Talking with Chris Yeardley after dinner
and discovered that I had at least 20 computers, not counting motherboards, in and around my
office, most of them functional and a number belong to Chris. Chris suggested that we
should write an article “101 uses for a dead computer”, so I brought out a handful of
laptops and we started playing around:
Yvonne bought some fly strips (or “Fly glue traps” according
to the packaging) in town today. They're supposed to be good against
the white fly
infestation I have in the greenhouse, and it looks as if they are. Within minutes I had
hundreds on the strip, and in less than a day I had what appears to be most of them:
At any rate, when I shake the leaves, no more flies fly out. There are still some there,
but they could be the dead ones. Now to wait for the remaining eggs to hatch and join their
ancestors.
I've made a couple of comments about Ben
Leach of the Sunday Telegraph and a doctor
friend of mine over the last couple of weeks, and out of respect for what has been going on,
I didn't mention names. But now he has published his article, so it's clear that I was referring to Rapinder Adekola, whom I met
last spring. I contacted her a couple of
times and at least got her to talk to Ben, for what difference that makes. She hasn't told
me enough for me to be able to verify her side of the story independently, but it really
does read strangely. About the only thing of interest is that they state that James used
Rapinder's email address for the matter. That could sound suspicious, but I confirmed last
year that yes, indeed, he did that to me too, for no obvious reason. But I can't see that
it's sinister.
Hopefully this article won't make things too bad for her.
Recently I have been inundated with spam with subject lines like Employment you've been
searching! and New job vacancy - see details. Much of it came from people I
know, notably in the FreeBSD project, but even
more came from me myself. The messages clearly come from combinations of user and ISP that
can be broken in to, and about the only thing they have in common is a line matching
the regexpplease reply to
.*@employmenteu.com,with, notably with a missing space after the comma.
So, is somebody trying to discredit employmenteu.com? Looking at the whois
data, it seems not:
Registrant Contact:
Jordan R. Harrison
Jordan Harrison supp@email.com
410-854-0150 fax: 410-854-3421
2102 Marie Street
Odenton MD 21113
That's a lot of information for a spammer, in particular that it was registered just as this
wave of spam started, though the address is invalid. And they only have an MX
record. No A record, no web site. And their allegedly authoritative name servers
(ns[12].jobcenter-coordination.com., a name significant in itself) don't want to know
about it. But there's enough information there for
a DDoS attack on them. That would be
amusing.
I baked another loaf of my sourdough bread yesterday. The recipe has pretty much settled now: flour, sourdough, water, salt, caraway
seed. But in the days when I was baking yeast bread, I had things like bread mixes and
“bread improver”, and there's still a lot of the latter left over. So yesterday I put in 30
g of it to see what difference it would make.
It didn't make much difference. It certainly didn't improve things. But the sourdough
aroma was missing, or at least greatly reduced. Did the “improver” bind it? In any case,
something not to be repeated.
I still can't get over how much my photo processing has improved since 2008, and today I
spent much of the day reprocessing. I started taking raw images on a regular basis from
about August 2008, and today I went through a number of collections dating from then.
One particular day surprised me, though. It didn't look
like the others, but it also didn't look “right”. Here is one image between the images from
the week before and the week after:
How did that happen? I set to looking at what was wrong and discovered I had already
reprocessed the directory. The creation timestamps of the images pointed to 18 January 2012, and the diary entry explained the rest.
Unfortunately, I didn't write down the settings at the time, and I didn't save the
information, something that I must investigate. But I suspect I used the “postcard”
settings, which are clearly too gaudy. I've found the also-gaudy “HDR Artistic” settings to
be more appropriate, so today I started reprocessing the images again with those settings,
giving myself no less than three different ways to look at the images. It'll be a while
before I have them finished.
All this photo reprocessing brings a danger, of course: I could rename or remove an image to
which I have referred on a web page. I already have a 404 document that sends me email if a
page on my site refers to a non-existent page, and that has greatly improved things. But
you don't get a 404 for a missing image. On the other hand, nearly all my images are
generated by a PHP function, so it's (relatively)
easy to check whether the image exists or not. The difficulty is mapping the URL to the
local path name. I got that done, and I was still having it claim that the images didn't
exist.
Debugging PHP is relatively clunky, and in general I put print statements in the
code. Gradually I was reduced to using explicit path names:
if ($localimagepath != "/home/grog/public_html/Photos/20081031/tiny/cj-2.jpeg")
print <<< EOS
<pre>
'$localimagepath'
'/home/grog/public_html/Photos/20081031/tiny/cj-2.jpeg'
</pre>
...
That makes sense: the image name gets sanitized to HTML entities where the characters could
be a problem. The character - by itself isn't a problem, but -- can be
interpreted as the end of a comment. All I needed to do was to take the original file name
before sanitization, and it worked, and I output code like:
And how it worked! Error messages started pouring in, for errors that most people, myself
included, had never noticed. So I removed the display and just sent the mail messages.
I'll reinstate when things become sane again.
Somehow didn't get much work done in the garden today. The cool weather doesn't help, but
it wasn't cool enough to stop weed growth, so spent a bit of time weeding. I'm still not
convinced I can stay ahead of the weeds.
Fixing up the photos for 6 September 2008 seemed to be a straightforward
enough thing when I started yesterday, but somehow it kept me going all day. First I
reprocessed the photos with the “HDR Artistic” settings, and discovered that there was
(almost) no difference. Clearly this was the way I had done it before. It took me
some time to realize that I had taken two complete sets of images on that day, one in the
morning and one in the afternoon, and had only processed the afternoon photos. So the
difference wasn't due to the processing: it was due to the time of day. Here the comparison
with the images from the morning, which clearly look better for this particular viewpoint:
Powercor isn't my favourite company
because of the pain they cause me, but there are other aspects which cause problems too.
Yesterday somebody came by to look at the vegetation on the property to see if it was
endangering the overhead power lines. He decided that yes, it did, and that he would be
back soon to trim the trees involved.
I'm not happy with that. I've had bad experiences with ETSA in the past, where the tree trimmers removed trees
that never could have endangered the power lines. So it was with mixed feelings that I read
a sheet of paper posted in our letter box today:
Isn't there something wrong here? Verbally he says he'll do it himself, and then I get a
letter telling me to do it. He doesn't say which trees, but Yvonne had noted a couple of trees that had been marked:
Why? Does X mean remove and T trim? Neither is appropriate. Those trees
don't come within 2 metres of the power line, and they haven't changed in the 5 years since
I first saw the property. Tried calling up Vemco, couldn't get in touch with the engineer
(apparently Travis Hewish). Hopefully we're not in for more annoyances.
Bad news from Amy Boyd of the Golden
Plains Shire Council: somebody (not yet known who, but we can guess) has put in an
objection to the erection of the NBN tower. That means it
goes to
the VCAT
to waste our time and their money. Now it's unlikely that it will be operational before
Christmas.
A flock of Estrildid finches,
which I know better as Prachtfinken,
have taken to bathing in the mini-pond that we have now placed in the Japanese Garden. We
didn't notice it until this morning, when all we saw was a lot of water spraying around
behind some of the bushes. Yvonne though that the sprinklers
were running at the wrong time.
But that's the problem: the bushes. We couldn't see the birds because of them. The main
one proved to be a volunteer yellow Marguerite daisy bush, and then there were lots of
weeds, including some large-leafed clover-like plants and some kind of vine, which was
gradually strangling the other plants in the area:
Spent a lot of time removing what I could of that; the whole area needs an overhaul, and in
the end we decided that, instead of transplanting the daisy, we would give it to Chris.
Gradually we have enough plants in the garden, including a surprising number of bulbs, which
I should really rearrange after they have finished flowering in spring.
Also removed a couple of the petunias in that area. It's the end of the season, and they
haven't flowered well. I think it's a little dark for them there, so I've planted them in
some of the hanging baskets currently spending the winter in the greenhouse.
dereel, my main machine, crashed (or rather, hung) this afternoon. Nothing in the
log files, which is normal enough. That's why I log remotely to another machine, in this
case cojones, the machine that is connecting me to the Internet until the NBN
radiation tower is finally complete. But
something went wrong there: syslogd hung itself up a month ago and I didn't notice,
so there's no evidence of what caused the hang.
On the bright side, this happens so seldom that it's worth mentioning
here. w3.lemis.com has been up for 1318 days (well over 3½ years), and it's still
going strong.
In particular, the Return key is completely different, and that was one of the
aspects I looked at before ordering it. I'll have to see whether I can come to terms with
it or not. In addition, it has British key caps (£ over the 3 key, for
example), and there's at least one extra key to the left of the Z key,
marked \ and |, but in the default map it produces <
and >.
The correct way to handle these keyboards is in the X configuration, but that would mean restarting X, something I'm reluctant to do. So I
just plugged it in to see what it would do.
It worked! It seems that just about all the keys that correspond to a normal modern PC
keyboard also generate the same codes. The ones on the left, of course, generate ones that
the default X configuration doesn't know. But that's what xmodmap is for. So I
played around with that for a while, not helped by the fact that it doesn't know whether to
output in hex or decimal:
It wasn't that difficult to remap everything, and now I have a keymap file that will do the mapping. I've mapped the 10 keys to the left of the
main keyboard to F1 to F10. On my old keyboard I have the keys F11
and F12 where the Help key is, but no Esc key to the right, so I've
mapped Help to F11 and Esc to F12. Ugly, but it works.
Somehow, though, it marks the end of an era. Commercial computers have been around
since UNIVAC I in 1950, about 62 years. I
have used the same keyboards, from Northgate and Avant, for 22 years, over a third of that
time.
And how do I like the keyboard? It's much quieter than the Northgate, and it doesn't
feel quite the same. In the course of time I'll get used to the different places for some
keys. And I have a whole lot of keys that I could use for strange remappings.
Chris Yeardley and I have had some documents which needed to be witnessed by a Justice of
the Peace. Normally we get Loes Pearson to do that, but we haven't been able to
contact her for over a month. I finally called up the Department of Justice to ask if
something had happened to her, and was connected with Jerry (a woman), who was only too
happy to find another JP for me—not what I had asked. It seemed that she had neither any
idea why nor concern for the possibility that a JP had gone missing. I asked to speak to
her superior, and got a call back a little later from Jan Szuba, who wasn't really able to
help much. They don't appear to track JPs, though he did note the fact. What happens when
one dies? There must be some way to keep the lists up to date.
You can find a JP online, of course, assuming you have the right browser and enough patience.
Some clever Javascript programmer has written a form that offers no features not available
in plain HTML, and implemented it so that the text disappears when you enter it, at least on
my firefox. And when you
finally get a result, you have to select it by letters of the alphabet, after which it
paginates in a far-too-small window and truncates text to make it illegible, as in the top
line here:
Still, Yvonne had told me that there was one on duty at the
Police station in Ballarat, so off there
and got my documents witnessed by a JP whose name I forget, and who doesn't appear to be in
the list, but who was quite informative. He complained about the list on the web site too,
and told me that Loes is on holiday and has sent her apologies. So why didn't the people at
the Department of Justice know that? “They don't know anything”.
It was nice sunny autumn weather today, so while in town, I dropped in at the Botanical Gardens and
took a few panoramas. They weren't the best: in sunshine it's almost impossible to avoid
shadows of the tripod in a 360° panorama, and in one case I missed out a shot, so I ended up
with:
Finally got a call back from Travis Hewish of the Vemco Group about the trimming work. It seems that the letter he left wasn't as
contradictory as it appeared, just misleading. It referred to the supply lines from the
transformer to the house (“low-voltage service line”). That should be clear enough, except
that some power suppliers in Australia use the term “low voltage” to refer to the 10 kV
supply lines.
This also explains the issue with the clearance. 60 cm is the minimum for the 230 V house
supply line, but it's round 5 m for the 10 kV lines, and the regulations have changed
recently. But I asked him to let me know before anybody went on to the property, and he
refused. “We're allowed on the property whether or not you want us there”. Clearly
somebody who has received new powers and is intent on wielding them. Looks like we'll have
fun there.
Continued with my reprocessing of old photos today, looking at the ones taken on
23 May 2009. I should have read the diary entry before doing so. I
had found duplicate photo series of the garden, and processed them all, but it seems that
the first series was taken with a graded grey filter mounted the wrong way round, which is
why I didn't process them. But I did today, and the results were much better. In
particular, I was trying to get more detail in the sky without losing shadow detail. With
DxO Optics "Pro" I got results that more than compensated for the graded filter. Here images with the
graded filter the wrong way round (left) and the right way round (right), first as processed
3 years ago, and then as processed today:
I'm planning to mount my projector higher, above the window in the lounge room. How do I do
that? I could hang a column from the ceiling, but it's quite high there, and the mount I
have uses four screws in a square about 25 cm apart, which doesn't match Australian ceiling
constructions. So I thought of putting a shelf there instead, but it proves to be
surprisingly difficult to find something appropriate. It took a long while for me to think
of looking for a wall mount bracket. And yes, they're available, and not significantly more
expensive than the shelf would be.
But what load limit? Some support 10 kg, others 15. What does my projector weigh? Sanyo's
“documentation” doesn't say, and neither does their web site. Finally found some specs elsewhere, even
containing a link to Sanyo's now-lost spec sheet. Answer:
it weighs 7.5 kg. So the smaller (and cheaper) ones should do the trick.
Finally I've put my photo stuff on the back burner and got round to doing some other
things—a little. More weeding south of the verandah. It's amazing how long it takes to
clear even a small area.
My new cage rat trap didn't do too well last time a rat entered: it was on a shelf, the rat knocked it over onto the ground and got out. So
this time round I put it in the barbecue with the lid down, where it wasn't so easy to knock
it over. But that's all right. It didn't need to:
I still have issues getting good shadow detail in my images. In the image above I went back
to the HDR techniques I've been using earlier, but in general that doesn't seem to be the
best choice. One of the main issues is that when taking panoramas, the canonical
instructions are to give each component image the same exposure. That gives rise to images
like this from the north view sequence (in this case without any further processing):
That's quite impressive, but looking at the cypresses to the right of the centre, it's clear
that the image has run out of steam. There's just not enough shadow detail in the original:
So why this rule that you should expose all components equally? Clearly it has to do with
exposure blending, and my experiments in the past with varying the exposure were less than
successful. But I've installed a newer version of enblend now, so decided to see if it could
handle them any better.
The results were mixed. Here the same panorama again, this time the one processed with DxO
(above) and the one created with automatic exposure. Running the
mouse over either image (preferably after enlarging) shows the alternate image:
On the other hand, the sky around the cypresses loses detail. That's clearly because it's
underexposed.
And the others? On the whole, the results were better with automatic exposure, though I had
problems with the one from the Japanese garden, coincidentally the first one I tried.
Another one that caused problems was the “garden SE” image, with both bright sunshine and
the shade area. Here the results looked much better, but not perfect:
The trouble here was the extreme contrast in the single image on the left. Enblend
rendered the plants in the shade area well, but the sky above was completely burnt out. I
had taken both series at the same time, so using the hugin mask functionality, I was
able to mix and match from the two series:
Here the image component outlined in red is from the manual exposure sequence, and the one
in cyan is automatically exposed. The results don't look at all bad:
Out to do some more weeding this afternoon, but it was really too cold. At least I have the
old “succulent” bed (the one south of the verandah) weeded now.
Had chicken tanduri for dinner this evening,
with Chris Yeardley as usual on a Saturday. You're supposed to eat
tanduri naan with that, but I've never been
successful making that, and I'm gradually getting my act together with chapatis, so that's what I made.
How many? Yvonne eats very little, and I only made one for
her, but I eat 3 or 4. Chris has a normal appetite, so I made a total of 8, a slow
business—it took me half an hour just to bake the things. And we had 2 left over. I can't
recall now whether I ate 4 and Chris only one, or whether it was 3 and 2. In any case, I
think I should base my calculations on about 2½ per person, Yvonne excepted.
Didn't do much today. It was quite cool, so spent some time playing around with yesterday's
panoramas and improving the selection
of EXIF data that I display on the web
pages.
Yvonne bought some unspecified species
of Scilla (cultivar “Blue Bells”)
and Tritelia (cultivar “Spring Star”) a
couple of weeks ago, and I've been puzzling where to plant them. But time's getting on
(though on the package both claim that they can be planted up to early winter), so finally
got round to planting where we had planned.
But that didn't work well. In the recently overhauled north part of the east garden,
discovered that the horse manure we had spread has its own grass seeds in it, so we're going
to have to let them germinate and then spray them. That's easier if there are no bulbs to
avoid. So I stopped there after planting a few Tritelias and spread the rest round the
north bed, the south-east corner of the east bed, and to the south of the verandah, where I
have also tidied up recently. Removed a large number of
giant Echium seedlings in the process.
They're pretty, but each of them grows to such a tree that I wouldn't know what to do with
them:
Lots of bulbs are popping up now, and it looks as if some
(Narcissus?) will flower in the next
couple of weeks, so gave them another helping of fertilizer. Also transplanted the
red Mirabilis jalapa seedlings,
which are now looking a bit better. It's amazing how much more slowly they came than the
yellow ones.
Yvonne has been experimenting with “chicken parcels”, chicken and accompaniments wrapped
up in puff pastry and baked. They don't taste bad, but somehow there's still something
missing. Today she used Parmesan cheese and bacon, but it didn't make a big enough
difference. Maybe we should approach with something like
a Chicken Cordon Bleu wrapped
in puff pastry next time.
Autumn is gradually drawing to a close, and today I took my flower photos. There's nothing special to
mention, but we seem to have flowers from all seasons at the moment. Here summer, autumn,
winter and spring:
It's been nearly a month since we got a new
remote control electric collar for Nemo. This one
worked—once. The contacts to the skin look dubious, but the real problem seemed to be
battery consumption. Since there were no instructions, and there appeared to be no way to
turn the receiver off, I removed the battery after every use. Despite that, the battery was
flat within an hour or two of use. So we bought a new one—$7.50, a significant proportion
of the $28 we paid for the collar. And it was drained in a similar time. So we ordered a
pack of 8 from China (also about $7.50 for the lot), and they arrived today.
Off down the road to measure the range of the thing. According to the packaging it's 200
metres. According to the seller, it's 800. Put one of the batteries in just before
testing, after measuring the voltage: 6.46 V. Yvonne stood
on the road and I drove off, stopping every 100 metres and holding the device out for clear
line of sight. At 100 metres she had to press several times before anything happened. At
150 metres there was no response. On the way back I got a response on about the third
attempt at 120 metres.
Back home, measured the voltage of the battery again. 6.40 V. Clearly the battery
consumption is as useless as the rest of the device. Sending it back is out of the
question: the
(registered) postage would cost $26. Sent the seller a message suggesting that this
warranted negative feedback, and got a response offering me another one, a waterproof
version of the one we gave up on last December.
Should we try it? I suppose it won't cost anything, and maybe it will work. But it seems
that these things are all just a waste of time.
Some months ago, in an interview, Margaret Throsby mentioned the idea of a high-quality
BBC documentary. At the time I was reminded
of Mohandas Gandhi's quoted reply to the
question about Western civilization: “I think it would be a good idea”. But today I saw one
that epitomizes all that is horrible in BBC documentaries.
At the end of the Second World War, millions of people were displaced.
Most left by foot with whatever goods they could carry. This was immediately after the war,
long before the escapes from the “so-called” DDR
(“German Democratic
Republic”). On checking, I discover it was about 12 million civilians, and just about
anybody who has lived in Germany knows some. My own father-in-law was one of them. His
sister saw how one of her cousins was eaten by a wolf.
So when I saw that the BBC had created a documentary entitled “The Long March To Freedom”, describing
exactly such an event, I was interested. Never mind the fact that the people who were
driven from their homes considered it exile, not freedom: it didn't bother to mention the
real problems. This film is about the (“only”) 300,000 allied prisoners of war who had to
do the same thing, without the encumbrance of their worldly goods.
OK, nothing wrong with that, though you'd expect some mention of the fact that there were so
many more civilians being displaced. But how they present it! Why does BBC so often throw
the (in this case particularly unappetizing) presenter into your face? Or tell the story of
individuals?
The rest of the (first episode) appears to show a specific story of the individuals
interviewed in the programme. Potentially there's something of interest there, and it's
certainly a topic that interests me. But the form of the presentation and the lack of
overview makes it completely uninteresting. And this is a “high class” documentary?
For years I have had a certain aversion to fish and seafood, somewhat to my own
disappointment and definitely to that of Yvonne, who would
gladly eat it most of the time. I suspect it's because of the very limited varieties
available in Australia (amazing when you think that most of the population lives on the
coast). Last week Yvonne bought some squid tubes, and we fried half of them in batter.
That didn't taste bad. Today I recalled a recipe I had eaten when I lived in Malaysia for
prawns and squid, and found a variant in
Passmore and Reid's “The Complete
Chinese Cookbook”. Surprise, surprise! It's basically just a kind of stir-fry, but
it tastes very good.
Last monthYvonne received an enforcement order notice for a speeding fine. This was the culmination of a
series of errors by the police, and in each case we were required to submit applications in
writing to get it revoked. Today we got a reply to the latest one:
Not sufficient grounds? The police have stuffed up twice, causing the extra costs. This
looks suspiciously like the third one. What grounds would they deem sufficient? It's
enough to shake any remaining confidence in the judicial system. But somehow, through no
fault of our own, we're getting deeper and deeper into a mess of the police's making. And
what happened to the principle of In
dubio pro re?
Spent a bit of time trying to improve the panoramas I messed up last week, using GIMP. The results are less than stellar. Running the
mouse over either image (preferably after enlarging) shows the alternate image:
Put it in the projector, and the results were less than illuminating. First indications
show that it has 10% less light output than the old, worn-out one, and the lighting is
uneven. Sent a message to the seller, and they claim that this is “normal”. Looks like
another fight, but first I need to investigate more carefully.
While taking things apart, also took a look at my old blood pressure monitor. I'm pretty
sure the inaccuracy is due to the too-fast release of pressure, and I suspected that it
might be possible to adjust it. Found the release valve under the circuit board (on the
right in the second image, connected by the blue cables):
For some months now I've had a dull pain round the thumb joint in my right hand. One doctor
told me that it was arthritis, and
definitely not gout, but didn't prescribe
any treatment. Another told me that it was pretty certainly gout, and that despite normal
blood urate levels, I should increase my
dose of Allopurinol. Clearly not
very trustworthy diagnoses.
So today, finally, off to see Heather at MCPhysio. As some variant on the demonstration effect, I had no pain in it, and she
had to prod a bit to get any sensation of pain. And she agreed with the first doctor:
probably arthritis. But she gave me a massage and a bandage and some hints on how to live
with it, and the good news that if it gets too painful, there are good replacements. She
should know: she has the same complaint herself. Apparently it's common among
physiotherapists.
While in the area, dropped in at the Ballarat Botanical Gardens to visit Genevieve Lowe at
the Friends. It seems they have a new
computer for the accounts, and there's no backup. Coincidentally Liz Gilfillan, the
president, walked in, and I got authorization from her to buy a 1 TB USB disk for backups.
Down to Officeworks, where the prices
for external disks blew me away—up to $279 for a 2 TB unit, and the cheapest was $127.
That's a big difference from last June,
when the cheapest 2 TB unit was $98. Is that still the effect of the Thai floods last year?
Even the 1 TB drives seemed to be over $100, but finally I found one with almost identical
specs to the other, from Seagate, for $78.
While at the Friends, Genevieve asked me how
to incorporate the FBBG banner in outgoing emails.
Aaargh! HTML mail with gratuitous images! The horror!
That's been my standpoint since HTML mail first came out, but I'm beginning to wonder how
tenable it is. HTML is still an issue for email, and certainly I'd hate to see us send out
messages only in HTML, but in some cases it makes sense. Is this one of them?
Maybe. The typical Friend is probably used to it, and getting a message only in monospace
text would probably appear unprofessional. I can see my stance softening, anyway.
I've known and used the term “In dubio pro re” (“in case of doubt, for the matter [at
hand]”) for decades, most recently yesterday. But when I looked it up in Wikipedia
yesterday, I couldn't find it. Instead I
found in dubio pro reo,
interestingly with a strong German flavour. But why reo? Clearly this is ablative,
and res is in the 5th
declension, so the ablative would be re. Even if there were a differing
ablative form, it would be more likely to be rea, since res is feminine.
Looking for in dubio pro
re on Google didn't help much, since Google seems to have lost the ability to perform
an exact search: it returned mainly hits for “in dubio pro reo”. After some time I came to
a different translation: the indirect object is not (ablative of) res (thing, matter)
but of reus or rea, an adjectival noun in the second and first declensions meaning “defendant”. So it does make sense. But is my usage just
incorrect memory, or from a different tradition? Certainly there are many others who use it
too, and also others confused by it, as indicated at the bottom of this exchange.
As far as I'm concerned, there's doubt, and in case of doubt I'll go for re.
It's annoying that my new projector lamp is even dimmer than the old one it was supposed to
replace, but it's puzzling that it also has uneven colour. And the seller doesn't want to
know about that, of course:
For the color discrepancy,i asked our egineer,it would be problem of color panel,please kindly inspect the machine.
Because the bulb only have one of function,it is lighting or no lighting.
So I replaced the old lamp and tried again. Colour even across the screen. Put in the new
lamp. Colour uneven again. How can that happen? At any rate got some images with the
relative readings of the old and new lamp:
When I lived in Germany, employment rules were very rigid. You joined a company at the
beginning of a month, you left at the end of a month (frequently the end of December of the
year you turned 65), and periods of notice were extreme, up to 12 months on either
side. There were exceptions, of course, and I seem to have been exceptional in many
ways. For example, 36 years ago today I joined Karstadt AG. And when I left, by some strange coincidence, not to mention a lot of
negotiations, I was able to reduce my period of notice from 6 months (to the end of a
quarter) to about 4, so it was on 17 May 1982, 30 years ago today,
that I joined Tandem Computers. How time flies!
Time may fly, but I don't. As if that wasn't enough, today is also the anniversary of the
last time I travelled by aeroplane, six years ago. I
don't miss that.
To the General Meeting of the Friends of the
Ballarat Botanical Gardens this morning, the first I've been to. It seems that it's
not annual, nor strictly bi-annual: there's one in May and one in August.
There's nothing much to report from the meeting, except that more people are gradually
coming round to the idea that we need a real web site. After the meeting we had a
presentation by John Ditchburn talking about
how he grows vegetables in his small (1100m²) urban garden. He made it sound so simple, but
it seems that I'm not the only one who has had trouble growing vegetables: Mike Sorrell told
a heartfelt story about growing capsicum (that's right, only one fruit) that reminds me of
my own efforts recently.
The real reason for my attendance was to complete some of the jobs I had with the Friends' computers: a “new” CRT monitor for one with a
damaged LCD monitor, setting up an account for non-privileged users on their new laptop, and
installing the new backup disk. The latter proved more difficult than I thought. Yes, it's
trivial to install a USB disk, but the box said it contained backup software. So it did,
too, an “extended trial”, after which you have to buy it. Somehow that's not my style, so I
gave up. I'll have to investigate what we can use instead.
While I was there, had an experience which explained a bit why people use mail services like
Gmail: no setup required. Lorraine Powell
wanted to give Raoul Dixon a file. Simple, right? Send it as an attachment to an email
message. Well, no, that box isn't on the net for some reason (the most obvious being a
missing net cable), so first she had to copy it to a USB stick and plug it into the new
laptop. And then? Set up Microsoft “Outlook”? No way. Use gmail and just attach it.
Yes, it's clunky, but anything that requires climbing through directory trees on the back of
a mouse is clunky, and at least this doesn't need any setup.
For a long time I've had continual problems with a number of different USB devices on a
number of different computers. About the only thing in common—most of the time, anyway—is
the operating system: FreeBSD. Is there some
problem there? Or are most USB devices just flaky? It's not as if I haven't had USB
problems with other operating systems, after all, and the fact that it happens less often
might just be due to the fact that I don't use them as often.
Today I had a number of disk errors while backing up my photos. It could be the disk, of
course, but there was something funny about them. I use rsync to do the backup, and it reported I/O errors
(EIO, or 5 as reported). But it continued. Looking at the system log, I saw lots
of:
May 18 11:51:45 lagoon kernel: (da0:umass-sim0:0:0:0): AutoSense failed
May 18 11:51:45 lagoon kernel: g_vfs_done():da0p1[WRITE(offset=443382513664, length=131072)]error = 5
May 18 11:51:45 lagoon kernel: g_vfs_done():da0p1[WRITE(offset=443382644736, length=131072)]error = 5
May 18 11:51:45 lagoon kernel: g_vfs_done():da0p1[READ(offset=1596956100608, length=2048)]error = 5
What does “AutoSense failed” mean? It sounds like a USB error, not a disk error. And the
write and read errors look strange for a number of reasons. I found another case in the log
files from the day before with many more errors, but the offsets didn't match today's ones.
In the older case, 125 errors were logged in a single second, spread over offsets
between 65536 and 1950230118400, just about the whole 2 TB disk. That's pushing the
capabilities of the disk.
The writes were frequently for 131072 bytes, MAXPHYS on a FreeBSD system. But
the SCSI subsystem writes a maximum of 65536 bytes.
As I found later by comparing the files, the file system data all made it to disk, and a
subsequent fsck found no problems.
The problems don't occur just when writing. I ran dd across the disk afterwards
and had errors at random places across the disk:
dd: /dev/da0: Input/output error
58221+0 records in
58221+0 records out
7631142912 bytes transferred in 399.033699 secs (19124056 bytes/sec)
dd: /dev/da0: Input/output error
dd: /dev/da0: Input/output error
146807+0 records in
146807+0 records out
19242287104 bytes transferred in 903.271487 secs (21302883 bytes/sec)
On these occasions, the log files had only an AutoSense failed message.
The offsets spanning the whole disk suggest that the file system is accessing the
superblocks. Offset 65536 is a typical offset for the main superblock. And clearly the
system did try to write there. But the other offsets don't seem to match, so this could
just be a coincidence. But disk errors don't normally span the whole disk. There's
something else going on here, and I suspect it is related to the first message: “AutoSense
failed”. Is that maybe a retryable error that isn't being retried? Or am I maybe making
holes elsewhere in the file system?
The last couple of days have kept me out of the garden, but today I got back a little: the
verandah has been getting increasingly messy, and the dropping vine leaves meant that I
really had to tidy things up. In the process, started moving some of the plants
elsewhere—it's gradually becoming too cluttered.
Ćevapčići for dinner tonight, for which
Yvonne wants a Weißkrautsalat, literally “white cabbage salad”, but close to the
English coleslaw. And my recipe calls
for apple, which we didn't have.
Went out on the web looking for alternatives, and found more that I could be bothered
looking at. None of them wanted apple. One went to great lengths to exclude olive oil,
others wanted olive oil and no other, and few resembled each other. In the end, tried one
with raw carrot and balsamic vinegar instead of the lemon juice I normally use. Also, for
the fun of it, tried garlic.
The results? It didn't taste very different. Certainly the carrot was barely noticeable,
as was the balsamic vinegar. The garlic came back and hit me later on, clearly far too
much. So I'll probably go back to the old recipe, possibly with carrot or nothing instead
of the apple. Clearly this recipe is particularly flexible.
Last week I did experiments with multiple alternate exposures from the same position and
aligned together, but stitching only one version. That allowed me to make several panoramas
with different appearance, and was particularly useful in the case of the “garden SE”
panorama that includes the shade area and the sky. Today I looked at two other panoramas.
One of the problems with the verandah panorama is the lighting in the back corner, so today
I put a remote-controlled studio flash in the room next to it and let it fire on that area:
On the whole, that didn't look particularly good. It's supposed to brighten up the area,
not throw a shadow of the window frame onto it. And the alternate image is clearly too dark
at the back. In addition, the flash didn't fire on the next image to the right, because the
on-camera flash was pointing in the wrong direction. But I tried stitching the results
anyway. I had three choices: the image with flash, both of the images above, or the image
without flash. The results of merging the images are interesting. The first attempt did
quite a good job of averaging the two photos:
Between those two images, I didn't do anything except move the centre of the image.
I need to read more about how exposure averaging works.
The other panorama was the one taken from the north-west corner of the house. This time I
took a complete series with and without flash. Here with flash, merged, and without flash:
This one was probably taken on the wrong day, since the shadows are not too bad even without
flash. But it's interesting how in this case the merged image looks almost identical to the
one taken only with flash.
The other thing I wanted to do was to get the angles of view of the panorama into
the EXIF data. To my surprise, hugin already does this, putting the
information into the User Comment section:
User Comment : .Projection: Miller Cylindrical (11).FOV: 360 x 180.Ev: 9.90
The only problem was that it doesn't copy most of the EXIF data, so I had been reinstating
it from an original. Fixed that by copying the appropriate fields from the
original TIFF output, and then spent some
time getting my PHP scripts to interpret that and display it in the JavaScript popup that
occurs when the mouse goes over the image, only to discover that the values are Just Plain
Wrong in most cases. For the verandah panorama it gave a horizontal field of view of 360°
(correct) and a vertical field of view of 180° (wrong; that would imply a complete sphere,
which I don't have. And it gives a field of 260° horizontal by 104° vertical for the image
above; in fact, the horizontal field of view might just approach 180°, and the vertical
field of view of the lens is only 87°, from which a certain quantity has to be deducted
because of the way the images overlap. I'd guess it wouldn't be more that 65° in that
image.
Looking at the way hugin does it, it's clear that the calculations need to be
modified. It seems to give the overall field of view of the Fast Panorama preview window,
it doesn't take cropping into account, and the calculations look wrong anyway:
Yvonne wanted to upload a video to YouTube today, but it was too long, so she asked me to
cut it into manageable pieces for her. Not a problem: I've been there before with avidemux2. So I fired it up, processed, saved it and... couldn't find the
result. On further examination, I found:
=== grog@defake (/dev/pts/0) /Photos/yvonne/20120425 13 -> avidemux2_gtk Maureen-on-Morena-1 *************************
Avidemux v2.5.6
*************************
(hundreds of lines of debug output omitted)
So the cwd
was /Photos/yvonne/20120425. But there was nothing there. Tried again and looked at
the save page and found:
This horrible program has completely ignored the current directory and saved it to the same
place it did last time. Why?
The answer is simple, of course: modern GUI environments have done away with many of the
conveniences of UNIX, including the concept of a working directory. avidemux2
has a working directory, but ignores it. Why do people put up with this? Is it the fault
of avidemux2 or of GTK+?
These were the same images as I had been processing all the time, and they lined up well.
And this was just after reading in the .pto project file that had already lined them
up correctly. What was wrong? After a bit of checking, discovered that during
my EXIF copying experiments, I had
accidentally copied the EXIF data from a panorama to one of its components. But why should
that have caused the problem? Possibly field of view issues. But I didn't think of that at
the time and just restored from a backup, so I will never know.
And so are the “Lilli Marleen” roses that I had investigated two weeks ago. Strangely, the ones from the shade are looking
better, and the ones that have been in the greenhouse all the time are looking worse. Here
two weeks ago and today, first the ones that were in the greenhouse all the time, then the
ones from the shade:
Time was getting on, so today I finally got round to assembling it. What a pain! I don't
think I've ever seen such a complicated assembly, and I certainly wasn't expecting it in
something as simple as a box:
I needed three different kinds of spanners and a heavy-duty screwdriver (nothing supplied),
and it took me the best part of an hour to assemble the lid. Then I got started on the
frame, which is held together by screws in aluminium profiles. But they don't fit!
According to the instructions, the bottom profiles should be attached to the verticals at
the same height:
The ends of the profiles foul each other, so in these photos the cutout for the screw on the
left can't come closer than about 1 cm from the channel where the screw belongs. It would
be possible to fix that by cutting off about 1 cm of the end of the inside profile, or by
mounting one above the other, but there's a simpler solution: it goes back. I wonder how
they assembled the box in the example. Did they maybe fix it on that one box and then not
bother to fix the rest? I can see them being overloaded with returns.
Cooking rice is not difficult, and I've known how to
do so since I was a child. You wash the rice, add water, boil until the rice absorbs all
the water, cover and wait for 20 minutes. But not everybody agrees. Even today there are
people who say you should boil the rice in lots of water, like potatoes, and pour away the
excess water and nutritional value at the end. More important, though, are cultural
differences. I learnt the method above in Malaysia, but as I discovered decades ago, many cultures put fat and salt into the rice as well. I do
this now with Indian food, where the fat is
usually ghee, but that tastes surprisingly
bad with Chinese food.
But how much water do you use? In the past I have gone by eye and sometimes used too much
(making the rice bloated and mushy) or too little (making it hard and possibly even
uncooked). Today, for the first time ever, I tried weighing things. For 423 g normal long
grain rice I added 504 g of water. After cooking, the result weighed 891 g, meaning that 36
g of water had evaporated during cooking.
But that's not important: the rice was too hard, probably because I cooked so much (normally
I only do 200 g) and didn't adequately compensate for the additional depth of the rice.
Next time I'll try 550 g.
None of this applies to Basmati rice.
There I typically use less water, but that's another experiment.
Our cat Lilac has not been well lately. She's 15
years old, not nearly as agile as she once was, and appears to have become incontinent. But
yesterday I saw her having difficulty urinating, so there's also the possibility that it was
just a case of Cystitis. Today
Yvonne took her to the vet for examination. Given her age, I
wasn't sure that I'd see her again. Interestingly, as soon as she saw the cat carrier, she
headed straight for it and didn't want to come out again:
That's unlike Piccola, who was coming with her for
a vaccination, and who resisted strongly. The vet found yes, bladder full, unable to
urinate, so much so that she was not even able to extract urine for a test. So we decided
to treat her for Cystitis and see if that helps. The alternative would be to put her down.
Antibiotics don't work immediately, of course, and she was quite unhappy in the afternoon.
Found her sitting in the garden, something I've never seen before, accompanied by Piccola:
We bought our first grapefruit tree
nearly 4 years ago, and it did so badly
that they gave us another tree for free 8 months
later. They both came with a fruit already on the tree, but apart from those, we only
ever harvested 2 others—a total of 4 fruit from 2 trees in over 3 years. But that seems to
have changed this year. The trees aren't much bigger, but they're bearing fruit, one of
them despite the fact that it is being attacked by far more
vigorous Hebes:
Also did some more weeding. I think I'm going to have to keep up an hour a day until I can
start all over again, and then probably do an hour a day for ever more.
Until about last Christmas I
took HDR sequences for
my weekly garden photos; after that, I only do it in exceptional cases, using DxO Optics "Pro" to
compensate for the limited dynamic range. Apart from other issues, this has reduced the
number of photos I take by over 50%. The effect on my battery life is interesting. Between
31 July 2011 and 24 December 2011 (146 days), I
took 10,309 photos and recharged the batteries 7 times. That's an average of 20.86 days or
1,473 photos per battery charge, and 70.6 photos per day. Between 24 December 2011 and today (151 days) I took a total of 5,257 photos and recharged the
batteries 7 times, an average of 21.57 days or 751 photos per battery, and “only” 34.8
photos per day. So it seems that the duration is more important than the number of photos
taken. That could be a coincidence, but it's certainly interesting how close the durations
are.
Once upon a time, life was simple. If a disk died, it was a head crash. If a domestic
light fitting failed, it was the globe. Neither seem to be the case any more. I've
already had trouble with the light dimmer in
the lounge room, but when the light failed there last night, I first replaced the globe. No
go. So today I took another look at the dimmer. Yes, one of the too-short wires from the
dimmer had come out of the terminal. Put it back, and things still didn't work. Spent some
time head-scratching, not helped by my own unclear documentation 3 months ago.
Much of the problems I had were due to the fact that the detail of the wiring didn't match
the layout of the device, and I had thought that possibly my particular dimmer had had the
switch mounted incorrectly, rotated by 90°. But looking at the installation notes again, I
saw a detail that I hadn't seen before:
I had followed the wiring, after correction, and put the second wire from the dimmer at the
terminal marked 2 in the last drawing, and that's where it came out again. But that
circuit doesn't make any sense anyway. At least in my circumstances, it makes more sense to
make the connection from the dimmer to the lamp via the Loop terminal, which is
conveniently closer to the dimmer connections.
But once again I'm flabbergasted. Why the incorrect documentation? And why did they mount
the dimmer so that the wires can only just barely reach the contacts?
I've been dragging my heels on getting the membership lists for the Friends of the Ballarat Botanical Gardens online. It's a
non-trivial amount of work, and I took the easy way out and used phpMyEdit to do the work. Put that up on the server
and found:
Fatal error: Call to undefined function preg_match() in /usr/local/www/data/pedit/phpMyEdit.class.php on line 2787
Huh? preg_match() is part of the base PHP installation, and has been for ever (well, since 4.2.0). Went and checked: my server is
deliberately not up to date (that would require rebooting and destroying the current uptime
of 1332 days), but it's not that old. On checking, discovered that it was 5.2.6, which
according to the manual requires serious effort to disable.
I've commented in the past about the tackiness of the FreeBSD ports for Apache, MySQL and PHP, and I really didn't want to look, but
somehow I had to get it working. Asked Google and found a predictably large number of hits. But the interesting thing is
that many of them are from FreeBSD, in particular this one, which referred to a
file /usr/local/lib/php/20060613/pcre.so. Checked: no file. There wasn't even that
directory, only /usr/local/lib/php/20060613-debug/ with the single
entry mysql.so. Checked on my local machine:
yes, /usr/local/lib/php/20060613/pcre.so was there, along with many others,
including mysql.so. Copied it across and tried again on the off chance. No
difference.
More experimentation, and restarted httpd to be on the safe side. Did it work? Of
course not. Further examination with find -atime showed that the file had not even
been accessed. But /usr/local/etc/php/extensions.ini had, and that looked
interesting. It contained one line for the mysql.so, so I updated it:
extension=mysql.so
extension=pcre.so
After that, things still didn't work. But /var/log/httpd-error.log was helpful:
[Tue May 22 05:36:45 2012] [notice] SIGUSR1 received. Doing graceful restart
PHP Warning: PHP Startup: pcre: Unable to initialize module
Module compiled with module API=20060613, debug=0, thread-safety=0
PHP compiled with module API=20060613, debug=1, thread-safety=0
These options need to match
That makes sense, especially given the name of the directory. Tried to build the port on
the server, but how do you set debug? The only indication was
in /var/db/ports/php5/options:
# This file is auto-generated by 'make config'.
# No user-servicable parts inside!
# Options for php5-5.2.6
_OPTIONS_READ=php5-5.2.6
WITH_CLI=true
WITH_CGI=true
WITH_APACHE=true
WITH_DEBUG=true WITH_SUHOSIN=true
WITHOUT_MULTIBYTE=true
WITH_IPV6=true
WITHOUT_MAILHEAD=true
WITHOUT_REDIRECT=true
WITHOUT_DISCARD=true
WITH_FASTCGI=true
WITH_PATHINFO=true
Still, maybe the port looked in there, so I tried to build it and—it worked! Given that it
is nearly 4 years out of date, I hadn't expected that, but fortunately I had the old tarball
on the machine. And after that, finally I got phpMyEdit to run. But what a
pain! And why was PCRE not installed in the first place? It's not my doing: I just
installed the software that I was supplied with the machine.
One of the reasons I wrote “The Complete FreeBSD” was to force myself to learn the things that I needed to document. And so, although it's
coming on 10 years since the last edition, I frequently refer to it. Today I had the task
of setting up access control for the private pages on the Friends of the Ballarat Botanical Gardens web site.
Simple: it's on page 498 of the online version. Set up a .htaccess file, create a
password with dbmmanage, and you're away.
And that's how I did it in the past. Problem is, it doesn't work any more. I got messages
like:
[Wed May 23 04:58:40 2012] [error] [client 59.167.11.50] user grog not found: /mypages/
That's not a password mismatch: the server just couldn't find the user. Tried with my own
local copy of the password file, and that worked, but that's not what I wanted. Looking at
the files, though, the format is completely different. My old files had entries that look
like this:
grog:jh45VzBDf2rlA
But the files that dbmmanage generated now were one of three different kinds of
binary format. Clearly I needed another authentication module. Went looking through the
web and found:
I shouldn't be using .htaccess in the first place. I always knew that it was
inefficient (it requires the server to climb up the tree to the document root to check
intermediate directories for .htaccess files), but this page, from
the horse's mouth, makes it clear:
Putting authentication directives in
a <Directory>
section, in your main server configuration file, is the preferred way to implement
this, and .htaccess files should be used only if you don't have access to
the main server configuration file.
I shouldn't be using dbmmanage (any more). The correct program is htpasswd.
So I should have fixed the config file, but I was too lazy. I should revisit that some
time. Instead I tried htpasswd, and it works. But it means that, now at any rate,
the information in “The Complete FreeBSD” is wrong. Was it ever right? Looking at current
documentation suggests that it wasn't, but I'm sure I tested it at the time. Should I fix
it? Put up an errata page? Probably.
More discussion of my dimmer problems on IRC today. It's amazing how different people's
viewpoints are. To summarize: I had difficulties with the dimmer because the documentation
is incorrect, and because either the wires from dimmer to the switch are too short, or it is
mounted in incorrect orientation.
But is the documentation incorrect? Callum Gibson says no. I'm referring to this image:
To me, this says: “Connect the dimmer between the terminal furthest from the dimmer and the
left-hand middle terminal. Connect the live (red) wires to terminals furthest from and
closest to the dimmer. Connect the neutral (black) wires to the terminal on the right in
the middle.” But Callum says that this is just a “block diagram”, and that there is no
conflict between the two images on the instructions:
I can't agree with this, and I suspect that Callum has the (dis)advantage of being told of
the details before having to think it through himself. To be clear: in both these
illustrations, the switch has the same orientation, with contact C on the left. In
the first, the dimmer is shown near terminal 1, in the second near
terminal Loop. My contention is that if people see these diagrams, they will assume
that the first image is an accurate illustration. That's certainly what the
instructions above that image state:
4. Connect the wires to the Light Dimmer as illustrated.
So that's an illustration, not a block diagram. In any case, there's absolutely no reason
to draw even a block diagram in a way that could easily be misinterpreted. All that would
be needed would be to put the dimmer to the right in the first diagram, or to dispense with
it altogether and show the wiring on the more realistic second image.
Then there's the question of safety. Connect the dimmer with the wires exactly in the
position shown in the first image, and nothing will happen (it won't even work). But the
wires coming out of the wall both look the same, and with correct connections that's not
important: it's symmetrical, and you could change the Power and Lamp
connections without any problem. Do that with this layout, though, and one switch position
would short-circuit live to neutral, because this is
an SPDT switch.
Callum says that this isn't important, because a licensed electrician will know what it is
really like, and he wouldn't read the documentation anyway.
So: is this a device that should really be installed by a licensed electrician? On the face
of it, yes: he is insured, so any problems would be covered, though as Callum observed, a
problem here could make for an interesting court case. But I don't buy the “licensed
electrician” thing in the first place. The only difference between this device and, say, an
power plug is that the dimmer is hard-wired. And there's no such requirement for power
plugs. Indeed, when I lived in the UK, there
were so many different power sockets (2A, 5A, 13A and 15A) that the manufacturers of
electrical goods took the cheap way out and supplied appliances without any plug at all: it
was up to the purchaser to buy and connect the correct type of plug. And other things in
the instructions are clearly not addressed to the electrician:
1. Turn power off at the switchboard.
I suppose it's typical of the instructions that they don't tell you to turn it on again when
finished.
Clearly the SPDT switch adds an element of danger to the construction. It's not needed: an
SPST switch with two non-connected terminals would be far preferable. To get it to work at
all, the power needs to be connected to one of the switch outputs, not the input. And then
when the switch is turned “off”, the dimmer is shorted out, which does no harm. I consider
this an ugly kludge; Peter Jeremy considers it elegant. But at the very least it's a
workaround for having used the wrong component. In principle an SPST switch would be
cheaper. Presumably there are so few of them in use that it proves more convenient to use
an SPDT switch instead.
The real surprise, though, is that Callum and Peter are both intelligent, level-headed
people with good technical knowledge (especially Peter). Why did we have this discussion in
the first place?
It's been over three years since we
planted the middle succulent bed, which at the time we called a rock garden. Things have
changed completely in that time:
As I guessed at the time, it's full. The only thing we planted in there after that time
were the Cannas at the back left of the
second image. And we have problems: things swallowing each other are only part of them.
There's grass everywhere, not to mention a potato that has been there longer than we have
lived here and won't say no, and there are tulips and daffodils trying to fight their way
through the succulents in the middle. What do we do? Cut back a lot, of course, but maybe
we should remove some of the plants. The tulips and daffodils in the middle are good
candidates.
No accompanying letter, just a pin indicating, amongst other things, that last week was
National Volunteer Week.
Based on other things I have received, I suppose it's for the work I've been doing for the
Friends of the Ballarat Botanical Gardens.
It's nice to be recognized, but I think that true volunteers don't do it for the
recognition, but for the public good.
Interesting message about Hugin in the mail today:
how to fill in untidy pieces of the panorama. Just what I need for the panoramas of the
Botanical Gardens I took two weeks ago, and which I tried to retouch with very limited success last week. Bruno Postle posted a link to a tutorial. The
trick is to include additional images (in this case copies of existing images) without
control points, but it left a few questions open.
After a lot of experimentation, found a way to do it. The rest of this entry will form the
basis of a tutorial.
It seems that the image gets positioned at the same place as the first image, which in this
case partially fills the hole, but in general there's no relationship between the position
of the image and where you want it to be:
To position it correctly, go back to the Images tab, select the image and enter a yaw
value, changing it until the image fits. Smaller values move left, larger values move
right:
That doesn't seem to matter, but it's cleaner to adjust the mask. Because of the way the
perspective changes, the bottom of the image is curved. It's probably better to create a
new mask with a V shape, which then fits the panorama better:
The preview image still looks pretty rough, but that's not important. enblend and enfuse will do their
magic. Crop the image and stitch it. The Assistant tab will show two disconnected
image sets, but that's not a problem:
That wasn't the way the original panorama was intended; they were supposed to be masked
differently. But here only one of the images is necessary. Remove one of the two images (I
chose image 1) and try again. Here's the result:
With a 360° panorama,
you need to move the image to the correct orientation first. The Move/Drag
functionality only moves the base panorama, not the added images.
Entering the yaw value numerically is not the most convenient method to move the
floating image. You should be able to drag it somehow. Cartola tells me that you can
do that with the Move/Drag function, but it took me a while to work out how. It
works if you can position the cursor on a part not covered by the rest of the panorama,
but you can't rely on that at the start.
On restarting Hugin and reading in the project file, I had problems with the
position of the floating image. I couldn't repeat that, however.
The current European
sovereign-debt crisis is worrying many people, ourselves very much included: it is
having a serious effect on the value of our investments. On the other hand, it's not clear
how much good excessive austerity will do, and both Greece and France are currently trying
to find another solution. Not surprisingly: in a democracy, voters are much more concerned
about their local situation than the global one. Next month's Greek election could start a
new, even worse phase of
the 2007–2012
global financial crisis.
What's the alternative? I don't know. Discussing it with Yvonne, I explained my view that the Germans don't want to live beyond their
means—very much a German trait anyway—and are thus resisting relaxation of austerity
measures. Doubtless they're reminded of the last time they tried to do that, starting in
1933, where the country grew on debt that finally was paid off by conquests, dispossession
and mass murder. Clearly nobody wants to go that way again.
But what's the alternative? There's a limit to how far you can tighten the belt. Then you
might end up cutting yourself off at the waist. As Yvonne
said, Übermensch
and Untermensch. And then we'd be
back in the same situation that the Germans want to avoid.
Horrible weather today, heavy rain and wind. Not only did I not do anything in the garden,
I didn't do anything inside either. Sometimes it's best just to watch TV.
House photo day again today, and a particularly dark and dingy one. I've been taking all
the photos at f/8, and the resultant shutter speeds went as low as 1/4 s:
I'm still thinking about whether to take all images at the same exposure or not, but at
least since enblend version 4,
it seems that I get better results from variable exposure than fixed exposure. Here a
couple of examples, first fixed, then variable, with mouseover
alternation:
Particularly in the second view, the lack of gradation the shadow of
the Buddleja is very apparent. So I
think that, for the time being at any rate, I'll stick with automatic exposure.
For some reason we've had continual damage to our letter box over the last couple of years.
On 3 May 2010, not for the first time, it was clearly
hit by something and the pole was bent, and on 31 December 2010 “horses” knocked it over. Some time later it was knocked out of the ground
altogether, and I had to put it back. By that time I didn't even bother to mention the fact
here. But today whoever it was completed his work:
Why do they do it? Once again there are tyre tracks from the neighbours, but we don't have
any reason to believe that they'd do a thing like that. The only people whom I know have
something against me are the McClellands, but I can't believe that they'd try these things
again and again. So it looks as if we'll need to find a replacement.
One of the things that Cartola suggested months ago was that I should use some kind of
browser plugin to animate my panoramas. One of the more promising looking ones was
SaladoPlayer, which I tried
some months ago and with which I ran into
documentation problems. Tried again today and got as far as being able to install and
display the demonstration panoramas,
but to run my own I had to convert the format, which involved the use of SaladoConverter, a Java application. I
have Java installed, but don't use it, and the first attempts to run it were less than
encouraging:
=== grog@dereel (/dev/pts/8) ~ 29 -> java SaladoConverter.jar Exception in thread "main" java.lang.NoClassDefFoundError: SaladoConverter/jar
Caused by: java.lang.ClassNotFoundException: SaladoConverter.jar
at java.net.URLClassLoader$1.run(URLClassLoader.java:200)
at java.security.AccessController.doPrivileged(Native Method)
at java.net.URLClassLoader.findClass(URLClassLoader.java:188)
at java.lang.ClassLoader.loadClass(ClassLoader.java:306)
at sun.misc.Launcher$AppClassLoader.loadClass(Launcher.java:276)
at java.lang.ClassLoader.loadClass(ClassLoader.java:251)
at java.lang.ClassLoader.loadClassInternal(ClassLoader.java:319)
That kind of error message reminds me
of Python. Took a
look with ktrace and found:
7010 java NAMI "/src/FreeBSD/ports/graphics/saladoplayer/SaladoConverter-0.5/SaladoConverter/jar.class"
7010 java RET stat -1 errno 2 No such file or directory
...
7010 java CALL write(0x2,0xbf9fe930,0xa)
7010 java GIO fd 2 wrote 10 bytes
"Exception "
7010 java RET write 10/0xa
7010 java CALL write(0x2,0xbf9fe930,0x11)
7010 java GIO fd 2 wrote 17 bytes
"in thread "main" "
7010 java RET write 17/0x11
7010 java CALL write(0x2,0xbf9fc8ec,0x33)
7010 java GIO fd 2 wrote 51 bytes
"java.lang.NoClassDefFoundError: SaladoConverter/jar"
What a way to say “Can't find jar.class”! I wonder what the real problem
is, but to work that out I'll have to learn a bit more about Java.
The only problem was, it didn't do anything. Only later did I discover that this is part of
Java's
inimitable way of reporting a SIGSEGV. It reported it not to the controlling
terminal (not even with a non-0 exit code), but to a file hs_err_pid83795.log with
about 120 lines of detailed information:
#
# An unexpected error has been detected by Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x822434d7, pid=83795, tid=0x8232f280
#
# Java VM: Diablo Java HotSpot(TM) Server VM (10.0-b23 mixed mode bsd-x86)
# Problematic frame:
# C [libc.so.7+0x714d7] free+0x47
#
# Please submit bug reports to freebsd-java@FreeBSD.org
#
--------------- T H R E A D ---------------
Current thread (0x8313ac00): Thread [stack: 0xbf9af000,0xbf9ff000] [id=-2110590336]
I suppose I should do that, but only after upgrading to the latest version. In the
meantime, I tried it on braindeath, my loner Microsoft box. First I had to install
Java, confusingly called release Java SE 7 update 4. And then, according to the
instructions, I had to (double) click on the “application”. What does that mean? I was
expecting an icon, but I didn't find any. So I painstakingly climbed down the tree
to z:\src\FreeBSD\ports\graphics\SaladoPlayer\SaladoConverter-0.5 and clicked
on SaladoConverter. Not quite what I expected:
So I clicked on OK and was taken to this page, which
claims:
File Type: Java Archive File
File Extension: .jar
Description: Java Archive files contain one or more files that have been
compressed and packaged into an archive file in the Zip format.
Unsupported major/minor version? Yes, SaladoConverter wants at least Java 1.6, but
my understanding is that I installed Java 1.7. But when I checked, I got:
1.4.2? That must be ancient. Removed all Java packages and tried again. The same Java
executable was still there. Searching the disk showed a couple of other file names starting
in JAVA.EXE with random appendages, some in other packages. This is a loaner
machine, so I don't want to interfere there. But at the end of the day, once again I had to
give up. At least it has brought home to me that FreeBSD isn't the only system with update issues.
Yvonne thought she could remember letterboxes on special at
ALDI recently, and I went looking through the
old weekly fliers without success. But on return from training, she brought one with her.
On more careful scrutiny, it proved that the special had been nearly 2 months ago, and
this one had been marked down to only $40.
Positioned the thing considerably further away from the Everett's driveway. In the first
image, the old box is roughly there where it used to stand. We concreted that one in, which
didn't really help: part of the damage was due to snapping off the pole just above the
concrete. So this time we took Chris Yeardley's advice and rammed
a Star dropper into the ground
and tied the pole to it. That way, if anything gets damaged, it'll be the dropper (or, more
likely, the car that runs into it).
Six weeks ago I planted some seeds
of Mirabilis jalapa, six each of
the red and yellow varieties. All of them germinated, though the yellow ones are much more
vigorous than the red ones. I have three of each kind in front of the window in the
bathroom and another three of each kind in the greenhouse. But one plant is strange:
instead of the usual leaf pairs, it has leaf triplets (second two photos):
My experience with SaladoConverter yesterday wasn't the best, but the promise of the rewards kept me
going today—all day long. Clearly the problem I had with braindeath had less to do
with Salado or Java than it did with the messed-up configuration on the box. I have another
couple of Microsofts, including one in
a VM, where I'm less concerned
about messing things up. So I installed Java on it, and how about that! It worked!
There are still a number of loose ends with the conversions. The SaladoConverter documentation states
that I need an equirectangular projection, and that's easy enough to make with Hugin. But SaladoConverter
refused some of them with the laconic comment Invalid image proportions:
pano.jpeg. After considerable investigation discovered that Hugin was stitching
invalid equirectangular images. By definition, they should have an aspect ratio of 2:1
(half as high as they are wide), but the “Assistant” was giving wider images. It seems that
the “Stitcher” does it right, once you find the “Stitch now” button hidden at bottom right.
With literally couple lines of .*xml configuration you can add navigation, maps, menus and
varius multimedia.
So I checked. The sample configuration file is not exactly a “couple lines”—205 lines in
all, not all of which are documented anywhere. Did some experimentation with the sample
file and finally came up with a couple of scripts: the first sets up the directory
structures to make it easy to do the typical Microsoft tree-climbing necessary to find
anything, and the second, run after the conversion, builds up the configuration file based
on what I have converted.
And of course it doesn't work. How do you debug XML files? Emacs nXML mode tells me
that the file is Invalid, but I can't find anything in the help to help me debug it, and in
any case the original file was also flagged Invalid. Still, it's progress. I did
manage to get some images to display, just not
with my automatically built config scripts.
Also managed to remove these silly Lens Flares that show up in most of them—it seems that
some people, including Cartola and Edwin Groothuis, like it, but after all the trouble I
have gone to to remove them from the images, I don't want some software package putting them
back in again.
The other thing that I discovered is that I need to revisit my panoramas. Really only the
verandah centre panorama makes any sense at all. I tried the view from the north for the
fun of it, but it's far too limited a view. In addition, they're not really high enough
resolution for this kind of use. The verandah centre panorama typically has a resolution of
about 9000 × 6000, but zooming in stretches the resolution:
Clearly I need to make images for the viewer at much higher resolution. And that means more
images, longer focal length, less depth of field... a real can of worms.
While I was at it with Microsoft, finally paid attention to the warnings that kept showing
up: “Your computer is in danger. Automatic updates are turned off”. I've always been
afraid of automatic updates, but maybe they have a point. So I turned them on again and got
97 updates for braindeath and 111 for smart, the VM box. And
after braindeath came back, I had no net connectivity. Given that I use rdesktop to access it, that's serious. Turned
the KVM to the monitor output and saw a message saying “Malware removed. Click here for
more information”, which disappeared before I could get the mouse there. After a while,
network connectivity came back, but rdesktop still didn't work: some component
(svchost.exe?) kept crashing shortly after logon. At least vncviewer still works, but that's a nuisance
because it's much slower and it requires a lower display size. Somehow my fears were
confirmed.
Years (I could almost say “decades”) ago, before I restarted keeping a diary, Yvonne and I first visited Jorge de Moya in Olivaylle. Sue, his
business manager at the time, made dinner and served fried (or maybe
baked) feta, which tasted very good. Yvonne
has been meaning to try it again, but our experience with baked and
fried Camembert made me hesitate.
Today we bit the bullet. Cut the cheese into cubes,
coat with breadcrumbs, and into the deep fryer at 180°:
I stopped when small globules of molten cheese started to come out, about 2 minutes, but I
think I could have left it longer. About the only thing to note is that feta is quite
salty, and it needs no extra salt.
Into the office this morning with a firm resolve to do something else than play with
panorama viewers. I failed. With a bit of comparison, it didn't take me too long to debug
my scripts, and I even found a way to find the error in the configuration file.
firefox complained about it:
And after fixing that, Emacs indeed agreed that the data was well-formed. Gradually
worked out other strangenesses, notably the tricks necessary to get Hugin to create correct equirectangular
images: in the Stitcher tab you need to set the field of view to 360×180, and to set
the crop to the complete image. Arguably this should be done automatically, but at the very
least it should be easier.
Finally I had a number of panoramas generated
“automatically”. Now I need to tune the configuration file.
I had just finished that when a message came through on the Hugin
mailing list: a new HTML5-based
panorama viewer, just what I was looking for when I decided to try SaladoConverter. Tried it out: it's
very new, and caused significant problems for my web browser and caused X to crash Yet Again. Still, another one to try.
Despite all, managed to do a bit of work in the garden. It didn't seem long enough, but I
managed a full basket of weeds. If I set myself a target of a basket of weeds a day, I
should be able to stay on top of things. But that's all I've done in the past 4 or 5 days.
Into town today to talk to Peter O'Connell about my investments. We're living in
interesting, but not good, times. What do we do? Move to cash? The interest rate is
dropping there too. We can just hope that things blow over soon.
While in town, dropped in to a couple of shops looking for a new mattress. My old one must
be 20 years old, and it shows. But where do I get a replacement? It should be
straightforward: it's a standard size, 90×200 cm. But Australian bed manufacturers don't
believe in standard sizes, nor even in
the metric system that has
officially been in force for 40 years. I can only get a “single” (3'×6'2" or
92×188 cm) or a “long single” (3'×6'8" or 92×204 cm) or even a “king single”
(3'6"×6'8" or 107×204 cm).
Never mind, they can be made to order, though the concept confused the hell out of at least
one of the saleswoman (“is 900 the width or the length?”). And the prices are not good.
The cheap mattresses feel just that even when they're new, with the springs very evident.
The cheapest I could find that might fit the bill would cost $359 at Forty Winks or $450 for a better one at The Bed Shop. That doesn't reflect the cost of
making a special size: it seems that mattresses cost that much nowadays. More searching
required.
While in town, also to the Friends of the Ballarat
Botanical Gardens to back up their computer. To my surprise, everything Just Worked:
on plugging in the disk, I was offered a number of choices of what to do with it. Possibly
out of embarrassment the option of using it as a backup disk was hidden beyond the end of
the too-short selection window, but once I found it it was relatively trivial to set it up
and start a backup. Next time I'm there I'll take a look at what it did.
It was five years today that we first inspected the house
we now live in. We liked what we saw, and we moved in less
than 6 weeks later. Since then, we've changed lots of things. I took photos then, so
we have a good comparison. Here some of
the more extensive changes:
When taking photos like today's comparative photos, it's good to have the original at hand
to compare. In the past I've done things like printing out a hard copy or dragging a laptop
around with me, but both are clumsy. Recently it occurred to me that my GPS navigator is
really an adapted tablet, and it does have software to display photos (as long as you
truncate the names), so today I copied the photos to the navigator and carried that around
with me for the comparisons.
Did it work better? Marginally. I can put the navigator in my pocket, but clearly not a
laptop. But holding it along with the camera is still clumsy, and I didn't reproduce the
viewpoints any better than on previous
occasions. That's due at least in part to the fact that this particular device really
isn't bright enough in even cloudy outdoor conditions, so it was difficult to see. But at
least it has made it clear that “one size fits all” is not sufficient. I want a real
keyboard when I use a computer, but for these purposes things need to be more compact.
Autumn is coming to an end, but some plants are only now coming into bloom. I suppose
that's normal enough for the Dahlia
imperialis and the Senna
aciphylla:
On the other hand, some of
the eucalyptus that normally flower at
this time of year are not in flowering now. It looks as if they have already done so,
though I don't recall seeing it.
This business with the Victoria
Police handling of Yvonne's speeding fine has shocked
me so much that it has taken me 2 weeks to do much in the way of following up. But we only
have another 3 weeks to object, so it's time to do something.
As last time, I started by calling the
RACV motoring advice line on (03) 9790 2190
and spoke to Paul, who was quite amused by the incompetence of the police. I suppose
there's a funny side to it. He suggested that I first call the Victorian Ombudsman on (03) 9613 6222. Did that
and was connected with Dagmar, who told me that they were not responsible for the police and
suggested I contact the Ethical Standards Department on 1300 363 101 or the Office of Police Integrity on 1 800 818 387.
This doesn't really seem to be a question of ethics, so I called the Office of Police
Integrity first. That was quick: they don't handle traffic offences. Why not? The
recording suggested I contact Civic Compliance Victoria on (03) 9200 8111, so I did, by no means for the first
time, and spoke to Claire. This time I got the additional information that in response to
Yvonne's letter in February an internal review was denied, whatever that may mean—probably
that they didn't think the matter serious enough to worry about. And of course she didn't
address the issue of police culpability, and she had no idea why the application for
revocation was refused, just suggested that we submit a second application, something that
wasn't mentioned in the refusal. She didn't seem to understand that there are two different
issues at stake here.
So I called the Ethical Standards Department and spoke to Senior Constable (aren't they
all?) Chapman, who told me that this wasn't an ethical issue, and so they couldn't do
anything. I asked who could do something, and she suggested the Ombudsman or Civic
Compliance. She gave me the email address ethical.standards@police.vic.gov.au, but
said “I doubt I'll get the response you're after”. She also didn't seem to understand that
there are two different issues at stake here.
So: it really seems that the police can do what they like in this area. They may address
the issue of ethics—internally—but not the issue of efficiency at all. I'm reminded of the
events that led up to Eureka
Stockade. I don't know of any modern Western police force with so little
accountability.
Nice weather today, and spent some time in the garden weeding. To make up for the days I
didn't do anything, managed two baskets full of weeds today, including some from the “far
east” bed on the other side of the eastern garden path.
The calendulas have taken over a fair
amount of the area, and there's plenty of difficult grass there, but it's looking better.
Still a long way to go.
Do you have a comment about something I have written? This is a diary, not a
“blog”, and there is
deliberately no provision for directly adding comments. It's also not a vehicle
for third-party content. But I welcome feedback and try to
reply to all messages I receive. See the diary overview for more details. If you do
send me a message relating to something I have written, please indicate
whether you'd prefer me not to mention your name. Otherwise
I'll assume that it's OK to do so.
