It didn't quite make it as hot as yesterday, but it started off a lot earlier, and it looked
as if it would get worse. As a result, spent just about all the day inside with the air
conditioners running. In the process discovered that the extreme temperatures last month
weren't as bad as in the previous two years, though the average was higher:
The Buddleja davidii that Mike
Sorrell gave me last week looks ready to bloom Real Soon Now. The flowers should be blue,
but as Mike said, the buds are almost black:
The following graph is smoothed, so it doesn't quite show the maximum:
As a result, of course, spent the day indoors again. But how reliable are these
measurements? This is the highest I have recorded with my weather station (over about 26
months). The previous high was 44.9° on 11 January 2010. But
is it correct? Other measurements suggest that the temperature was far from the all-time
high we've measured here: that was on 29 January 2009,
where we measured 46° on the spirit thermometer on the verandah.
So what's a difference of 0.6°? Within the range of the accuracy of the old thermometer.
But the location is important. I still have another thermometer on the verandah (now
electronic), and it recorded a maximum of 39.6°, quite a significant difference. In
addition, the other temperatures on the WunderMap round Dereel were noticeably lower, by 2 or 3 degrees.
So: the question is how representative the weather station values are. It's interesting to
note that, unlike on other days, the highest temperatures were in late afternoon, round
18:00. Is the temperature sensor affected by the sunlight? To be investigated.
It did give us a chance to measure the water temperature in the pond, though. Yvonne had been concerned that it would be too hot for the fish, but in
fact the temperature at the surface was as I had expected, about 27°. And at the bottom it
was only 22°, still a little higher than the 20° I had expected.
Not everything in the garden has grown the way I wanted. Last year I planted
a Hosta, just to watch it wither and die. And
the Pandorea that we planted
over two years ago has never really
flowered much. But lately both have come into their own:
So maybe the hosta was just fading away for winter.
The Delphinium “Völkerfrieden” that we
bought at Lambley Nurserylast year also faded on us, and I thought it was dead and
nearly threw out the remains. But in the end I replanted it, and it's now coming up with
buds. I really need to understand the plants better.
After getting the Olympus E-510 in August 2007, It took me quite
some time to get good results from the postprocessing. Some of it was a learning
experience, some was lack of suitable software. I've learnt a lot since then, and I've
found ways to improve the quality of the images, but can DxO Optics "Pro" do
better? Trawled through my old diary entries and got as far as three years ago, when I had serious issues with exposure and gradation. It proves
that, for some reason, some of the images really were underexposed, but DxO brought much
better results. Here the comparison of processing the same image with ufraw 0.13, ufraw 0.14.1, dcraw and DxO. Mouseover shows the next image for comparison.
Also reprocessed the panoramas. I couldn't completely fix the verandah panorama, because I
took it without parallax correction, but the results look a lot better.
How well could I have done without DxO, just with better tuned ufraw? I don't know,
and I don't intend to look yet. First there are other images to compare.
The Bureau of Meteorology had originally
expected today to be even hotter than yesterday, but at the last minute decided that the
temperatures would be 3° less. In fact, they were closer to 15° less: although the
temperature shot off early in the morning, it hit 30° and hung around there all day:
A good thing too: the heat was really getting on our nerves.
Spent much of the day with further reprocessing old photos. The results were quite useful.
On 4 April 2009 I had another case where a panorama changed from
“useless” to “acceptable”:
If it hadn't been for the reflections of the sun in the lens, it wouldn't have been at all
bad, though it's clear that it challenges the limits of the dynamics.
Then I looked at some comparisons I had done on 12 April 2009, where I
had had extreme problems with contrast. At the time I hadn't realized that some of the
camera functions, such as “vivid” and “soft” images, only applied to
the JPEG output. In addition, I didn't find
a camera JPEG image for all the comparison photos, so today's comparisons aren't all they
could be.
A few days later I received a comparison
photo for the first sequence from Michael <no surname> in South Australia,
processed with Photomatix. In the following
comparison I have the original from the camera, the best I could do to improve it with
ufraw, the results from
Photomatix and the output from DxO Optics "Pro". As
usual, moving the mouse over the image gives the next in sequence.
I think it's fair to say that DxO produces by far the best image, and probably about as good
an image as I could hope for from this particularly difficult scene. The shadows are more
dynamic than Photomatix, and the highlights are much more natural. Conceivably Photomatix
could do better—Michael said that he didn't try too hard, which is understandable—but then
DxO might do better too, and there's only so much work I want to put into these images.
The other image wasn't really a good comparison 3 years ago, but here it is with DxO, which
once again does a good job:
A couple of months later I tried some
different scenes, this time comparing images from the Olympus E-510 and the
E-30. No hard
contrast, just overall image gradation. Today I was more interested in the postprocessing,
and tried the following: JPEG original from the E-30, DxO with no correction, DxO with
default correction, DxO with “realistic” HDR-style tone mapping, and DxO with “artistic”
HDR-style tone mapping.
The tone mapping doesn't seem to make much difference, presumably because the dynamic range
was limited, but the “artistic” version makes it gaudy enough. Better than default? Here's
the comparison:
So far my experience has shown that DxO Optics "Pro" is quite
a useful program. The only problem, apart from the fact that it needs Microsoft to run, is
that it uses so much memory that I can't run it on VirtualBox on my 32 bit machine. I started trying to upgrade to amd64 (64
bits, without the memory limitations) nearly 8 months
ago, and gave it up two weeks later because of X issues. I then tried again in August, but the pain of upgrading was just too
much.
By today I had forgotten some of the pain, and so started trying to upgrade again. I've had
the machine running as a test machine all this time, so all I really needed to do is to
upgrade the kernel and the ports. The kernel upgrade was trivial. The ports, once again,
were not.
First,
===>>> Warning: java/jdk16 is interactive, and will likely
require attention during the build
So? Why does it stop there and wait (maybe all night) for me to take notice of it for the
hundredth time? The same happens for postfix.
Then there was one I hadn't seen before:
===>>> Launching child to update php52-5.2.17_1 to php52-5.2.17_5
===>>> Port directory: /usr/ports/lang/php52
===>>> This port is marked DEPRECATED
===>>> PHP 5.2 series is not supported anymore, migrate now
===>>> If you are sure you can build it, remove the
DEPRECATED line in the Makefile and try again.
OK, so what's the replacement? php53? No, they've done almost the right thing and
removed part of the version number from the preferred version. It's called php5,
though it should be called php of course. But I had to go looking for that.
People tell me I should be reading (/usr/ports/)UPDATING, but why? Why can't that
information be output when the build fails? In any case, I couldn't find any mention
of php52 there (in a file nearly 100 pages long). And after that, of course, I
could wait for the messages about java16 and postfix.
The next one was a superb shoot-in-foot incident:
===>>> Upgrade of portmaster-3.9.1 to portmaster-3.11 succeeded
===>>> Delete portmaster-3.9.1.tar.gz? y/n [n] y ===>>> Delete portmaster-3.9.1.tar.gz.asc? y/n [n] y ===>>> Returning to update check of installed ports
===>>> Launching child to install devel/qmake4
/usr/local/sbin/portmaster: /usr/local/sbin/portmaster: not found
Why that? It says it has upgraded it, so there was no reason to keep the old version. But
there was no new porrtmaster installed. Why not? I couldn't even find the backup I
had asked for, so I had to reinstall it first.
And that's as far as I got in a whole day. This process should run in the background
without my intervention and give me a summary of anything that fails. And failure should be
a rare event, not like every time I've tried to upgrade ports in the last few years.
No wonder Kirk McKusick has given up on straight FreeBSD.
The weather is finally mild enough to work in the garden, so out to inspect the north bed.
It's looking very good now, but the weeds are also feeling at home, so spent a lot of time
removing surprisingly few weeds.
Continued with my ports build today. Came into the office and found an old acquaintance:
cd x2p; LD_LIBRARY_PATH=/src/FreeBSD/ports/lang/perl5.12/work/perl-5.12.4 /usr/bin/make s2p
make: don't know how to make s2p. Stop
*** Error code 2
Stop in /home/ports/lang/perl5.12/work/perl-5.12.4.
In summary: I still don't know how to build perl from the Ports Collection. Maybe if
I blew everything away and started again, it would be better. But that's not sure, and it
could be even more pain. As I did last time, I just downloaded the package, once I found it:
=== root@defake (/dev/pts/0) /usr/ports 90 -> pkg_add -r perl5 Error: Unable to get ftp://ftp.freebsd.org/pub/FreeBSD/ports/amd64/packages-8-stable/Latest/perl5.tbz: File unavailable (e.g., file not found, no access)
pkg_add: unable to fetch 'ftp://ftp.freebsd.org/pub/FreeBSD/ports/amd64/packages-8-stable/Latest/perl5.tbz' by URL
=== root@defake (/dev/pts/0) /usr/ports 91 -> pkg_add -r perl Fetching ftp://ftp.freebsd.org/pub/FreeBSD/ports/amd64/packages-8-stable/Latest/perl.tbz... Done.
Removing stale symlinks from /usr/bin...
Skipping /usr/bin/perl
Skipping /usr/bin/perl5
Done.
Creating various symlinks in /usr/bin...
Symlinking /usr/local/bin/perl5.12.4 to /usr/bin/perl
Symlinking /usr/local/bin/perl5.12.4 to /usr/bin/perl5
Done.
Cleaning up /etc/make.conf... Done.
Spamming /etc/make.conf... Done.
Cleaning up /etc/manpath.config... Done.
Spamming /etc/manpath.config... Done.
=== root@defake (/dev/pts/0) /usr/ports 92 ->
But that's perl-5.12. There's a 5.14 out there. That must be newer. Why have I
been given an old version of perl? See if I care.
Then there was docbook. I seem to have 7 different versions installed. It
reinstalled 6 of them:
drwxr-xr-x 2 root wheel 512 Jan 5 11:44 docbook-1.4
drwxr-xr-x 2 root wheel 512 Aug 8 15:26 docbook-4.1_4
drwxr-xr-x 2 root wheel 512 Jan 5 11:44 docbook-4.2
drwxr-xr-x 2 root wheel 512 Jan 5 11:44 docbook-4.3
drwxr-xr-x 2 root wheel 512 Jan 5 11:44 docbook-4.4_2
drwxr-xr-x 2 root wheel 512 Jan 5 11:44 docbook-4.5_2
drwxr-xr-x 2 root wheel 512 Jan 5 11:44 docbook-5.0_1
I had intended to do some more weeding in the north bed today, but looking around, it became
clear that a number of plants have suffered from the recent hot weather, in particular in
the “old” succulent garden and south of the house:
Continued with my attempts to update the ports tree today. More pain. First it died
in libass, whatever that may be:
===> Configuring for libass-0.10.0
cd: can't cd to /home/ports/multimedia/libass/work/libass-0.10.0
env: ./configure: No such file or directory
===> Script "configure" failed unexpectedly.
What's that? Took a look in the build directory, and sure, ./configure is there.
But the directory name is libass-0.9.13, not libass-0.10.0. Why that? Did
a make clean, built, and it installed.
It went on, though, although I removed the C option. Later I got a crash after
this message:
===>>> pkg-message for compat7x-amd64-7.3.703000.201008_1
*******************************************************************************
* *
* Do not forget to add COMPAT_FREEBSD7 into *
* your kernel configuration (enabled by default). *
* *
* To configure and recompile your kernel see: *
* http://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig.html *
* *
*******************************************************************************
And of course COMPAT_FREEBSD7 was set in the kernel config. Ran things again, and
the message didn't occur again. Instead, firefox died with:
configure: error: Perl 5.006 or higher is required.
===> Script "configure" failed unexpectedly.
Please report the problem to gecko@FreeBSD.org [maintainer] and attach the
"/home/ports/www/firefox/work/mozilla-release/config.log" including the output
of the failure of your make command. Also, it might be a good idea to provide
an overview of all packages installed on your system (e.g. an `ls
/var/db/pkg`).
*** Error code 1
That's ridiculous. I had just (with much pain) installed version 5.12.4. Did some old
version get left behind? I should have checked config.log, but instead I checked
with ktrace and found the same thing:
46354 sh CALL execve(0x800c06e08,0x800c06eb0,0x800c2c608)
46354 sh NAMI "/usr/local/bin/perl5.10.1"
46354 sh RET execve -1 errno 2 No such file or directory
46354 sh CALL write(0x2,0x800c16080,0x32)
46354 sh GIO fd 2 wrote 50 bytes
"./configure: /usr/local/bin/perl5.10.1: not found
"
46354 sh RET write 50/0x32
46354 sh CALL exit(0x7f)
Well, it's right there. There is no perl5.10.1. But the error message is
misleading: it suggests that there is an old version of perl there. And where did it
get that from, anyway? It didn't try anything else. This one is apparently (marginally) my
fault: I ran portmaster with the C option, which tells it not to run
a make clean before building. Of course, the build process should have noticed that
the config.status was out of date, but that one is understandable. Moral: don't use
the C option.
So finally firefox version 9 was
installed, so portmaster went off and started building firefox version 5.
Why? And how? We don't have a version 5 in the Ports Collection any more, but there's a
/var/db/pkg/firefox-5.0,1 alongside /var/db/pkg/firefox-9.0.1,1. Did
a pkg_delete on firefox-5.0,1, and of course tore firefox-9.0.1,1 apart
too:
=== root@defake (/dev/pts/0) /usr/ports 118 -> pkg_delete /var/db/pkg/firefox-5.0,1 pkg_delete: '/usr/local/bin/firefox' fails original MD5 checksum - not deleted.
pkg_delete: '/usr/local/include/firefox/Allocator.h' fails original MD5 checksum - not deleted.
...
=== root@defake (/dev/pts/0) /usr/ports 120 -> pkg_delete -f /var/db/pkg/firefox-5.0,1 pkg_delete: no such package 'firefox-5.0,1' installed
=== root@defake (/dev/pts/0) /usr/ports 121 -> wh firefox No files found
So I had to reinstall it. That was fast, but ineffective:
=== root@defake (/dev/pts/1) /usr/ports/www/firefox 40 -> make install === root@defake (/dev/pts/1) /usr/ports/www/firefox 41 -> firefox su: firefox: command not found
The problem here is that the Ports Collection notes in the build directory whether the
package has been installed or not. To get it to reinstall, you first need to remove a
secret file:
Edwin Groothuis tells me there's another target, reinstall, that does this
for you.
Looking at what was going on, this seems to be a bug in portmaster. It finds a
directory /var/db/pkg/firefox-5.0,1 and establishes that it's out of date, and that
the directory in the Ports Collection is /usr/ports/www/firefox, which currently
builds irefox-9.0.1,1. But it doesn't check whether that is installed or not. Maybe
it would have rebuilt it and reinstalled it. If it does a make clean it won't see
that it's already installed any earlier.
Still, finally I had firefox upgraded. Easy, isn't it? Enough pain for a day.
We have two different kinds of Clematis
on the south side of the verandah, planted mainly because we preferred the second one, which
is darker. This year, at any rate, it also flowered much earlier, so we had double the
time. Today I found a single flower on the old vine, while the other one is in full flower:
It's been over a week since I pulled
the Gazanias out of the north-central
bed, and gradually the remaining weeds are dying off. I suppose I should mow the area, but
did some thinking and decided that it's probably better to remove the
remaining Aeonium first, so did that.
The real issue is that we need to recreate the brick border of the bed, which has been
knocked around in the course of time, and that seems like too much work.
Instead, grabbed some of the Gazanias and planted them round the south side of the old
succulent bed to the south of the verandah, where it had been so dry that even some of the
Gazanias died off.
Also attended to the shade area to the west: the creepers have now made it to the top,
and are somewhat hindered by the shade cloth there, so it's time for that to go—almost.
There's still not enough shade from the creepers, so I'll just leave it loose for the time
being.
We watched an old film this evening, Brothers in Law, made in 1957. The film itself is mixed—I liked it, Yvonne didn't. But one scene got us: two men and a woman were sitting
at a table. Woman gets up and goes. Men almost stand up as she goes. Used to be normal,
but where do you see that happen nowadays?
The real issue, though, was the almost. I was taught to stand up completely, not
just to pretend to get up. And we even had a term for this incorrect behaviour: “Raise
Bottom Slightly”. That seems to have disappeared into history; I couldn't find any
reference to it on the web. Should I add it to Wikipedia? I don't have any references. I suppose this will be here soon, but that
doesn't really help.
A good three years ago we bought
an artichoke plant from Diggers. It hasn't borne much fruit, but we did
manage to get two artichokes last March.
And so when we visited Lambley Nurserylast October, I wasn't overly keen on buying
more. But we did, and it was a good thing too: they're bearing nicely. Why did we have so
little success with the stuff we bought from Diggers? I've been wondering that since we
went there, and an obvious reason is that I was then an even less experienced gardener than
I am now. But that doesn't explain why plants in the same bed perform so differently.
How long do you cook artichokes? We keep discussing this. This time it was 20 minutes, not
enough. Somewhere between 30 and 45 minutes is probably right. But the artichokes tasted
very good—finally something worth growing at home.
Another windy day for the house photos, but managed to get most of them anyway. I've been
taking these panoramas for over three years now, but I continually run into new experiences:
A few weeks back I changed the use of flash in the verandah photos to use TTL flash for
the first image, which meant taking the first image with flash, turning off the flash
unit, taking the other two, turning the flash unit on again and moving on to the next
position. And from time to time the camera would hang when the flash was turned off.
It happened again today. After turning it on again, the camera still hung—but only for
a while. Then there was some memory activity (the LED flashed for a while) and it
continued.
This happened twice, and each time it was after the flash fired at full intensity
instead of normally—something that I need to work on. So there appears to be some
connection there.
One of the panoramas didn't close properly: Hugin found control points for all except one overlap. It was a 360°
panorama, so that was “enough”, and it told me I had a good match: average error
1.3 pixels, maximum error 3.8 pixels. But when I looked at the preview, things were
anything but good. The trees on the left don't line up:
Adding control points at that overlap didn't help; the preview still showed this
enormous discontinuity. Strangely, though, the result looked OK: something in the final
stitching must have improved the situation.
This seems to be another case where history makes a difference with Hugin. In
this case, it had determined that the effective focal length of the lens was 9.11 mm,
when in fact it was closer to 8.9 mm, and I think this might have made the difference.
It was a pretty grey day today, and the images looked pretty washed out. Tried some new
settings with DxO
Optics "Pro", and they seem to help. Here the normal and then the enhanced
view, as the mouseover shows:
That was done with the “Artistic” variation of the “single shot HDR” “preset”. In
general it looks too gaudy, but it's probably what I need here.
Did more exposure comparisons with automatic and manual exposure settings for the
components of the garden-se photos. Today there was no sun, so the exposure differences
weren't as extreme as they might have been.
The second image was taken with automatic exposure, offset by -0.7 EV by mistake. But
the results look almost identical. Clearly this kind of lighting isn't as much of a
problem as the strong contrasts in sunshine.
The photo from the south-east corner of the house also looks like it could do with
automatic exposure. In particular, the right-most component is extremely underexposed
with uniform manual exposure:
Probably another one to try with automatic exposure, but the “HDR” processing
and Hugin generated a reasonable result, though the shadow detail on the right
leaves something to be desired:
The hot weather of the last week or two has made its mark. The grass is brown, and the dam
is at its lowest level since September 2010. Here four
weeks ago and today:
Didn't have much time for my ports build today, but that didn't mean things went smoothly.
While building nmap, got this message:
c++ -c -I../libdnet-stripped/include -I/usr/include -I../nbase -I../nsock/include -O2 -pipe -I/usr/include -fno-strict-aliasing -Wall -fno-strict-aliasing -DHAVE_CONFIG_H -DNPING_NAME=\"Nping\" -DNPING_URL=\"http://nmap.org/nping\" -DNPING_PLATFORM=\"amd64-portbld-freebsd8.2\" -D_FORTIFY_SOURCE=2 NpingOps.cc -o NpingOps.o
gmake[3]: *** No rule to make target `nmap.h', needed by `utils.o'. Stop.
That wasn't a portmaster issue: it happened when I tried to build it in the normal
manner. On three different machines. With two different versions of nmap.
And on checking, discovered that there is no mention of nmap.hanywhere in
that directory. The file itself was in the parent directory.
Discussed it on IRC, and Jürgen Lock suggested that I try it without any preset environment
variables (su - root -c /bin/sh). And yes, that worked. So it's an environment
variable, but which? Or is it? Went back to build the port again, this time with my
standard bash shell and environment, and it worked too. Just more pain.
While watching the build information fly by—I have collected 336212 lines, or about 5,500
pages of it so far—discovered a number of these:
===> Registering installation for poppler-glib-0.18.0
So far there have been 9 of them. Why? And what is this port?
Things continued as far as:
===> Installing for linux-f10-fontconfig-2.6.0
===> Generating temporary packing list
===> Checking if x11-fonts/linux-f10-fontconfig already installed
cd /home/ports/x11-fonts/linux-f10-fontconfig/work && /usr/bin/find * -type d -exec /bin/mkdir -p "/compat/linux/{}" \;
cd /home/ports/x11-fonts/linux-f10-fontconfig/work && /usr/bin/find * ! -type d | /usr/bin/cpio -pm -R root:wheel /compat/linux
617 blocks
ln: /compat/linux/etc//fonts: File exists
*** Error code 1
Then I gave up for the day. But the Ports Collection wasn't done with me. Tried to back up
the photos I had been processing and got:
=== root@defake (/dev/pts/1) /usr/ports 98 -> syncphotos Sat Jan 8 18:43:43 EST 2012
rsync: not found
rsync: not found
Where did rsync go? Was it in the
process of building it? No, it seems that it thought it had built it, but it was gone.
Looking through the log files, I find:
===>>> Upgrade of rsync-3.0.8 to rsync-3.0.9 succeeded
Some success! It seems that I shouldn't have even tried to use portmaster. But in
general, the amount of trouble I've had here is an order of magnitude worse than any other
system I have used, and it's not the first time. If I could find something better that
didn't want to tell me how to think, I'd move to it immediately.
It's been two weeks since I received
my hard disk docking station. The first experience wasn't good, and I didn't have
a SATA disk handy to put in it; the only one
was my 2 TB backup disk, and after seeing what happened with an SD card, I didn't want to
risk that. Today finally got round to taking a spare SATA disk out of the external housing
and putting it in the docking station, in the process discovering confirmation of what I had
suspected, that the eSATA interface was
only for the SATA drive:
The SATA disk fitted better than the attempts with the PATA disk, and it powered up. And
that was all. The system didn't recognize the disk, so I rebooted—I've had trouble before
with another low-quality eSATA external housing—but it still wasn't recognized. What's
wrong with it? Do I care?
On with the ports upgrade today, without running into many problems. It wasn't until I
rebooted defake that I discovered that I could no longer communicate with
it: xterm had gone away. Why? That's the second program that just disappeared,
after rsync.
rsync wasn't to be outdone, though. It went away again! I don't understand
that, since it happened when I thought I hadn't done anything. But I can't be bothered
crawling through the equivalent of 5000 pages of log information to try to piece together
what went wrong.
Michelle Miles along today to visit and look at some horses, staying overnight. Cooked
cous cous, in the process running into some
dubious quantities: 20 g of pepper? I don't know where that came from, but it wasn't from
the original recipe, and it was too much. Also tried some salted lemons, which completely
threw the delicate salt balance. Nonetheless, not too bad a meal.
===> Registering installation for poppler-glib-0.18.0
...
===>>> Re-installation of poppler-glib-0.18.0 succeeded
===>>> Update check of installed ports complete
That must be about the twelfth reinstallation of poppler, but it's finished. Well, it
reinstalled poppler many times, but it removed xterm, and it's still gone. Why?
What else is gone?
It's been 5 days since I started this update. The intention had been to do it faster than a
complete rebuild of the ports. But it doesn't seem to have been. What went wrong? Jürgen
Lock suggests that the presence of the FORCE_PKG_REGISTER environment variable or
the -C option (don't run make clean) that I had at the beginning might be to
blame. But I don't care. It has just taken too long. Went back to reinstall the ports from
scratch, once again going through the pain of hundreds of default configurations. And then
Jürgen told me about the -P option for portmaster, which downloads the
precompiled packages. Maybe I should have tried that. I probably will next time.
Finally today the rotator arrived that I had bought in the USA four weeks ago. Into town to pick it up, in the process taking the
panorama bracket I bought 6 weeks ago to ask Paul Sperber of Ballarat Automotive if he knew anybody who could remove the bottom
rotator. He did find a screw to remove, but it didn't help. Later I found I could then
take the rotator apart by removing the locking screw, and inside found two more screws
attaching it to the rail. But even after that, I couldn't get it off. It appears to be
glued to the rail with epoxy, and it's hard as nails.
The rotator is a Sunwayfoto DDP-64M, which looks very much like the Manfrotto 26018 that I already
have, and which their terminally broken web site calls a Manfrotto 300N.
The similarity is intentional: the Manfrotto is part of the panorama bracket I bought 8 months ago, in fact the
only part that I'm happy with. But I'm planning to sell the bracket, and it's firmly
attached, so the Sunway is the replacement.
How do they compare? Manfrotto has a reputation as one of the leaders in this area, while
Sunway is a Chinese newcomer. They look pretty much the same. Here the Sunway, then the
Manfrotto, both mounted on the Manfrotto 3416 levelling base:
They both have the same number of increments, from 5° to 90°, and they're set up the same
way, by screwing a screw into a specific hole. There are two other screws, one for locking
the mechanism so that it can be removed from the base, and the other for positioning the top
plate relative to the detents. And they both work well. Comparing them, there are minor
differences:
Both rotators tie the selecting screw to a wire, and the other end is attached to the
lock screw. That works well with the Manfrotto, though in the past I've had the wire
stick out into the field of view, but on the Sunway the layout is somewhat different,
and for some angles the wire bends strongly, as shown in the first image.
The Sunway has somewhat firmer click action than the Manfrotto. That's good: from time
to time I find myself overshooting the position with the Manfrotto, and I have to go
back again.
The Sunway has 5° marks starting at 0°, going to 90°, then back to 0° again. The
sequence repeats on the other side. By contrast, the Manfrotto has a single scale going
to 360°:
That's a matter of preference, I suppose, but I prefer the 360° method. That makes it
less possible to miss out a shot.
The lock knob on the Manfrotto is circular, while on the Sunway it's a lever. The
circular knob is not easy to tighten enough; the lever works better.
Both heads have the same diameter, but the Sunway is slightly shorter in height than the
Manfrotto.
In summary, then, the Sunway is a good replacement for the Manfrotto. It would have been a
lot cheaper, too, if I hadn't had to order it from the USA. Postage cost $31.92 (the exact
cost of postage), and it took 4 weeks to get here. In that time, coincidentally, I ordered
a screw from Hong Kong on the same day as the rotator. It arrived (free postage), I liked
the look of it, and ordered another one. The second one also arrived today.
So now I have all the components to build a functional panorama bracket that does almost
everything I want. It took me a while to think out how to put the hardware together to do
it, particularly since I wasn't able to remove the toy rotator at the base of the bracket,
but now I can mount the camera horizontally or vertically and rotate about the other two
axes:
The Manfrotto levelling base. It's not very clever, since it has three adjusting screws
with limited travel. Something like a pan and tilt head would be better, and I'm still
thinking of replacing it. But for the time being it will do.
A focusing rail to move parallel to the horizontal axis of the sensor.
The bottom part of the BF3 panorama bracket. If I could get the rotator off this thing, and if
it were just a little wider, I wouldn't need the rail in the point 3; I could just mount
the bracket directly onto the rotator with an appropriate clamp. But as the photos
show, it's too short when the camera is mounted horizontally.
The camera is mounted on another rail moving along the lens axis, to move the entrance pupil
to the correct position relative to the horizontal rotator on the panorama bracket. For
vertical (“portrait”) orientation it fits directly into the clamp on the rotator (first two
photos above). For horizontal (“landscape”) orientation it is mounted on the L bracket,
also part of the panorama bracket, and the L bracket fits into the rotator (third photo).
How does it match up? Total cost was:
$89.95 for the panorama bracket.
$27.50 for two focusing rails.
$170.92 for the rotator.
That's $288.37, more than I had hoped for, and 60% of that price was the rotator. I suspect
that price would come down by about 30% if somebody in China or Hong Kong would offer them.
I haven't mentioned the levelling base, for which I don't have a price, since I bought it
with the Manfrotto panorama bracket (which together came to $387.00 including postage). At
B&H it costs $94.50 new. I'm planning to get rid of it anyway, not because
of the price, but because it doesn't do its job very well. I'd hope to find a suitable
leveller for about $50.
I've been looking at panorama brackets for years now, and so far I haven't found anything, at any price, that fulfils all my
requirements. And those prices can be high, well over $1000. Here the latest incarnation
of my requirements list, and how the present setup stacks up:
Rotate the camera in three directions about
the entrance pupil of the lens.
No bracket allows rotation in three directions. They all allow rotation about the
vertical axis (otherwise they wouldn't work at all), and the better ones allow rotation
about the horizontal axis perpendicular to the lens axis. The third axis, roughly the
lens axis, is really the question of horizontal or vertical orientation, and I can do
that too.
Level the camera independently of the tripod, to make it possible to rotate the camera
about a vertical axis.
The levelling base does that, not very well. I'll replace it with something offering a
better range of adjustment some time. In principle a pan and tilt head would do the
job.
Rotate the camera in specific increments for equally-spaced images. At the very least,
provide angular markings so that you can set them yourself.
This is what the rotator does.
Provide a level indicator (usually
a spirit level).
Currently I have this in the leveling base. It really should be on the panorama
bracket, and I'll see what I can find for that.
Provide scales for accurately positioning the camera, in particular a scale parallel to
the lens axis to help set the entrance pupil position.
The FOTOMATE rails have scales. The upper one is too short for some of my lenses; I'll
have to investigate further, but I have other rails in the meantime.
For rotators: allow setting the start position.
Yes. That's the other lock screw, which allows moving the top plate independently
of the detents.
It's been two months since I've paid much
attention to the Nickel-Zinc
batteries I bought in October. I haven't
been demanding so much from the flash unit, but today it finally told me they were
discharged. And indeed they were: the voltages were down to 1.549 V, 1.486V, 1.437 V and
1.381 V, the last one that I had already marked as not being quite up to the same level as
the others. It's difficult to be exact, because it was clear that they were still
recovering from the flash, and the voltages were gradually rising. Still, a long way below
the 1.6 V in the specifications. After charging, the voltages were 1.832 V, 1.832 V, 1.829
V and 1.826 V (this last the marked battery).
As planned, continued building the ports on defake with my old, incomplete build
method. In principle it should work, but for some reason today the build died at the start
because I had two rules to build X. How did that
happen? Checked out the last version of the rules file again, and discovered it was already
in there. My best bet is that previously the duplicate rules didn't do any harm, while now
they caused the build to fail.
What did do harm was checking out the rules file. It seems that in so doing I
overwrote some changes I had made since the last checkin, and I was missing rules for a
number of ports. In the course of the day, had to update the rules file a number of times,
but nothing else went wrong.
In the process, found the reason for my FORCE_PKG_REGISTER environment variable:
===> opera-linuxplugins-11.60 is already installed
You may wish to ``make deinstall'' and install this port again
by ``make reinstall'' to upgrade it properly.
If you really wish to overwrite the old port of www/opera-linuxplugins
without deleting it first, set the variable "FORCE_PKG_REGISTER"
I noted yesterday that it took twice as long for my rotator to arrive from the USA as it did for the screws from China.
It also cost a lot more, $31.92. This was sent by “Priority Mail
International”, USPS' allegedly slightly-better service (at least it's more
expensive). Today I received another item from the USA, sent “First Class Mail” (the
cheapest) on 29 December 2011. So, bearing in mind the 1 day difference
between the USA and here, the more expensive service took 27 days, and the cheap one took 11
days. This isn't the first time I've seen this: last September I noticed exactly the same issue, and not for the first time. I
wonder if this is a general problem.
The package that I got today was an oven thermometer, very similar in function to the one
that failed. As far as I could tell externally, the probes were identical:
That wasn't the case, though. I put the old probe in the new thermometer, and it indicated
a temperature of 54°. The correct probe indicated 21°, as did the other probe in the
thermometer for which it was intended.
I suspect that the problem with the old thermometer was damage to the temperature probe
cable. It works fine at normal temperatures, but stops working when it gets warm. Under
those circumstances, it's interesting to note the lengths to which the manufacturer went to
tie the probe cable in knots:
The ports build continued relatively smoothly today. multimedia/dvdauthor failed:
===> Applying FreeBSD patches for dvdauthor-0.7.0_1
Ignoring previously applied (or reversed) patch.
1 out of 1 hunks ignored--saving rejects to src/subgen-image.c.rej
...
=> Patch patch-src__subgen-image.c failed to apply cleanly.
Typical reasons for that are out of date ports, so synchronized my repository, ran cvs
up -Pd, and tried again. No improvement. Further investigation revealed:
=== grog@defake (/dev/pts/1) /usr/ports/multimedia/dvdauthor/files 2 -> l total 1
drwxr-xr-x 2 grog lemis 512 Oct 2 11:29 CVS
-rw-r--r-- 1 grog lemis 1549 Oct 23 2003 README.FreeBSD
-rw-r--r-- 1 root lemis 2867 Jan 11 14:10 patch-src__subgen-image.c
-rw-r--r-- 1 grog lemis 2522 Jan 11 14:05 patch-src__subgen-image.c.orig
=== grog@defake (/dev/pts/1) /usr/ports/multimedia/dvdauthor/files 3 -> cvs update -Pd cvs update: Updating .
cvs update: conflict: patch-src__subgen-image.c is modified but no longer in the repository
C patch-src__subgen-image.c
What's that doing there? The whole idea of the d option is to remove things. This
is clearly because it was modified. But how? Did I do that?
Nope, wasn't me. And it was a long time ago. What went wrong there? Anyway, removing the
files solved the problem. And, apart from further entries missing in my portrules
file, that was the only problem. Certainly easier than portmaster so far.
A few months ago we bought a hamburger press
on eBay and made a few home-made hamburgers.
The problem with the press was that the diameter was too small, and the hamburgers shank
further on grilling, so we were left with a thick piece of meat which didn't fill the bun.
Recently Yvonne found a hamburger press at the local
supermarket—at $2.50, only 10% of the price I paid on eBay, so she bought one, and today we
tried it out.
And? It works, and the size is better. Of course, we also took the opportunity to try a
different composition for the hamburgers.
There's plenty of work waiting in the garden—I don't think that will ever change. But today
I just couldn't face it, though some things are getting urgent, such as the vegetable patch,
where the potatoes are waiting for harvest before the rats eat them all.
In this morning to discover that my portrules file had been overwritten again with
the last checked-in version. Do I have some hidden automatic checkout somewhere? I don't
know where it would be. Fortunately I had a backup of the last version, so I was able to
continue, but somehow this is all just too much pain, so the next time it stopped, I left it
until tomorrow.
My Internet connection has been surprisingly reliable in the last few months. If I had
known about it 4 years ago, I would have set it up then, but the threshold of installing new
hardware for an uncertain outcome, combined with the terrible service I got from Telstra ensured that I
didn't try it until I had confirmation from others that it was a viable alternative.
I've been keeping a statistics page for the link, in line
with older ones for satellite and ADSL. But it doesn't really give much useful information
any more. Many short dropouts, including restarting the PPP process, go unnoticed, and
others are due to issues beyond the ISP's control, such as this horrible Huawei 1762 modem
hanging (solution: pop the modem from the USB slot, replace, wait for one failed connect).
Apart from that, things have been good for months. In the previous 3 months I had only 8
dropouts, some doubtless for other reasons. Today, though, things were worse:
Popped the modem, replaced it, but the system didn't reconnect: the authentication failed.
Several times. Finally it worked, but the throughput was terrible, with the old 2 minute
ping times. Called up Internode support and
spoke to Jesse, who went through the usual stuff, and promised to call back when he had
something. He didn't, but the network came good after a couple of hours. Hopefully this is
only an isolated incident.
Finally got round to looking at the potatoes in the vegie patch, and harvested about a
hundredweight (50 kg). How do you do it correctly? Gardening books don't give you that
kind of information. If you just dig them up with a spade, you run the risk of damaging the
potatoes. If you dig by hand, which is what I do in this surprisingly sandy soil, you run
the risk of missing some. And I clearly missed a lot: I appear to have harvested a number
of potatoes from the previous year, like the one on the right here:
One way or another I managed to harvest a large number of potatoes, some very big, some over
400 g. But the ones I planted late in spring weren't among them: they had hardly grown at
all. Is this because of the time I planted them, or because they were commercial food
potatoes? No idea.
More of our gladioli are blooming.
Some are quite pretty, but there are clear issues with buying bags of mixed bulbs, as the
clash between this one and the
surrounding Watsonias shows:
So Saturday will be my first day with the new panorama
hardware. But I still don't have a really clear way of getting the lighting right for the
verandah panorama. Spent some time today stitching together individual components of the
panorama I took on 24 December 2011, since that seemed the best candidate
of recent weeks.
The following image represents the two that I considered the best: without mouseover it's
comprised of the individual 3 image HDR shots that I have been doing for months now. With
mouseover, it's the single image created by DxO Optics "Pro" with the
“HDR” function:
The DxO version still looks too dark, though it's infinitely better than the original
images. I could increase exposure, of course, but there's only so much I can do before the
highlights get burnt out. Here a comparison with the “+0EV” version (really overexposed by
one stop, and with flash) and then again with the “+1EV” (overexposed by two stops), each
with the DxO version on mouseover:
The image comparisons I made yesterday gave me something to think about. In many ways the
version processed with DxO Optics "Pro" showed promise, but of course it was too bright. That's not
surprising: it was overexposed by one stop and illuminated with flash, also overexposed by
one stop. As part of a sequence to create an HDR image, that's just what the doctor
ordered, but it's not what I need here. The correction is clear, particularly when
comparing the result with the “normally” exposed image without flash (mouseover swaps each
of these images for the other):
On with the ports very much in the background. With the help of locate found an old
copy of the portsrules file that greatly helped fill in the puzzles. Nevertheless
found one port that didn't build:
...
Fl_Print_Dialog2.cxx:1839: error: expected primary-expression before ')' token
Fl_Print_Dialog2.cxx:1839: error: expected `;' before 'w'
*** Error code 1
Stop in /home/ports/graphics/flphoto/work/flphoto-1.3.1.
This time it wasn't a CVS consistency issue. Clearly there's something wrong with the port.
But it builds in the build cluster, so what's the issue? I can't be bothered. I wasn't
even sure what flphoto is
(it's a basic image display program, which I seem never even to have tried out), so I
installed it from the package. But why is this all so painful?
FreeBSD 9.0 was released today (or yesterday in the USA; it wasn't intended to be on Friday the
13th). And for the first time I recall, it had a dedication:
The FreeBSD Project dedicates the FreeBSD 9.0-RELEASE to the memory of Dennis M. Ritchie,
one of the founding fathers of the UNIX™ operating system. It is on the foundation laid by
the work of visionaries like Dennis that software like the FreeBSD operating system came
to be. The fact that his work of so many years ago continues to influence new design
decisions to this very day speaks for the brilliant engineer that he was.
When we had ham for Christmas, I went off trying
to find some indication of how my mother made it decades ago. I found a bare mention in my diary of 45 years
ago. But I got reading: that was an interesting time of my life. I had just left
school, not unhappily. I had been
treated so badly that I developed an aversion to anything English, some of which continues
until today. In the 12 months from late October 1966 I
returned to Malaysia, learnt
(reluctantly) to run printing presses, drove overland
from Tamil Nadu in southern India to
England, hitch-hiked from Paris
to Hamburg, learnt more German
in Lüneburg, and matriculated in the
University of Hamburg.
My writing has never been good, and it was difficult to read the diary. So I started typing
it in, and so far have managed about two months. It's interesting stuff: I remember the big
issues, of course, but I had forgotten many details. There are other documents too, such as
my photo processing records, which help piece things together. And yes, much of my activity
45 years ago was pretty much what I'm doing now—it was just a few decades of having to earn
my living that got in the way of my life.
And the people! Many at school are referenced only by nicknames, others names that I had
long forgotten, such as Nylander (surname—we almost never used first names at school).
And then today I get a mail message from Ladikpo Nylander, from whom I have really not heard
for over 45 years. He was looking for the Baudouy family who used to run Netherton House,
the holiday home which was the closest thing to home for me when I was in England in the
1960s. From recollection, Marcel Baudouy was
a pied-noir who married a Scot, Sonia,
and settled in England after retirement in the early 1960s. They had four children, Gui,
twins Sylvain and Simone (usually called Miranda), and Alain, all with the distinction of
being the first people I met who had dual nationality (French and British). I lost track of
them sometime round 1970, but it seems that Ladi had kept in contact with them until only a
few years ago, but he has now lost contact and thought that I might be able to help.
Clearly I can't. I also can't find the reference to his name again. If it's in my diary, I
don't seem to have typed it in yet, and I can't find it in the photographic records.
As if that weren't enough of a coincidence, I received a “connect request” on LinkedIn from an Andrew Ferguson in Melbourne, saying
that he was a friend. The only Andrew Ferguson I recall is an estate agent in Ballarat, but
this one runs a computer company. I couldn't recall meeting him, so I sent him a message.
He, too, is from my old school, though we only just overlapped: he started at school in my
final full year at school. Still, what a coincidence. By comparison, I have only met one
other schoolfriend, Rupert Stanley, since 1968.
While watching TV, teevee hung. NFS problems. It turned out that dereel had
crashed, apparently because of still more USB disk problems. If only I could get this eSATA
stuff working well!
Today I tried out the new panorama head for the first time. It does what I expected, though
it took me a while to work out how to set things up so that the arrow pointing to the degree
scale was on my side of the camera (simple: adjust the rotating table with the top lock
screw). The arrow is also hidden by the angle select screw when it's set to 45° increments.
There were a couple of other surprises: the rails I bought have proved to be too short to
position the entrance pupil correctly. It's just about enough for the Zuiko Digital ED
9-18mm F4.0-5.6 set at 9 mm if I set the rail at its maximum length and clamp it
offset by about 3 mm:
In addition, the L bracket for the horizontal orientation isn't exactly 90°. Hugin tells me that it's off by about 1.5°,
which, surprisingly, is noticeable in the viewfinder. I can live with that; it doesn't
cause Hugin any problems, but it's one indication of where my original requirement
for rotation around 3 axes would have solved the problem.
The panorama hardware wasn't the only issue. I'm still trying to expose the verandah images
correctly, and, as planned, took single photos today with flash. A couple of things still
managed to go wrong: even normal TTL flash is relatively bright (I set the lens to f/8, and
the darkest corner is about 4 metres away), and it takes a while to recharge fully, though
it's still prepared to fire.
Stitching the verandah panorama was less than good. There's more at work than the changes
I've been making: the ornamental vine growing round the roof makes it very difficult to find
control points, and despite all my efforts I ended up with parts of the roof that didn't
line up:
How did that happen? The exposure was 1/100 s on a tripod, and the lens was set to 9 mm (18
mm full frame equivalent), at which focal length you should be able to get better hand-held
shots at 1/8 s. Is this misbehaviour on the part of the image sensor?
Hugin wasn't able to find any control points, but with a bit of manual work I found
them, and with masking I was able to eliminate most of that image and still get a complete
panorama.
The other thing I changed today was to use brighter colours for the garden photos. I've
been concerned for some time that they looked a little washed out, so I've tried a couple of
settings in DxO Optics
"Pro": “Landscape postcard” and “HDR (artistic)”. They seem to interact, but both
produce poppy colours, and I think this is one place where they fit. Here the view from the
south-eastern verandah a year ago, last week and today:
The difference becomes particularly obvious looking at
the Clematis blooming on the verandah,
to the extreme left of the images. Here last year and today:
That was heartening enough that I went back looking at old photos. The oldest images I
found with taken in Olympus' raw image format were on 21 May 2008, and the first
house photos were on 19 July 2008. And yes, I wasn't happy at the
time. Today I went back and converted all the images for that day with DxO, with good
results. Here some with the new processing; mouseover gives what I got 3½ years ago:
A database error has occurred. Did you forget to run maintenance/update.php after upgrading? See: http://www.mediawiki.org/wiki/Manual:Upgrading#Run_the_update_script
Query: UPDATE `user` SET user_touched = '20120114015610' WHERE user_id = '288471'
Function: User::invalidateCache
Error: 1205 Lock wait timeout exceeded; try restarting transaction (10.0.6.42)
But this one was different. It was from Wikipedia. It wasn't repeatable.
sent 28676608 bytes received 86878 bytes 68079.26 bytes/sec
total size is 108183083 speedup is 3.76
rsync error: some files/attrs were not transferred (see previous errors) (code 23) at main.c(1042) [sender=3.0.7]
*** Error code 23
A repeat was successful. What's that? A network problem?
Despite a lot of effort, I couldn't get yesterday's verandah panorama to stitch properly.
It's the last real issue I have, so today tried a couple of alternatives. One was the
assumption that there wasn't enough overlap between the top row and the zenith, so I raised
the top row. Too much—the two halves didn't overlap enough:
Tried again, this time in sunshine, with more careful overlaps, but after a lot of manual
lining up, still couldn't get the zenith to close. Here two attempts with and without the
zenith:
There are at least two issues here: firstly, the vine over the roof is gradually closing the
gaps, making it difficult for Hugin (and me) to find control points. And secondly, the lighting is still wrong. It's too
bright at the top, and it varies from one image to the left. Here the zenith and one of the
upper row:
No wonder the control point detectors had difficulty. Maybe I should try the HDR version
again, but that has the distinct problem of ghosting. I need to consider the issue of
lighting yet again. The real problem is getting it wide enough.
One thing that the photos brought to light is that the climate on the verandah is changing.
There are other indications: the roses on the south side are no longer blooming, though they
don't look sick. And the petunias in the
hanging baskets are not looking as happy as they once did. Presumably all are not getting
enough light. And the vines are getting unruly. Spent some time adding additional wires
for them to grow along, but it's tiring work, and I didn't get finished.
Also cut down the giant Echium, which was
looking decidedly past its prime. Collected the flower stems, each of which has an
estimated 30 seeds; the total must be well over 1000:
The stems are prickly, so I'll leave them to wilt before extracting the seeds.
Those weren't the only
seeds. The Mirabilis jalapa
plants aren't looking very happy—I suspect the weather is too warm for them—but they, too,
have generated lots of seeds which look as if they'll germinate, so collected them too.
Time for another attempt to propagate plants from seed.
I have finally finished my ports upgrade. Building from scratch worked well, with
ultimately only one port (flphoto) requiring to be fetched as a binary package. And now FreeBSD 9.0 is out. Time to start over again?
In the meantime tried upgrading the build machine. Found a couple of 1
GB SIMMs lying around, so put them in the
machine, and also an nVidia display card that I
had been given recently. I had been told it was PCI, and I didn't really check, but on
examination it's a 16 lane PCIe board. And
it only has a DVI
connector. And I don't have anything I can connect that to. I'll have to mess around yet
more to get it up and running. About the only thing I could confirm is that Hugin crashes when I try to start it with a
display set to dereel. But I think that's a known bug: it doesn't understand
X properly, and only works on local servers.
It's now been a month since I went off
the blood pressure medication, and
I've been measuring my blood pressure twice a day. The results look good: so good that I've
been doubting the accuracy of my blood pressure monitor (which the Americans apparently call
a sphygmomanometer). Sometimes
the indicated diastolic pressures go as low as 40 mmHg. So today I took it in to the
doctor, and we did a simultaneous measurement with my monitor and hers, one on each arm.
Mine showed 146/81 (incidentally the highest systolic value in the month), hers showed
141/82. So there doesn't seem to be anything wrong with the monitor, and the doctor's not
worried about the low diastolic values, so I'll remain off medication.
And that's all I did. I had intended to go shopping as well, but I had forgotten my handbag
(with my wallet), so just back home again.
Today was the first day of linux.conf.au, in
Ballarat. I was involved in the
organization at the beginning, but I dropped out, and I didn't go to the conference. Why
not? Somehow it's difficult to accept, but my conference days are over. And times are
changing. I think what really broke it for me was the discussion of the Code of Conduct. As I said
last year, the necessity for something
like that turns me off. And now I hear it's printed in the conference handbook. The
programme isn't. What does
that say?
It seems I'm not the only one. A number of the regular visitors from the past are not
going, including Rasmus Lerdorf,
who once spent half his life going from one conference to another. We're all getting
older—my first LCA was 11 years ago—but I don't think that's all. Times are changing, and
this is maybe one indication.
The temperatures this year are either too high or too low. Today was in the “too high”
category, a maximum of 36.4°, and I didn't do much in the garden. But the verandah is
becoming urgent, so I strung another couple of wires and tied up some of the vines. More to
be done.
So now I have finally built all my ports. But FreeBSD 9.0 has just come out, so it seems a good idea to install it before cutting
over to 64 bits.
How do you check out a new branch from the
FreeBSD Subversion repository?
I had it written down somewhere, a “brain dump” by Peter Wemm. But where did I put it? I
couldn't find it in my HOWTOs (though it's there now). Went
looking on the FreeBSD web site. Couldn't find anything there either. Should it be in the
handbook?
That's for end users, who don't have access to the Subversion repository. Should it be in
the Developers' handbook? Arguably, but that's mainly technical stuff, and it's not
there either.
Finally Edwin Groothuis told me: http://wiki.freebsd.org/SubversionPrimer. How did he find that? Google.
Somehow that's an indication that something's wrong with the navigation on site.
So I checked out and started building. Then I ran into an old friend:
===> gnu/usr.bin/groff/font/devX100 (obj)
"Makefile", line 6: Could not find /Makefile.sub
make: fatal errors encountered -- cannot continue
*** Error code 1
Despite a lot of searching, I hadn't been able to find the cause of that problem. I had had
a number of repeats, but then it went away, and I haven't had it at all in the year or so
that I've been trying to upgrade to 64 bits. Why has it shown up again now? We had already
discussed the problem of environment pollution in regard to ports, so once again I tried to
build the system with a clean environment. Success.
So it's pretty clear that it's an environment variable (or just barely possibly a function).
But which? Started removing things at random (well, alphabetically), but it's a slow
business, and I didn't get finished. But somewhere in the middle of the alphabet I got past
the problem, so it must be one of the 20 or so environment variables that I removed on that
occasion.
Also tried playing with the new nVidia display
card, but didn't get very far: the system doesn't recognize it. Is that the kernel, the
card or the BIOS? The BIOS settings are a little funny—by default only one display is
enabled, and there's one on the motherboard—so possibly it's that, but since it was a
freebie, it's possible that the card itself is defective. More experiments necessary.
We cut that off and put it in a vase, where it's still flowering happily. Now there's
another one in its place, this time much better matched to the Watsonias:
My network problems still don't seem to be over. Today had a number of dropouts, and even when I was reconnected, I
only got a GPRS connection, which is so slow
that you couldn't tell the difference from being disconnected. But
the RSSI was showing 16, which is 16 to 18
dB better than I normally have. Popped the modem and it reconnected with normal signal
strength and HSPA. It
looks like I had been connected to a tower with only GPRS (there's one to the east
somewhere). Is this the sticky result of a failure on the correct tower, or is it the modem
itself? How can I tell?
Continued this morning looking for the cause of yesterday's build failure. It's obvious
once it's been found:
MAKEFLAGS='-I ..'
That's something I put in my environment about 3 years ago to enable me to have only
one Makefile in the parent directory of my daily photos. And it seems that this
one Makefile didn't reset it, so it went looking for files in the wrong directory.
Remove it and all is well.
Well, almost all:
=== grog@dereel (/dev/pts/4) ~/Photos/20080906 8 -> unset MAKEFLAGS === grog@dereel (/dev/pts/4) ~/Photos/20080906 9 -> make web make: don't know how to make web. Stop
So this kludge was a little too general. How do I Do It Right? There are many ways to fix
things for the photos:
Run make -I .. web
Run make -f ../Makefile web
Create a link to the parent Makefile in the current directory.
Create an alias mp equating to make -f ../Makefile.
None of them seem particularly elegant, though I'm tending to the last. I'm left with a
feeling that I've forgotten some make magic that would do it better, but I can't find
anything obvious in the documentation.
It's been some time since we have suspected that our fridges have been using too much
electricity, so when somebody offered a relatively new, big one on Freecycle today, we jumped at it. Off
to Mount Clear to pick it
up and discovered that it was almost identical to our big Westinghouse.
We were there with the trailer, so we took it anyway; we can compare later which is better.
It's interesting that the model numbers vary: 381 and 423 look very much like the volume in
litres, but in fact the inside of each is exactly the same. Which is older? Possibly the
external trim (about the only difference) gives the clue. Here the one we picked up today
and our old one:
On 6 September 2008 I took a large number of photos in raw format only. It was intended to be a recurrent thing,
like the flowers in garden pages that I
started keeping in 27 June 2010. But the quality of the photos was
less than satisfactory, so today I reprocessed them with DxO Optics "Pro", to
great advantage. I've done a page comparing all the photos, but here are a couple of examples (again, mouseover shows the
original conversion, and without the mouse is today's version):
The protests against SOPA are coming to a
head: a number of prominent sites, in particular English language Wikipedia, went offline at 16:00 local time today (05:00 UTC):
Or did they? The whole thing is implemented in JavaScript. Disable JavaScript and at least
normal lookups are as good as they ever were, though I could imagine that editing pages
would pose a problem. But why did they do it this way? Surely they knew this. My best bet
is that they're leaving a back door for people “in the know” to get in anyway, in the
assumption that people stupid enough to think that SOPA
and PIPA will work will also not be able to
work out how to access the site. Or maybe they're trying to show that you can circumvent
just about anything.
And the effect of the protest? I fear it won't work. The people who don't understand the
web won't miss it as much as those who do. And it's the kind of protest you can't do too
often. I fear that, one way or another, we're in for a rough time until law-makers learn
what the Internet is.
This evening visitors came from Germany: Christiane Keppler and Andreas Hähnle. Yvonne knows Christiane vaguely from one of her horse mailing lists, but
we had never met. All the more surprising to discover how closely our paths have crossed.
Andreas lived 450 m away from us
in Rosbach vor der Höhe
roughly until we moved in; he even visited the previous owners of our house there. Both he
and Christiane went to university
in Frankfurt am Main, only
a few hundred metres from where Yvonne worked
at the Kreditanstalt
für Wiederaufbau. And then they moved
to Nieder-Gemünden, 10 km from where we lived
in Schellnhausen. What amazing
coincidences!
They're now beekeepers in Knüllwald, and they're in Australia
on holiday, but also to see how people make honey here. Interesting people—pity they can't
stay longer.
They were off pretty soon, though, with a couple more appointments with beekeepers
throughout the day. They intended to visit somebody south
of Geelong and then continue to
the Mornington peninsula, so
the obvious way to go was to take the ferry across
the mouth of Port Philip Bay. But it turns out that for a car and two people, that
would cost $69! The alternative around the bay is nearly double the distance, but according
to Google Maps it only takes about 17 minutes longer, so they probably ended up taking that.
Another day when I didn't do very much. Why? In principle I want to do things, but I just
can't get off my backside. Hopefully it'll pass.
A little work in the garden, removing the odd weeds, pruning some creepers (and planting the
prunings; I wonder if anything will come of them), and that was about all.
Last night Piccola didn't come home again, despite
all our attempts to find her. When we got up this morning, she still wasn't there, but the
food we left out for her in the laundry had been eaten. And a little later I found her
again, though after a while in the house she quickly disappeared in the direction of the
garage. Presumably she smelt a rat.
But what do we do about it? We got her in in the evening, but she wasn't happy and wanted
to go out again. I fear we're in for some conflicts.
I'm still playing around with my amd64 system, though soon I should start to cut
over. One of the issues I've had for some time is that newer versions
of Emacs don't respect the settings
that I have been using for years, and use fonts that produce a window far larger than the
screen. Of course I should look for the reason, and it's certainly part of what I need to
do before I can consider my installation complete, but for the time being I've just been
manually resizing the window.
And today that didn't work! It wouldn't let me make it any smaller, though I could have
enlarged it. But why did that happen? People on IRC suspect my environment, and possibly
that's the case; again, when I ran it with a shell with no environment, it worked. But it
still uses this GTK thing, which I don't
really like, and Peter Jeremy provided me with his defaults, which don't use GTK and seem to
work better, and which didn't display the problem, so I'll go with that.
The whole thing makes you wonder about the value of environment variables. I still haven't
worked out a good way to handle the MAKEFLAGS issue. I should probably investigate
the Makefiles and fix it; in the meantime, the idea of an alias has proved to be
workable, but very inconvenient. I keep typing make instead of mp. If I
can't fix the build, it would be easier to remove the MAKEFLAGS before building.
Last week's house photos weren't the best I've ever
taken, for a number of reasons. The big concern, as ever, was the verandah. I had uneven
illumination and great difficulty finding control points among the leaves on the roof.
Today I made a number of changes: I used flash as before, but lowered the intensity by
one EV. I also physically lowered
the camera by about 35 cm, which both gave a larger area of view of the roof and also meant
that I could get more of the floor in the image, since the tripod legs weren't in the way.
And I cheated and hammered some additional nails into the beams of the roof. I already have
a number for hanging pots off, so it wasn't as bad as it sounded.
The results? The exposure was more even—in fact, good enough to not worry about any more,
at least when the sun isn't shining. But again I had too little overlap between the lower
and upper row, though I was able to find some manual control points. And the roof wasn't
much better than before, though again I was able to find enough control points to get a
reasonable panorama. But it took me the best part of an hour. Here the images for the last
three weeks:
Before I continue, I should decide on a representation format. The roof looks different
now, of course, because it's taken from a different perspective. But probably I need one of
these rotating panorama things where you can turn the panorama to look in a particular
direction.
About the only other thing of interest was a minor problem with one of the panoramas:
somehow the pitch changed from one image to the next, by 6.7° over the course of the shots:
This was the only image like that this week, though I saw others last week. Once I get my
new rail I should pay more attention to ensuring that nothing slips.
Most of the day was taken up with photos, of course, but I finally found time to do a little
work in the garden. Managed to set out some weed mat for one of the trees that desperately
needs planting, and even put some mulch on top, but I couldn't manage a second one. I used
the wind as my excuse.
The roses aren't looking happy, with the possible exception of the one I transplanted
this spring. It lost all its leaves,
but it has come back and even has a couple of buds on the way. But others are not
flowering, and I'm wondering if there's too much competition or not enough fertilizer in the
area. Spread fertilizer over the north bed area, including the shade area, and also around
the Ginkgo. I should do much more.
After my photo processing today, started a backup. Or I tried:
=== root@defake (/dev/pts/0) /home/grog 2 -> mount /dev/ada0p1 /photobackups mount: /dev/ada0p1 : Operation not permitted
That's mount's inimitable way of saying “look at /var/log/messages to see
what's up”. As I feared, it said:
Jan 21 18:40:19 defake kernel: WARNING: R/W mount of /photobackup denied. Filesystem is not clean - run fsck
This is an eSATA drive. Did I forget
to umount it before disconnecting it? Started fsck, which failed with
unreadable sectors starting at 128. It also vomited all over the kernel message buffer and
the log file:
Jan 21 18:41:17 defake kernel: siisch1: Timeout on slot 30
Jan 21 18:41:17 defake kernel: siisch1: siis_timeout is 03240000 ss 60000000 rs 60000000 es 00000000 sts 801e0000 serr 002c0000
Jan 21 18:41:17 defake kernel: siisch1: ... waiting for slots 20000000
Jan 21 18:41:25 defake kernel: siisch1: Timeout on slot 29
Jan 21 18:41:25 defake kernel: siisch1: siis_timeout is 03240000 ss 60000000 rs 60000000 es 00000000 sts 801e0000 serr 002c0000
Jan 21 18:41:25 defake kernel: siisch1: port is not ready (timeout 1000ms) status = 001f0000
Jan 21 18:41:25 defake kernel: siisch1: port ready timeout
Jan 21 18:41:25 defake kernel: siisch1: trying full port reset ...
Jan 21 18:42:32 defake kernel: siisch1: Timeout on slot 30
Jan 21 18:42:32 defake kernel: siisch1: siis_timeout is 01040000 ss 40000000 rs 40000000 es 00000000 sts 801e0000 serr 00080000
Jan 21 18:43:02 defake kernel: siisch1: Timeout on slot 30
Jan 21 18:43:02 defake kernel: siisch1: siis_timeout is 01000000 ss 40000000 rs 40000000 es 00000000 sts 801f0000 serr 00080000
Jan 21 18:43:02 defake kernel: (ada0:siisch1:0:0:0): lost device
Jan 21 18:43:32 defake kernel: siisch1: Timeout on slot 30
Jan 21 18:43:32 defake kernel: siisch1: siis_timeout is 00000000 ss 60000000 rs 60000000 es 00000000 sts 801f0000 serr 00000000
Jan 21 18:43:32 defake kernel: siisch1: ... waiting for slots 20000000
Jan 21 18:43:32 defake kernel: siisch1: Timeout on slot 29
Jan 21 18:43:32 defake kernel: siisch1: siis_timeout is 00000000 ss 60000000 rs 60000000 es 00000000 sts 801f0000 serr 00000000
Jan 21 18:43:32 defake kernel: (ada0:siisch1:0:0:0): removing device entry
Jan 21 18:43:47 defake kernel: REDZONE: Buffer overflow detected. 8 bytes corrupted after 0xfffffe0015f24310 (16 bytes allocated).
Jan 21 18:43:47 defake kernel: Allocation backtrace:
Jan 21 18:43:47 defake kernel: #0 0xffffffff809ab0dd at redzone_setup+0x4d
Jan 21 18:43:47 defake kernel: #1 0xffffffff802c138e at camq_init+0x4e
Jan 21 18:43:47 defake kernel: #2 0xffffffff802c13ec at cam_ccbq_init+0x2c
Jan 21 18:43:47 defake kernel: #3 0xffffffff802c404d at xpt_alloc_device+0xdd
Jan 21 18:43:47 defake kernel: #4 0xffffffff802cc2fd at ata_alloc_device+0x1d
Jan 21 18:43:47 defake kernel: #5 0xffffffff802c3b88 at xpt_compile_path+0xa8
Jan 21 18:43:47 defake kernel: #6 0xffffffff802c3da6 at xpt_create_path+0x66
Jan 21 18:43:47 defake kernel: #7 0xffffffff802cbeaa at ata_scan_bus+0x14a
Jan 21 18:43:47 defake kernel: #8 0xffffffff802c7443 at camisr_runqueue+0x63
Jan 21 18:43:47 defake kernel: #9 0xffffffff802c7861 at camisr+0x101
Jan 21 18:43:47 defake kernel: #10 0xffffffff807043f4 at intr_event_execute_handlers+0x104
Jan 21 18:43:47 defake kernel: #11 0xffffffff80705bb4 at ithread_loop+0xa4
Jan 21 18:43:47 defake kernel: #12 0xffffffff80700a8f at fork_exit+0x11f
Jan 21 18:43:47 defake kernel: #13 0xffffffff80a3d78e at fork_trampoline+0xe
Jan 21 18:43:47 defake kernel: Free backtrace:
Jan 21 18:43:47 defake kernel: #0 0xffffffff809ab056 at redzone_check+0x176
Jan 21 18:43:47 defake kernel: #1 0xffffffff8071b2e8 at free+0x38
Jan 21 18:43:47 defake kernel: #2 0xffffffff802c39ae at xpt_release_device+0xbe
Jan 21 18:43:47 defake kernel: #3 0xffffffff802c3a1a at xpt_release_path+0x1a
Jan 21 18:43:47 defake kernel: #4 0xffffffff802ca295 at xptscandone+0x15
Jan 21 18:43:47 defake kernel: #5 0xffffffff802c7443 at camisr_runqueue+0x63
Jan 21 18:43:47 defake kernel: #6 0xffffffff802c7861 at camisr+0x101
Jan 21 18:43:47 defake kernel: #7 0xffffffff807043f4 at intr_event_execute_handlers+0x104
Jan 21 18:43:47 defake kernel: #8 0xffffffff80705bb4 at ithread_loop+0xa4
Jan 21 18:43:47 defake kernel: #9 0xffffffff80700a8f at fork_exit+0x11f
Jan 21 18:43:47 defake kernel: #10 0xffffffff80a3d78e at fork_trampoline+0xe
As the messages said, there was no repeating: the device entry had been removed. So I
powered down the disk, rebooted defake and tried again. Same thing. Were the
sectors really not readable?
=== root@defake (/dev/pts/0) /home/grog 5 -> dd if=/dev/ada0p1 of=/dev/null bs=128k count=100 conv=noerror dd: /dev/ada0p1: Input/output error
1+0 records in
1+0 records out
131072 bytes transferred in 0.011207 secs (11695439 bytes/sec)
dd: /dev/ada0p1: Input/output error
dd: /dev/ada0p1: Input/output error
5+0 records in
5+0 records out
655360 bytes transferred in 0.025838 secs (25364058 bytes/sec)
dd: /dev/ada0p1: Input/output error
^C^C5+0 records in
5+0 records out
655360 bytes transferred in 147.566084 secs (4441 bytes/sec)
=== root@defake (/dev/pts/0) /home/grog 6 ->
So there appear to be two hardware faults on the disk, and after the second it wasn't able
to continue. So much for the increased reliability of eSATA. And this disk had 1.2 TB of
data on it. It's all recoverable, but the time! Copying the data over 100 Mb/s Ethernet
will take about 36 hours. That's for tomorrow, though: it was time for dinner.
I'm moving my monthly flower photo day to the middle of the month, by taking them every 4 weeks until the time has changed.
Today was part of the transition, and by April I should make it.
I've been concerned aboout the roses, for good reason: almost none of them are flowering.
The Rose “Phyllis Bide” is looking happy enough, but the Iceberg roses on the south side of
the verandah are only barely ticking over.
And those are the only roses that are flowering at all—a far cry from last month. Still, my concern for the south
side of the verandah is not completely justified.
The Clematis are flowering like never
before:
A number of plants are looking happier than before, notably
the Pelargonium “Rhodo” and
the Lobelias, which never seemed to
survive where we planted them, though the Lobelias self-seeded vigorously. But now they're
looking relatively happy:
I wonder why I bother with some plants.
The Anchusa capensis are flowering, but
they don't seem to be worth the trouble. They're less interesting than some of the weeds,
like this orange one that I haven't been able to identify:
Still, I'm quite happy with the Leucospermum, which has flowered for some time, and has
already grown considerably.
On the other hand, some flowers that flowered in profusion last year are looking decidedly
less happy this year. Here
the Stachys byzantina last year
and this.
I suspect that part of the problem is that it was very moist last year, while the weather
this year was more normal (today a top of 37.2°).
The Osteospermums are not living up to my expectations of easy-to-maintain floral ground
cover. Only the purple variety are flowering at all, and that not very much:
I think they need replanting every year, and that means that I should limit the area of the
planting.
The Salvias are puzzling. We bought one
of unspecified species a year ago on Australia Day, so I have called
it Salvia diesaustraliae. It has grown like fury, but it hasn't flowered. I'm going
to have to trim it back at the very least, but if we don't get any flowers, it can go.
The Salvia leucantha and
the Salvia officinalis are
both flowering happily, and the Salvia “Phyllis Fancy” is also flowering. But it's a little
disappointing compared to the leucantha:
Some of the ground covers are finally coming int their own, like this one that was planted
only a couple of months ago. And we have some interesting kinds of grass to which I should
pay more attention:
Connected yesterday's defective disk to lagoon, Yvonne's computer today, with a USB connection. Yes, it required fsck. No, no other
problems. So there's clearly something wrong with
the eSATA connection to defake. Is
it the newest version of FreeBSD? Or just the
fact that fsck was required? The backtraces indicate memory allocation failures,
though it looks more like this was a consequence of hardware problems, not the cause of the
reported problems. Probably I should look at the driver in more detail. But not today.
Darah and Carlotta have been at Chris Yeardley's
place for some time now. Now that the weather is getting hot, Yvonne decided that we should bring them back on foot. It's been nearly a year since I last rode her, as the result of the
greasy heel, which still has not completely disappeared. I would have preferred to wait for
cooler weather—it was over 30° by the time we got them here—but Yvonne was insistent.
In fact, it wasn't too bad, and despite not having been ridden at all for such a long time,
Darah behaved perfectly. That's amazing for any horse, let alone a pure-bred Arab. Here
some photos as proof:
Somehow today was a day of little irritating problems. To the clinic Yet Again: it seems
that there's some programme on for free (or heavily subsidized) preventive medical checkups
for people like me, and in to find out more.
From recollection, this was because I am (marginally) diabetic, though it had nothing
to do with the case at hand.
The background was a slight pain in my hand that might be arthritis. I could go to a
physiotherapist and have it looked at, but it would cost (a little) money, and here I could
get it for free.
Things didn't start well: on the way I discovered that the petrol tank was nearly empty, so
called up Yvonne and asked her to wait until I had filled up
before leaving the house. I made it to the petrol station, filled up, and tried to call
Yvonne—and that previous call had been the last on my prepaid card. Bought a new voucher,
but I was already late, so first went to the clinic and then attended to the recharge, by
which time an impatient Yvonne had called me.
The discussion was somewhat pedestrian, a nurse reading a checklist. No talk of
physiotherapist (probably my fault), but ended up with a referral to a podiatrist. And an
hour for all that.
Shopping at Bunnings, then to pick up the glasses I had ordered last month. They had been waiting for me since the end of last month,
but it seems they messed up informing me. By way of apology they gave me an enormous
“Eyewear care kit” with a retail price of $20, which is good of them.
The hot weather continues—a high of 38.7° today, compared to 38.8° yesterday, definitely
air-conditioner weather. So the power
failed, not just briefly, but for 1½ hours, from 17:43 to 19:16—exactly the time we
wanted to eat. We had planned lamb chops and chips, but clearly we couldn't run the
friteuse, so we ended up with “Rösti”
instead
We had planned to grill the chops on the barbecue—and the gas cylinder ran out. That's the
third occurrence of that kind of issue today. Not a big problem, since we could still do
them in the kitchen, but why do things like this happen all at once?
One of the potential causes for a longer power outage is a bushfire. Bushfires happen on
hot days with strong winds. So after a failure like this, it made sense to call up the
bushfire information hotline. No, no bushfires round here—the closest was
in Flowerdale, on
the Broadford Road—where our
family lived 70 years ago. And that was just a grass fire.
But while I was on the phone, I asked for a solution to a question I have had for some time:
which “total” fire ban district am I in? There's no information about other kinds of fire
bans, and very little of any kind on the web site. This page purports to
offer information on fire bans, but instead, it just gives a link to another page, and for
that you need to know your district. The toy map doesn't help much; we could be in the
Central district, or we could be in the South-West district. No information there about how to find out; for that you need to go back to the
previous page, which leads you to the list of fire districts.
And that page lists municipalities by columns, at least as far as the North East district,
after which it truncates it and requires a slider to move the table across the middle of the
screen. No Dereel, of course, just bigger
places.
So Ryan went looking, commenting as he went. He had trouble too, and in the end gave up.
Then he had a bright idea:
what Shire am I in?
That's Golden Plains Shire,
and it seems it all belongs to the Central district. On closer examination of the poorly
formatted columns, yes, it's there. But that's not a municipality, so unless you read all
the columns sequentially, you're not likely to find it.
Sent off a message to the CFA with that suggestion. Based on past performance,
even after the catastrophes of 3 years ago, I don't have much hope. And to add insult to
injury, they ask for “additional information”, in which I am expected to set
an emoticon!
Yvonne had trouble booting her machine this morning. On
asking, she told me “Could not mount /Photos”. That's the photo disk on my machine,
mounted on hers via NFS. And yes, I couldn't
get it to mount. In log file I found:
Jan 24 19:18:33 dereel kernel: WARNING: /Photos was not properly dismounted
Jan 24 21:00:07 dereel mountd[1325]: can't delete exports for /Photos: Device busy
Jan 24 21:00:07 dereel mountd[1325]: could not remount /Photos: Device busy
Jan 24 21:00:07 dereel mountd[1325]: bad exports list line /Photos -alldirs -maproot
Jan 25 07:23:56 dereel mountd[1325]: mount request denied from 192.109.197.134 for /Photos
Jan 25 07:24:56 dereel mountd[1325]: mount request denied from 192.109.197.134 for /Photos
The last two messages repeated every two minutes. But what's that? I haven't seen that
before, and the problem persisted even when I restarted mountd. I had to
first umount/Photos, and then remount it, and then things worked. It's
clearly something to do with the power failure—the first message is from the reboot
following restoration of power—but that's about all I'm sure about. What happened at 21:00?
That's when my nightly cleanup scripts run, but that doesn't help, since they don't back up
that file system. Maybe it was background fsck. Wouldn't it be nice for it to say
when it is finished?
I need Facebook to submit comments to a
government agency? Are they out of their minds?
Well, not completely. I was given a choice: ““Get Satisfaction”, Google Mail or
Twitter. What happened to real email? And
why does a government agency do this sort of thing? Considered sending it from my gmail
account (yes, I do have one, because some broken sites insist; it just forwards to my real
email). And, not surprisingly, I got a message:
I most certainly do not approve; in fact, I disapprove strongly. What business does a
government agency have to ask me to share my personal details with a non-government agency I
have never heard of? And why is there no warning?
Further investigation shows that the previous page had given me the option to send the email
via other means, including normal email. But that's very much the exception, and I would
guess that fewer than 0.1% of all correspondents use it. I'm horrified. I suppose I should
put in a formal complaint, but that would only be sensible if it was handled by somebody who
understood the issues, and it seems amply clear that there are very few such people around
in the government (or even banks, for that matter—they of all institutions should know
better).
Over the last few days, I've been reprocessing old photos with DxO Optics "Pro",
sometimes with spectacular results. But there's a problem: most of the photos I took in the
first year with my Olympus E-510 were taken only
with JPEG format, and some of the DxO
“presets” are intended for raw images only.
Or are they? I tried the “one-shot HDR” preset, supposedly only available for raw images,
and found that it works with JPEG as well. Probably not as well as with raw images, but the
photos of the barbecue four years ago are amazingly much
better:
And that was the series of photos that eventually got me going the way
of HDR images. For the time
being, I think I can accept that I don't need that any more.
The weather's cooler—a little. Today the temperatures were mainly in the high 20s, and I
finally got up enough energy to put down some more weed cloth and plant a couple of trees
that we bought at the end of winter: the
Melaleuca incana “Sea mist” and the Hymenosporum
flavum, the so-called “Australian frangipani”:
In the evening, decided to eat a few mixed nuts. ALDI has quite a good selection at a good price, so we keep a few bags in reserve.
Tonight I wanted to open one. But which? The oldest, of course.
It took me 5 minutes to find the use-by dates. When I did, I needed a torch to read it by
reflection. The date is written in black on dark blue. It's difficult to read by daylight,
but by artificial light it's almost illegible:
That's the best I could do, first with a camera, then with a scanner. The text is there, on
the right above the bar code.
Why do people do this? I get the impression that they really don't want people to know.
Others write them on plastic foil that has to be removed to open the package, in places
outside the label, or in formats that need prior knowledge of the format to decipher,
like 110212. It's high time for some kind of international regulation on the way
use-by dates are written on products.
Australia Day again today, and once
again there was a fair car boot sale at the site of the old General Store.
Last year we found a number of things to buy
there, so off today to see what we could find.
And we did find some things, though possibly not as much as at other times of year. And
once again we were left wondering what some of the plants were. One looks like the
Japanese Iris that we bought three years ago (first image):
The difference is that the Japanese Iris flowers in early spring, and the other
plant—whatever it is—flowers now, in the middle of summer, and is about 3 times the size.
Bought two of them, and also a similar plant, not in bloom, that allegedly has yellow
flowers.
Apart from that, a Bromeliad that
caught Yvonne's eye, another mystery ground cover that should
be a good choice for under the Cypress
trees, and a Gasteria that looks pretty
much the same as the Gasteria carinata var. verrucosa that flowered 18 months ago and then died on us. The seller told me that
I had probably watered it too much. But none of these plants seem to have any roots,
including the new one.
But wait! There's more! Also bought
a Plumbago auriculata and
a Limonium perezii, both with
pretty blue or violet flowers. The Limonium is supposed to be frost-tender, so we'll have
to think of where to put it. There are plenty of areas on the garden that never get frost,
but is that enough?
Back home, discovered that our ground cover is an only slightly different version of a
succulent that we have in profusion.
We established a number of shrubs to plant there, and it occurs to me that this might also
be a good place for the Grapefruit
and Lime trees that we planted to
the north of the house 3½ years ago, which
have never done well, and which are now being engulfed by
the Hebes. Here the photos after planting (August 2008) and now (taken from the other direction due to the
growth):
Started digging some holes for the other shrubs, and discovered a layer of gravel
underneath. It was too hot for that, so gave up Yet Again and planted the deep red
Pelargonium on the west side of the house:
3 Obtain a young grapefruit tree in a container. Grapefruit trees are readily available
from many nurseries.
What does that have to do with transplantation? Nothing. But somehow it seems typical of
ehow.com. Still, there are other hits, nearly
all from forums. The very first one gives no advice at all, just degenerates into a fight between inhabitants of Arizona and
California. The others aren't much better.
The issue here seems to be that answers on forums are treated as being as reliable as each
other. In all probability there's a good document there written by somebody who knows what
he's doing, but I couldn't find one.
Another frustrating day in the garden. I had planned to plant
the Irises (or whatever they are) in front
of the steps of the south-west verandah, where I had planted some of the variegated Japanese
Iris with the similar flowers. But the soil was hard as rock. It's only gradually becoming
apparent to me how little attention I paid to soil condition when I started gardening. No
wonder so many plants look less than happy. So another job that will have to wait for
cooler weather so that I can exert myself a bit.
Instead pottered around a bit, sprayed some weed spray—for once it was hot and not too
windy—and pruned some plants, tidied up the shade area a bit, and potted a couple of pieces
of Crassula falcata that had
broken off. Not much to show for a day's work.
Another round of house photos today, this time in bright sunshine. After last week's
problems stitching the verandah panorama, took more careful aim setting the heights of the
middle layer. The bottom one is clear: it goes down as far as the tripod will let it. I've
improved that angle by lowering the tripod, and now I have a circle with a diameter of about
40 cm at the bottom. The zenith image is also clear: straight up, now a larger area since
I've lowered the camera. But the one in between? I had hoped to be able to have the top
edge of the layer pointing straight up, almost eliminating the need for the zenith. But
that leaves too little overlap with the lower layer, and in some directions that overlap
isn't much use for finding control points. So today I set the lower edge to be horizontal—I
thought. Yes, it was, but only a few degrees lower. Here an image from last week and this
week:
Still, it made the difference, and there were few problems between the layers. But the
automatic control point detectors still couldn't fit the zenith image. Spent half an hour
doing it manually, then somehow lost the image and the PTO file. Somehow I need to find a
better way.
The other issue was the saturation. Last week was cloudy, and I used the “Artistic” “HDR”
“preset” (what a number of dubious terms!) with DxO Optics "Pro". That
came out well. This week it was sunny, and I still need to decide whether the results
aren't too gaudy. Here results for 17 December 2012, before I started
using DxO, then last week, when it was shady, and then today:
In particular the clouds look a little unnatural. But last month's image looked washed out
by contrast—except in the shade. Here I'm clearly reaching the limit of the DxO “HDR”
rendering: the difference in brightness between the light and shade is much larger in
sunshine, and there's just not enough detail in today's image. I'm going to have to think
of something more conventional, such as additional lighting.
Another hot day spent mainly inside, but that meant first doing a lot of watering. My seed
raising attempts last winter were spectacularly unsuccessful, but at least I was able to
raise a couple of morning glories,
which only now are getting round to flowering. They're also climbing at a rate comparable
to Hops. Here the status yesterday and
today:
It's not immediately apparent—I didn't think it had grown so quickly—but it seems that it's
growing at 12 cm per day, or 5 mm per hour, measured in a straight line. Clearly the real
value is slightly faster. In these photos, taken almost exactly 24 hours apart, it's not
quite, that much, only about 9 cm (the wires are 5 cm apart).
Spent some time today thinking about the correct parameters for processing my house photos.
DxO Optics "Pro" (new version out today) has three levels of intensity when processing “HDR”: “Slight”,
“Realistic“, and “Artistic”. I'm beginning to understand the meaning of the term “Preset”:
they're all just basic settings which can be tweaked continuously. But today I decided to
start by comparing the results of the three presets. Here the shadow area with Artistic,
Realistic and Slight respectively. Running the mouse over the
images gives a comparison with the next version, wrapping around. This appears to work best
with “small” images (i.e. one size larger than the default “thumbnails”).