Greg
animals language
brewing multimedia
food and drink music
fiction health
opinion gardening
photography general
technology history
Stones Road house politics
rant
Greg's diary
April 2013
Translate this page
Select day in April 2013:
Su Mo Tu We Th Fr Sa
1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30
Select month:
2012 Sep Oct Nov Dec
2013 Jan Feb Mar Apr
2013 May Jun Jul Aug
Today's diary entry
Diary index
About this diary
Previous month
Next month
Greg's home page
Greg's photos
Network link stats
Greg's other links
Copyright information
    
Groogle

Monday, 1 April 2013 Dereel Images for 1 April 2013
Top of page
next day
last day

FreeBSD to drop support for i386 architecture
Topic: technology, opinion Link here

Mail from Eitan Adler in the FreeBSD mailing lists today: a proposal to drop full support for the 32 bit Intel i386 architecture (the one others call ia32). His reasoning:

Computers are getting faster, but also more memory intensive. I can not find a laptop with less than 4 or 8 GB of RAM. Modern browsers, such as Firefox, require a 64bit architecture and 8GB of RAM. A 32 bit platform is not enough now a days on systems with more than 4 GB of RAM. A 32 bit core now is like 640K of RAM in the 1990s. Even in the embedded world ARM is going 64 bit with ARMv8.

Secondly, the i386 port is unmaintained. Very few developers run it, so it doesn't get the testing it deserves. Almost every user post or bug report I see from a x86 compatible processor is running amd64. When was the last time you booted i386 outside a virtual machine? Often times the build works for amd64 but fails for i386.

Finally, others are dropping support for i386. Windows Server 2008 is 64 bit only, OSX Mountain Lion (10.8) is 64-bit only. Users and downstream vendors no longer care about preserving ancient hardware.

I hope this email is enough to convince you that on this date we should drop support for the i386 architecture for 10.0 to tier 2 and replace it with the ARM architecture as Tier 1.

Its amazing how few people read the exact wording of that last sentence. A lot of serious discussion went on, both for and against. Somehow the world is becoming too serious a place.


Interference causing reception problems?
Topic: multimedia, opinion Link here

I've been monitoring the quality of recorded TV programmes for a couple of years now. There seem to be two separate issues:


Tuesday, 2 April 2013 Dereel Images for 2 April 2013
Top of page
previous day
next day
last day

Faster than light: it works
Topic: technology, opinion Link here

Finally the specifications for Faster-Than-Light Communication have been published. They work, as I reported over 22 years ago.


Too stupid for Facebook
Topic: technology, opinion Link here

I almost never get up on my hind legs and proclaim that I'm a computer expert. I've been using computers for 45 years, and networks and email for over 30, so I suppose I should qualify. Now there's this Facebook thing, with which I really can't identify. But it can't be difficult—after all, every man and his dog uses it. And a number of them have asked after our welfare after the bushfire last week. If I had had any sense, I would have posted a status on Facebook.

Well, it's not too late, so today I did it, along with a link to my diary. And Facebook, being its usual obliging self, offered a choice of photos out of that page to display along with the entry.

Problem is, there are 140 photos on that page, covering the whole month. Facebook gave me a choice of three of them, none of them relevant, and none of them an obvious choice. In particular, they weren't the first three. It's difficult to tell which they are from the sub-postage-stamp size, given that many images are similar. So I told it to remove the photo from the preview. It did. It also suppressed my message. I hate Facebook!

So I reentered the status. How can you possibly write serious texts in Facebook? It won't even allow you to end a paragraph. I had to use It's all text to create the entry. And I managed to misspell the URL not once, but twice, and the entry still has a photo of my verandah in it. Clearly I'm not clever enough to use Facebook.


Nele and Magda visit
Topic: general Link here

Nele Koemle and her mother Magda Delva around for dinner this evening, and even Chris Bahlo was allowed to come. And once again Yvonne served Pavlova for sweets. That must be at least the third time; last time I noted that Nele doesn't like it (well, just the fruit on top, in fact).

A pleasant time had by all, even if our current rearrangement of the lounge room furniture left Chris on the floor with the computers and dogs:


https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130402/big/Chris-2.jpeg
Image title: Chris 2          Dimensions:          3024 x 4032, 2833 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Tuesday, 2 April 2013, thumbnails          All images taken on Tuesday, 2 April 2013, small
Diary entry for Tuesday, 2 April 2013 Complete exposure details

   
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130402/small/Chris-4.jpeg
Image title: Chris 4
Complete exposure details
Dimensions: 600 x 450, 109 kB
Dimensions of original: 4032 x 3024, 2615 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Tuesday, 2 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130402/small/Chris-5.jpeg
Image title: Chris 5
Complete exposure details
Dimensions: 530 x 510, 151 kB
Dimensions of original: 2833 x 2728, 2515 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Tuesday, 2 April 2013:
thumbnails    small images    diary entry


Wednesday, 3 April 2013 Dereel
Top of page
previous day
next day
last day

More house developments
Topic: general Link here

Finally got to talk to James Iles, the Town Planning Consultant (phone 0408 577 880), about subdividing our property in Dereel. It seems that there's little we can do if the Shire doesn't agree, and that VCAT would not accept the case because current legislation doesn't allow for subdivision. The only glimmer of hope on the horizon is that it seems the government is reconsidering the classification of the rural living zone with a view to reducing the required block size to 2 ha—exactly what we want. But it would take at least a year before any such change came into force, and the Shire could still reject it. He suggested that I try the Department of Planning and Community Development, though he doubts they will be able to give me much information. Do I really want to tilt against windmills?

Looked around the properties for sale on the web—no update in a couple of weeks, apart from the fact that Bernie Massey has changed the photos for 5 Inglewood Drive. And then he called me up, only about a minute later. It seems that 3 Inglewood Drive is also for sale, and he gave me first refusal. Off to take a look at the place—it's considerably smaller, but he expects the same price for it. No way. But it has fewer trees, so we could negotiate.


Thursday, 4 April 2013 Dereel
Top of page
previous day
next day
last day

More potato harvesting
Topic: gardening, opinion Link here

Somehow found the energy to harvest more potatoes today. The harvesting alone is more work than seems worth it. And though the potatoes (Kipfler) looked better than last year (hardly any damaged ones) and are difficult to find, they didn't look as good as the ones you can buy, and they didn't taste any better. From now on we only plant vegetables we can't buy at the supermarket.


ABC reception problems: why?
Topic: multimedia, technology, opinion Link here

More investigations of my interference problems (if that's what they are) today. Here's the output of femon -l (log format), starting when tuned to a commercial channel, and continuing with ABC.

2013-04-04 10:19:40 Adapter 0: status SCVYL     signal 197, S/N 218, noise -22
2013-04-04 10:19:42 Adapter 0: status SCVYL     signal 198, S/N 218, noise -21
2013-04-04 10:19:43 Adapter 0: status SCVYL     signal 196, S/N 219, noise -23
2013-04-04 10:19:44 Adapter 0: status SCVYL     signal 197, S/N 217, noise -21
2013-04-04 10:19:45 Adapter 0: status SCVYL     signal 197, S/N 217, noise -21
2013-04-04 10:19:46 Adapter 0: status SCVYL     signal 165, S/N 188, noise -23
2013-04-04 10:19:47 Adapter 0: status SCVYL     signal 167, S/N 190, noise -24, 1234 block errors
2013-04-04 10:19:48 Adapter 0: status SCVYL     signal 165, S/N 191, noise -27, 1234 block errors
2013-04-04 10:19:49 Adapter 0: status SCVYL     signal 168, S/N 193, noise -25, 1234 block errors
2013-04-04 10:19:50 Adapter 0: status SCVYL     signal 167, S/N 193, noise -27, 1234 block errors
2013-04-04 10:19:51 Adapter 0: status SCVYL     signal 167, S/N 192, noise -26, 1234 block errors
2013-04-04 10:19:52 Adapter 0: status SCVYL     signal 161, S/N 189, noise -29, 2444 block errors
2013-04-04 10:19:53 Adapter 0: status SCVYL     signal 164, S/N 191, noise -27, 2444 block errors
2013-04-04 10:19:54 Adapter 0: status SCVYL     signal 165, S/N 191, noise -27, 2444 block errors
2013-04-04 10:19:55 Adapter 0: status SCVYL     signal 168, S/N 195, noise -27, 2444 block errors

No prizes for guessing when the tuner was tuned to ABC. And this is good: there are no uncorrectable errors, so this particular recording was acceptable. It's interesting that the error rate seems to be measured every 5 seconds. It looks very much like the noise isn't the issue: it's the signal. But how do I get ABC to believe this?


Friday, 5 April 2013 Dereel → Ballarat → Dereel
Top of page
previous day
next day
last day

Into town again
Topic: general, opinion Link here

Into town today to pick up a prescription and have my hair cut. While there, also took a look around display homes: we discovered that the total living area (kitchen, dining room, lounge) of our current favourite house was 2 m² larger than the kitchen of our present house. OK, our kitchen is badly laid out and has a lot of waste space, but it was still a bit of a surprise. Did some looking around, and it really seems that that much space is sufficient if it is well used.

In the process, discovered a new development site for display homes, Lucas, in Alfredton. It's clearly very new-only a couple of houses are complete—but one of them, by Simonds, appeared immensely spacious—and it turned out to be only slightly larger than the ones we're looking at. So I'm a little reassured on that topic.

Why is it, that when I ask people in this metric country how big the house is, they say something like “22 squares”? I asked that question at both place I went, and followed up with “How much is that in square metres?“. Neither of them could tell me: they had to look it up.


Saturday, 6 April 2013 Dereel → Geelong → Dereel Images for 6 April 2013
Top of page
previous day
next day
last day

More house inspections
Topic: general, opinion Link here

One of the builders that Bram Gunn recommended to us was Hotondo. We visited their Ballarat office last month, but they don't have a display home in the area. The closest is the other side of Geelong, 90 km away. Off today to take a look. We weren't very impressed by the finish, but it's apparently a franchise, and it seems that a lot depends on the individual builder. In general we get the impression that houses in Geelong are different from houses in Ballarat—for example, most of the houses in Geelong come without floor covering. Did a bit of looking around—another company that we don't know in Ballarat is Burbank, who had a ridiculously large display home with far too many doorways, but who were interesting because they claim “7 star insulation”, whatever that means. At the very least the house had double-glazing. Unfortunately, the ones in our price class don't.

We were going to go and visit another site nearby, but somehow we're completely fed up with looking at houses which all seem to be the same. So we headed back home, stopping only to buy some specials at ALDI and a particularly bad Kabanos instead of lunch. Nearly 200 km with nothing much to show for it.


Sunday, 7 April 2013 Dereel Images for 7 April 2013
Top of page
previous day
next day
last day

Olympus Viewer 3
Topic: photography, technology, opinion Link here

Now that I have the new Microsoft box I can more or less come to terms with photo processing with DxO Optics “Pro”. But there's one thing that doesn't work: it doesn't know the characteristics of my Olympus Zuiko Digital 18-180mm F3.5-6.3 lens, so it can't correct distortion and chromatic aberration. That's a particular problem because that lens probably has the most distortion and CA of all my lenses.

Today I heard that Olympus has released version 3 of Olympus Viewer. I've tried the previous version before and was irritated by the interface, so it seemed a good idea to try it out and see if they have improved it. It probably was, too. But what a horrible interface! They seem to have completely changed it. Here's the startup screen of Viewers 2 and 3:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130407/small/Viewer-2-1.gif
Image title: Viewer 2 1
Dimensions: 610 x 442, 138 kB
Dimensions of original: 1026 x 744, 168 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 7 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130407/small/Viewer-3-1.gif
Image title: Viewer 3 1
Dimensions: 610 x 442, 52 kB
Dimensions of original: 1029 x 746, 66 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 7 April 2013:
thumbnails    small images    diary entry

This is after I found out how to put the menu bar at the top of the Viewer 3 display, but I still haven't found out how to remove old images.. You don't “open” the files: you “import” entire “folders”:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130407/small/Viewer-3-2.gif
Image title: Viewer 3 2
Dimensions: 401 x 507, 52 kB
Dimensions of original: 401 x 507, 52 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 7 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130407/small/Viewer-3-3.gif
Image title: Viewer 3 3
Dimensions: 614 x 439, 132 kB
Dimensions of original: 1029 x 736, 256 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 7 April 2013:
thumbnails    small images    diary entry

It tells me that it doesn't copy them, and indeed I can't find anywhere where it might have copied them, but the import takes a very long time and can't be interrupted. This display shows both the original raw images and the JPEGs converted by DxO. Still, this is just an experiment. How do I process them? “File” maybe? The only likely selection there is “Open file destination folder”, and it is greyed out:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130407/small/Viewer-3-4.gif
Image title: Viewer 3 4
Dimensions: 404 x 413, 56 kB
Dimensions of original: 404 x 413, 56 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 7 April 2013:
thumbnails    small images    diary entry

“Tools” seems like another good choice, but that only offers you useful functions like “Share” and “Print” and “Set as wallpaper...”.

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130407/small/Viewer-3-7.gif
Image title: Viewer 3 7
Dimensions: 316 x 225, 12 kB
Dimensions of original: 316 x 225, 12 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 7 April 2013:
thumbnails    small images    diary entry

The correct answer is “Photos”/“Open Raw Development Window”. You can also get there via the item “Menu” at top right. And then there are selections with the same options as Viewer 2 has, but with different tab names. “RAW 2” is the interesting one:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130407/small/Viewer-3-6.gif
Image title: Viewer 3 6
Dimensions: 609 x 443, 160 kB
Dimensions of original: 1024 x 745, 336 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 7 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130407/small/Viewer-3-6-detail.gif
Image title: Viewer 3 6 detail
Dimensions: 290 x 364, 10 kB
Dimensions of original: 290 x 364, 10 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 7 April 2013:
thumbnails    small images    diary entry

So Viewer does claim to know about the distortion characteristics of the lens, but not the CA characteristics. That puts it at a disadvantage to DxO for combinations that DxO supports.

But those aren't the photos I wanted to process; they just got added because they were there. The ones with the 18-180 mm lens were taken yesterday. Here's one of them processed by DxO, Viewer and ufraw respectively. Run the cursor over each image to see a comparison with the next one. Click on the images up to three times to get successively larger views.

The comparison is complicated because Viewer crops the image differently from DxO. The native raw images of the Olympus E-30 are 4096×3084 pixels in size, but most software trims it to the maximum JPEG size that the camera delivers, 4032×3024. ufraw doesn't, which is why I have included it in this comparison. And Viewer does not centre the cropped image: it cuts off more at the top and less at the bottom. Run the cursor over the second and third image to see this: the second image (Viewer without the cursor, ufraw with the cursor) shows that more appears at the top than at the bottom of the image. The third image (ufraw without the cursor, DxO with the cursor) shows that it cuts off equal amounts from the top and the bottom.

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130406/small/Geelong-DxO-1.jpeg
Image title: Geelong DxO 1
Complete exposure details
Dimensions: 600 x 450, 120 kB
Dimensions of original: 4032 x 3024, 2864 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 6 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130406/small/Geelong-Oly-Viewer-3-1.jpeg
Image title: Geelong Oly Viewer 3 1
Complete exposure details
Dimensions: 600 x 450, 112 kB
Dimensions of original: 4032 x 3024, 3200 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 6 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130406/small/Geelong-ufraw-1.jpeg
Image title: Geelong ufraw 1
Complete exposure details
Dimensions: 599 x 451, 108 kB
Dimensions of original: 4096 x 3084, 3408 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 6 April 2013:
thumbnails    small images    diary entry

Out of the box, the Olympus output is brighter and appears sharper. I'm sure that DxO can be made to give the same appearance, but it's not the standard. But the corrections I'm looking for, for distortion and chromatic aberration, don't seem to have been done at all.

Chromatic aberration is another matter: it seems that DxO does do some reduction, while Viewer doesn't. Here details from top left of each image. There's clear CA on the pole and the edge of the house, but DxO seems to handle it best:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130406/small/Geelong-DxO-1-detail.jpeg
Image title: Geelong DxO 1 detail
Complete exposure details
Dimensions: 450 x 600, 62 kB
Dimensions of original: 450 x 600, 60 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 6 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130406/small/Geelong-Oly-Viewer-3-1-detail.jpeg
Image title: Geelong Oly Viewer 3 1 detail
Complete exposure details
Dimensions: 450 x 600, 56 kB
Dimensions of original: 450 x 600, 56 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 6 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130406/small/Geelong-ufraw-1-detail.jpeg
Image title: Geelong ufraw 1 detail
Complete exposure details
Dimensions: 450 x 600, 42 kB
Dimensions of original: 450 x 600, 42 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 6 April 2013:
thumbnails    small images    diary entry

I thought that the lack of distortion correction may be specific to this particular lens (Zuiko 18-180 mm), so I tried again with a different image, this time taken with other lenses: Zuiko Digital ED 9-18mm F4.0-5.6 and Zuiko Digital ED 70-300mm F4.0-5.6. Again, Viewer claims to be able to correct distortion, but not chromatic aberration. Here the results:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130211/small/Office-during-1-DxO.jpeg
Image title: Office during 1 DxO
Complete exposure details
Dimensions: 450 x 600, 56 kB
Dimensions of original: 3024 x 4032, 912 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Monday, 11 February 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130303/small/Cockatoo-1-DxO.jpeg
Image title: Cockatoo 1 DxO
Complete exposure details
Dimensions: 600 x 450, 79 kB
Dimensions of original: 4032 x 3024, 1770 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 3 March 2013:
thumbnails    small images    diary entry

It's interesting that the results for the second image look better from DxO than from Viewer. But it seems that Viewer does not perform any distortion correction at all. What use is it? In any case, I've written enough that I've split it out as a separate review.

It seems that Viewer can correct for distortion, as I discovered some months later. Presumably I had forgotten to click on some non-obvious button in the horrible interface.


Monday, 8 April 2013 Dereel
Top of page
previous day
next day
last day

More laziness
Topic: general Link here

Somehow I haven't been doing anything in the last few days. There's plenty to do, but I just can't be bothered.


Pension application revisited
Topic: general Link here

It's been over 6 months since I filled out an application form for my German pension. I filled out the form with great difficulty, saved it and did some thinking about what documents I could dig out. Such was my distaste that I didn't do anything more until a couple of weeks ago, when I tried to access the document again, but it didn't accept my password. So I sent an email to the Deutsche Rentenversicherung and got a long reply telling me all about the background I already knew. I thought that they had not read the message, but after 2 pages of text I found the statement:

Wie Sie an die Daten des Online-Antrages ohne Ihr Passwort gelangen, da können wir leider nicht weiterhelfen.

(We can't help you recover your document without your password).

Isn't that helpful? But it had to be done, so today I started again from scratch. Surprise, surprise! It was much easier. I now know that „Beitrittsgebiet“ means “ex-DDR”, but other questions just didn't get asked, like caring for children and evidence that I was legally in Germany. Even the list of documents they wanted was much shorter. In the end I added a question asking them which documents they wanted; we'll see how that progresses.


Tuesday, 9 April 2013 Dereel Images for 9 April 2013
Top of page
previous day
next day
last day

“Coalition” NBN plans
Topic: technology, opinion, multimedia, general Link here

The Australian Federal Opposition parties (“Liberal” (in my mind really conservative) and National, who have been in a coalition so long that they're generally referred to simply as “The Coalition”) look increasingly likely to win the next Federal election on 13 September 2013, despite their leader. While I don't have much love for the incumbent Labor Party, they did one thing right by initiating the National Broadband Network. The Coalition consider it unnecessary and had previously stated that they would cancel it if they came to power.

It seems that they have changed their tune, and today they announced their new plans. The NBN offers people in built-up areas glass fibre connections to the premises (FTTH), with current speeds of up to 100 Mb/s, and clearly that's not a hard limit. The Coalition has made it known that this is too expensive, so I was very interested to hear what they had to say. No information on the TV programme, of course, and so I just recorded the midday news. That proved not to be the midday news, but about the last 60% of the presentation.

Callum Gibson found the whole thing online in low resolution, so off to the lounge room to look at that. No go. It's Adobe Flash, and I still haven't installed flash on teevee (long story; the installation is out of date, and I really need to upgrade lots of other things first). I was about to go back to the office when Callum reminded me that my TV has an IP connection, so tried that. Still no go; it may be able to display MPEGs and such, but not flash.

Finally got round to looking at the presentation, by Malcolm Turnbull (shadow minister for this sort of thing) and Tony Abbott (leader of the Coalition and thus prime minister in spe). They claim that Australians need faster broadband than “this Government” can deliver. That, it seems, translates to a minimum of 25 Mb/s and normally 50 Mb/s by the end of 2016, and a minimum of 50 Mb/s by the end of 2019. This, it seems, is faster than 100 Mb/s because everybody will have it. They came up with some figures: in 2010 the plan was for 1.3 million connections to the NBN by 30 June 2013. In August last year they revised the figure to 340,000, and last month the figure was stated to be round 200,000. Clearly “this Government” can't deliver!

The other issue is the cost of FTTH. The alternative of FTTN (Fibre To The Node) is, says Turnbull, 75% cheaper. Instead of laying fibre to every house, they'd lay it to the ends of streets and use existing copper for the connection to the premises, using technologies like VDSL.

And the speed? Turnbull says “25 Mb/s should be enough for anybody”, and gave the example (borrowed from British Telecom) that a high-definition TV signal only takes 6 Mb/s, so a 25 Mb/s connection would be enough for 4 of them. I was watching on a 720p recording, which proved to be at slightly over 8 Mb/s. And the quality isn't that good; it's a compromise because of the limited bandwidth of DVB-T. A better 720p MPEG-2 stream would probably take 12 Mb/s. So 1080p would be more like 27 Mb/s, too much for the link. And when do the real high definition images come? Planning is already under way for ultra high definition television with at least 2160p. That would imply another fourfold data rate, coincidentally saturating a 100 Mb/s link. Clearly FTTN and VDSL can only be an interim solution. It's interesting to note that, though they quoted British Telecom repeatedly, they didn't mention that they are rolling out FTTH.

Despite these limitations, all this sounds like a sensible way to go. Rather than tearing out existing infrastructure and starting again, delivering maximum speed to only a few people, let's do what we can for as many people as possible and upgrade from there. It's clear that even 100 Mb/s won't be enough for TV of the future—but the costs of such upgrades could be covered by the gradual decommissioning of TV transmitters.

And Wireless? It wasn't mentioned. Would the Coalition NBN have given FTTN to people in Dereel? I greatly doubt it.


NBN plans: the other side
Topic: technology, opinion, general Link here

We had a lot of discussion of the matter on IRC, of course. And there are lots of reactions. I like this one:

http://i.imgur.com/w3ZsBoC.jpg

More interestingly, though, somebody came up with slides from a presentation that Simon Hackett made. It's fascinating: FTTN would cost money that would be wasted when moving to FTTH, it uses more power, is less reliable, and just changing the infrastructure plans could take years. And the final link, currently VDSL-2, is already stretching technology to its reasonable limits:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130409/small/FTTH.gif
Image title: FTTH
Dimensions: 580 x 465, 60 kB
Dimensions of original: 1165 x 935, 122 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Tuesday, 9 April 2013:
thumbnails    small images    diary entry

Another point that came out of these slides is why the current NBN is so far behind schedule: your friend, Simon's friend and mine, Telstra. It took NBN a long time to come to an agreement with them, which greatly held up the deployment. So now they have an agreement with Telstra, and the Coalition want to renegotiate it. As one of the reporters at the press conference pointed out, Telstra is one of the biggest legal firms in the country. The negotiations could take as long again. Simon's final slide sums it up:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130409/small/FTTN.gif
Image title: FTTN
Dimensions: 487 x 554, 134 kB
Dimensions of original: 844 x 960, 256 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Tuesday, 9 April 2013:
thumbnails    small images    diary entry

Some of the details of this mock-up are worth noting:

That's just the tip of the iceberg, of course. I can see things changing a lot before the election.


Chiles: what's the point?
Topic: gardening, food and drink, opinion Link here

Most of my vegetables have proved to be either too much work or of insufficient quality, and in principle I've decided not to grow anything I can buy in the supermarket. But there are others. This year my chiles poblanos have turned out quite well, apart from a couple that got infested with spiders:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130409/small/Chile-1.jpeg
Image title: Chile 1
Complete exposure details
Dimensions: 600 x 450, 124 kB
Dimensions of original: 4032 x 3024, 1584 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Tuesday, 9 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130409/small/Chile-2.jpeg
Image title: Chile 2
Complete exposure details
Dimensions: 600 x 450, 90 kB
Dimensions of original: 4032 x 3024, 1232 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Tuesday, 9 April 2013:
thumbnails    small images    diary entry

Incidentally this is the same bush from which I got last year's crop: it seems to be a perennial. But what do I do with them? Originally I wanted to dry them and use them as chiles anchos for Mexican food. But there aren't enough for that. Last year I used some for pollo en mole verde, some of which I froze and by chance ate again today. So I chopped up some raw chiles and tried them. They tasted alright, but why bother?

My tortillas also caused grief today. I had lots of difficulties when I first started making them, but things have got a lot better—until today. Once again I had them sticking to the comal. I mixed the mixture some time in advance; maybe it was too moist like that. But it makes me wonder: why use a (relatively rough surfaced) comal when you can use a normal frying pan? To be investigated. In the meantime I'm lowering my quantities of water for the Casa Iberica masa from 190/120 to 180/120.


Wednesday, 10 April 2013 Dereel
Top of page
previous day
next day
last day

Upgrading ports, Yet Again
Topic: technology Link here

Years ago I discovered phpMyEdit, which allows you to edit MySQL tables. It hasn't weathered well: PHP changes have removed the functions it uses, and I have to keep a separate down-rev web server in a VM to be able to use it at all. Clearly I need a replacement.

Recently Anthony Curtis has been showing up on IRC, so I asked him. He didn't know of anything either. But Andy Snow came up with a suggestion to use adminer, which does much more, but also claims to be able to edit tables. So: how about installing it?

The first thing the port did was to try to upgrade PHP. That way madness lies on a production machine. So it looked like time to upgrade ports on my build machine. What fun!

Stale dependency: emacs-24.2,2 --> ImageMagick-6.7.9.4 -- manually run 'pkgdb -F' to fix, or specify -O to force.

So I did that and it started installing other software, which I hadn't expected and thus didn't log. And at the end the error was still there. Never mind, this is a build machine, so I just blew away Emacs and reinstalled it. And then the eternal portupgrade, continually stopping to ask me to configure dependent ports that I had never heard of. Roll on pkgng!


Kepala pertama
Topic: language Link here

Watching a film on TV today, Prime. Yvonne asked what “prime” means. Good question. It's not clear what it means in the title, but it's clear that the word derives from Latin ”primus”, meaning “first”. And both “first” and “primus” derive from the same stem, which includes Sanskrit प्रथमः (prathamaḥ) and Greek πρῶτος (protos). In Germanic languages the initial pr became fur or for.

There are other words like that. Years ago I learnt two everyday Malay words that I later discovered were loan words ultimately from Sanskrit. One was pertama, the meaning of which should be clear. The other was kepala (head), which as a boy I found amusingly close to kelapa (coconut). So “kepala pertama” means “first head”, which Google Translate recognizes as Indonesian (unavoidable) and translates as “head first”.

The Sanskrit word is कपाल (kapāla), which corresponds to Greek κεφαλή (kephale) and Latin caput. Old Greek φ or ph was not pronounced f—that's why the Romans transliterated it rather than using the letter f. It's an aspirated p, [pʰ], much closer to the Latin and Sanskrit pronunciations. It's not clear whether the English word head is related or not; it derives from Old English héafod and a presumed old Germanic *hauƀuđ, which corresponds relatively well with caput, but neither OED nor Grimm accept that as being close enough; Grimm supposes a root *kabh.

Still, it's funny to note that Malay, a completely unrelated language, maintains old Indo-European forms that have even gone out of use in languages such as Hindi and Farsi, and changed so much in English that they're no longer recognizable. It reminds me of the Finnish (also not Indo-European) word for king: kuningas, derived from Old Germanic *kuningo-z.


Thursday, 11 April 2013 Dereel
Top of page
previous day
next day
last day

More opinions on LiberalNBN
Topic: technology, opinion Link here

Not surprisingly, there's been a lot of talk about the “Coalition” plans for the National Broadband Network. I'm surprised how negative they all are. This one is a little extreme, but it gives a good feel for what people are thinking.

And how fast will 25 Mb/s or 50 Mb/s be in 3 or 6 years' time? Did some searching and came up with this document from the OECD, obnoxiously only available in Microsoft “Excel” format. It's 18 months old, and at that time it states that the average advertised download speeds exceeded 50 Mb/s in 8 countries, and they exceeded 25 Mb/s in 24 countries, paradoxically including Australia (with a claimed 34.64 Mb/s). The origin of this document suggests that it's correct in some manner, though the use of the word “average” seems dubious.

This page takes a different approach: the results of Speedtest by country. That's not link speed, of course, and the results are significantly lower. But it's probably more representative of what people really have. No country hits 50 Mb/s, but 17 are faster than 25 Mb/s. Australia comes in 43rd place with 12.6 Mb/s. That probably corresponds to over 25 Mb/s link speed.

One way or another, it's difficult to believe that delivering 25 Mb/s five years after the date of this survey is any form of progress. And at least one person I know, Jari Kirma, has a 100 Mb/s downstream rate now. To add insult to injury, he gets unlimited data for € 8.90 per month. That certainly puts the Coalition at the bottom of my ballot come next election.


Another house possibility?
Topic: general, opinion Link here

Our property search is stalemated at least because we haven't heard back from Bernie Massey. In the meantime, discovered there's another house on the market in Dereel, a relatively good-looking house on the Colac-Ballarat road. Around to take a look at it this morning. The property is “9 acres” (about 3.6 ha), and the back end of it is far enough away from the main road that it wouldn't be a problem. But there are lots of trees on the property; we'd need to remove a number of them, if we're allowed.


More tortilla pain
Topic: food and drink, opinion Link here

After my experiences with tortillas the other day, I tried the things I suggested: less water and in a smooth iron pan. I also left them thicker. And still they stick, though not as badly. I wonder what I'm doing differently.


Friday, 12 April 2013 Dereel
Top of page
previous day
next day
last day

Backing up the Microsoft box
Topic: technology Link here

In the past my approach to Microsoft systems has been very much “hands off”: the thing's there, I don't understand it, so don't try. Backing up is simple like that: boot FreeBSD on the box and copy the disk partition.

But now that the machine is running all the time (usually hibernated), that's no longer appropriate. And surely there's some kind of backup software for it. Yes, indeed, though only because this operating system is “Business”; it seems that “Home” systems don't need backups. I fired it up and was thoroughly confused. It offered to back up either files or my entire computer (what, hardware too?). No, as it says, “Create a Windows Complete PC Backup and Restore image of your entire computer, which can be used to recover from a hardware failure”.

That looked like what I wanted, so clicked on that button and got a window asking me whether I wanted to back up to (“hard”) disk or DVD. Disk, obviously. But I couldn't enter a pathname; it just offered <Refresh>. And off it went looking for backup devices, came back and laconically selected “DVD Backup” for me. Presumably its way of saying “Wrong, fool!”.

So maybe it was clever enough not to back up to the same disk, and not clever enough to back up over SMB? Dragged out a USB disk with what I think is a FAT-32 file system and connected it up. Yes, it saw it. Click on the icon: “Please insert a disk into Removable Disk (F:)”. What does that mean? I didn't find out. Right clicking on the icon and selecting Properties gave me the information: WDC WC800 disk drive.

Was there something wrong with it? Removed it and put in a WDC 1200 instead—but the message didn't change: that was the system disk, conveniently the same kind and displayed under the properties of the USB disk. A good thing I didn't try to format it. My USB drive was just that, “Generic Storage Device USB Device”, and it was working properly. But I couldn't access it. I still can't. Wouldn't it be nice to have more information?

OK, how about the other way round? Mount the Microsoft file system on eureka? Yes, indeed, there's an smbfs, though there's no man of that name; instead, it's mount_smbfs. But on IRC people suggested I should back up from the Microsoft side. How? You can back up to SMB file systems, it seems. So I tried—and it worked! Almost:

You need Full Control permission for the network share. Make the user a co-owner of the share, with Full Control..

What does that mean? No idea. But it seems that this is Microsoft's way of saying “permission denied”. I had created a directory to dump to, as root, but my file system permissions were as myself. A chown solved that problem, and I was able to back up.

But why didn't it work the first time around? It still doesn't, in fact: I was only able to back up the files, not a system image. My best guess is:

Why am I writing all this? For myself, mainly. I'm getting more proficient with Microsoft, but I'm still a rank amateur. Next time I run into the problem I want to have a record of last time's pain.


Saturday, 13 April 2013 Dereel Images for 13 April 2013
Top of page
previous day
next day
last day

Network problems of a different kind
Topic: technology Link here

I've been moaning for months about the terrible quality of my Internet connection, in particular the wireless link—I've seen ping times of over 6 minutes! Lately things have been better there, and I've typically had ping times of under 200 ms, which I consider acceptable (a good ping time for this link is 60-80 ms).

But for the past few days I've been trying to upgrade my FreeBSD ports. The last time was in January, but it seems that since then just about everything has changed, and I'm continually downloading new tarballs and having (dependent) ports that I've never heard of hang, asking me what options I want. I don't know how much longer it's going to take, but clearly I need to keep a tree up to date more often.

And today I had some of the worst throughputs I've seen in a long time. Here's an observation of downloading mutt:

=> Attempting to fetch ftp://ftp.mutt.org/mutt/devel/mutt-1.5.21.tar.gz
-rw-r--r--  1 root  lemis    262,144 13 Apr 12:28 mutt-1.5.21.tar.gz
-rw-r--r--  1 root  lemis    524,288 13 Apr 12:30 mutt-1.5.21.tar.gz
-rw-r--r--  1 root  lemis    602,112 13 Apr 12:31 mutt-1.5.21.tar.gz
-rw-r--r--  1 root  lemis    606,208 13 Apr 12:33 mutt-1.5.21.tar.gz
-rw-r--r--  1 root  lemis    610,304 13 Apr 12:35 mutt-1.5.21.tar.gz
...
-rw-r--r--  1 root  lemis  3,716,886 13 Apr 13:06 mutt-1.5.21.tar.gz

In the 7 minutes between 12:28 and 12:35, I downloaded 348,160 bytes, a speed of 830 bytes per second. Normally I could expect 100 to 200 kB/s. Clearly things got better after 12:25: the remainder came across at 1.67 kB/s. But the ping times were OK: the link wasn't the problem. traceroute showed me the probable cause:

=== grog@eureka (/dev/pts/5) ~/public_html 82 -> traceroute ftp.mutt.org
traceroute to ftp.mutt.org (82.165.34.161), 64 hops max, 52 byte packets
 1  lns20.mel4.on.ii.net (150.101.212.44)  82.454 ms  79.542 ms  80.205 ms
...
 4  gi1-0-0.bdr1.mel4.on.ii.net (150.101.210.134)  98.464 ms  99.633 ms  155.656 ms
 5  ae2.br1.syd7.on.ii.net (150.101.33.28)  414.001 ms  419.149 ms  420.408 ms

In passing it's interesting (and maybe a little disturbing) to see that Internode's network now has the domain on.ii.net. Is that part of the problem? In any case, this big jump in response time between Melbourne and Sydney continued all day, as did my connectivity problems. I ran the tests at speedtest.net and got 0.04 Mb/s downstream, 0.08 Mb/s upstream—that's particularly bad. I can't be the only person to suffer. Why is it lasting so long?


Finally: the Dereel mobile phone tower
Topic: general, technology Link here

Jeffrey Kirsten has saved Dereel! He has erected his own mobile phone tower:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130413/small/Phone-tower.jpeg
Image title: Phone tower
Complete exposure details
Dimensions: 449 x 600, 38 kB
Dimensions of original: 359 x 480, 30 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 13 April 2013:
thumbnails    small images    diary entry

I'm trying to get him to publish another one with him underneath wearing a tinfoil hat.


Danish food and pronunciation
Topic: food and drink, language, opinion, technology Link here

Watching the Danish Food Safari on SBS Television today. Like the Polish Safari a couple of weeks ago, it's interesting because it has things in common with German food. I had a second interest: understanding how the Danes pronounce things. It's different from Swedish, which in the past I've tried to learn, but like the Swedes they tend to drop sounds. Of course, an Australian commentator doesn't help: is the dish gravlax or gravad lax? The presenter used the former, the Danish person doing the preparation the latter—I thought. It seems that in Danish it's either, but the latter is spelt “gravad laks”. I supose it's typical that the SBS recipe is titled “Gavlax (Gravet Laks)”. The former is clearly a typo, and the latter appears to be Norwegian, though the Norwegian (Bokmål) page suggests it's the Danish spelling too.

One other thing they had was Rødgrød med fløde, which they say is difficult to pronounce. I don't know why: ø is pretty much the most common vowel in English, usually the one used to substitute for real vowels in words like “berg” and “burg”. I did hear a Dane pronounce it, but she mumbled so much that it's difficult to tell which letters she omitted. But Google helps: there are lots of hits on pronounce rødgrød med fløde. But the ones I tried, like this one, all mumble too. Maybe the mumbling's part of it.

To my mind rødgrød med fløde looks like German Rote Grütze, and with good reason, as the English page demonstrates. But I had already guessed that, and to confirm it I tried Google translate, which tells me it meant “A long and winding road”.

Huh? I don't speak Danish, but that looks wrong. So I tried translating it into Swedish. “En lång och krokig väg”. That looks right. Translate back to Danish? “En lang og snoet vej”. That too looks plausible. So why am I getting it translated like that? Misspelling, maybe? A little experimentation with the spelling was interesting. “Rødgrød med flød” translates as “Rødgrød with cream“. “Rødegrød med flød” translates as “Red Porridge with cream”. So do “Rødegrød med fløde” and “Rødegrød med flode”. But on checking, the correct spelling is “Rødgrød med fløde”, and that got consistently translated as “A long and winding road”. Google translate can be dubious, but this is so far from any form of truth that I suspected something more basically wrong. And indeed I seem to be right: by the time I wrote this article, it had changed. Rødgrød med fløde now translates as “Fish and Chips”. It looks as if somebody has found a way to corrupt the translations.

This episode is also interesting because it includes a recipe for (full corn, though they don't say this) rye bread. It's really full of all kinds of grain, but it does have some reasonable suggestions.


Let's fake a nadir
Topic: photography, opinion, technology Link here

My spherical panoramas are gradually maturing—except for the nadirs. A couple of months ago I took some panoramas in the Great Otway National Park which would have been almost perfect if I had been able to get a nadir. It's almost impossible to represent a spherical panorama on a flat surface:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130219/small/Triplets-lower.jpeg
Image title: Triplets lower
Complete exposure details
Dimensions: 769 x 351, 168 kB
Dimensions of original: 9090 x 4145, 13772 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Tuesday, 19 February 2013:
thumbnails    small images    diary entry

The flash version looks much better, but the hole at the bottom is unpleasant. But the floor is simply wooden planks, and I have enough images of that. Today tried an experiment: take part of the floor of an adjacent image and replicate it, then manually place control points to put the new image under the tripod:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130413/small/Pano-1.gif
Image title: Pano 1
Dimensions: 556 x 485, 186 kB
Dimensions of original: 1641 x 1432, 1152 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 13 April 2013:
thumbnails    small images    diary entry

There are a number of issues with this approach. There's no way to know exactly where to put the control points, and even a small divergence will distort the image. Secondly, it's very difficult to guess where the thing should be pointing. Still, it's an experiment. Under the circumstances it worked extremely well. To my surprise I covered nearly all the area under the tripod. Of course the planks are pointing in the wrong direction, but that's not immediately obvious. Now I need to do some more thinking about how to refine the method.


Sunday, 14 April 2013 Dereel
Top of page
previous day
next day
last day

More network debugging
Topic: technology, opinion Link here

Yesterday's network problems haven't gone away: my file download speeds remain round 10 kB/s. Finally got round to ringing up Internode Support, where I had to explain to Dan that the problem was networking and not remote file access. Pointed him at the traceroutes that I had done; coincidentally Daniel O'Connor had also done one from Adelaide, which showed a similar step in response time when accessing the Sydney node.

[ur 21:26] ~ >traceroute ftp.mutt.org
traceroute to ftp.mutt.org (82.165.34.161), 64 hops max, 52 byte packets
 1  ns (10.0.2.1)  5.671 ms  2.706 ms  3.740 ms
 2  lns20.adl6.on.ii.net (203.16.215.174)  68.297 ms  70.227 ms  41.383 ms
 3  te3-3.cor1.adl6.on.ii.net (150.101.134.209)  45.237 ms  40.820 ms  54.298 ms
 4  xe-11-0-0.cr1.adl6.on.ii.net (150.101.225.229)  40.998 ms  47.384 ms  69.414 ms
 5  ae4.br1.syd7.on.ii.net (150.101.33.34)  549.616 ms  408.192 ms  357.494 ms
 6  te0-2-0-3.br2.sjc2.on.ii.net (203.16.213.158)  356.865 ms  484.262 ms  358.881 ms

Finally he managed to see what I was saying, and put me on hold while he talked to the technical people. When he came back, he had the answer: it's all BigPond's fault: they're overloading the link, and there's nothing they can do about it.

Somehow I've heard “There's nothing we can do about it” too often from Internode. That must also be the most implausible response I can think of on a link that is clearly within Internode's network. Asked for his supervisor, and got connected to Troy, to whom I needed to explain the problem over again: Dan had told him I was having problems downloading files from mutt.org. In the end he gave me the ticket number 5590659 and said that Vijay would call back within the hour and discuss the problem with me.

He did that, just as I had started lunch, and we went through a number of things. In particular, he pointed at the files at http://mirror.internode.on.net/pub/test/, which I could reach with less than 100 ms. And they, too, came down with the same speed. He contacted Optus, who as usual said that they didn't have any congestion—in this case, I'm inclined to believe them. He then went off to check what was going on at the Point of presence, and discovered very low loads. All in accordance with my observations. Is something artificially throttling the traffic? I tried doing multiple downloads of http://mirror.internode.on.net/pub/test/10meg.test, and watched the usage jump to 60 kB/s for a while, though it then settled down to about 30 kB/s—still much more than I had been seeing with a single transfer. But why should they limit the data rate of individual transfers?

Another thing of interest: a couple of months ago we established that cell 8fc8e4a was on the Rokewood tower. Today there was almost no cell-hopping: I've been on cell 8fc48e8 all the time. And today Vijay tells me that this, too, is in Rokewood. Are Optus telling the truth? It's a pity I didn't log cell IDs before Rokewood came on line.


Another power failure
Topic: general Link here

Another power failure today at 12:35. It was a little unusual: a couple of bounces, and then out. While I was calling Powercor, the power came back on: a total of about 2 minutes down. Told the consultant so, and thought that the matter would be over.

An hour or so later two men from Powercor showed up to fix my power. They were aware of the short outage, but apparently the consultant had called them out anyway. I'm getting the feeling that the average phone consultant has a disconnect with the company they work for.


Cat talk
Topic: animals, language Link here

Piccola is a Siamese cat, and thus very talkative. I've established a number of different sounds that she makes, including a long, loud groan which seems to signify dissatisfaction. Lilac makes the same kind of noise from time to time, but typically she ignores Piccola's utterances.

Today was different: groan from outside the office, and Lilac, on my lap as usual, jumped off to investigate. Piccola had a mouse. That's not unusual in itself, but it was clear that Lilac knew. The groaning noise sounded the same to me, but obviously not to her. I wonder how many different expressions cats are really capable of.


Monday, 15 April 2013 Dereel Images for 15 April 2013
Top of page
previous day
next day
last day

More network troubleshooting
Topic: technology Link here

My network connectivity hadn't improved today. It's clear that my initial suspicion of the link between Melbourne and Sydney was wide of the mark, but one result was that I didn't go back to the beginning and consider the other alternatives. One was packet loss. Under those circumstances you're not filling the pipe, so concurrent transfers, such as the one I tried yesterday afternoon, can improve the total throughput.

Fired up wireshark to take a look. Yes, indeed:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130415/small/packet-loss.gif
Image title: packet loss
Dimensions: 693 x 390, 114 kB
Dimensions of original: 2560 x 1440, 336 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Monday, 15 April 2013:
thumbnails    small images    diary entry


https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130415/small/packet-loss-detail.gif
Image title: packet loss detail          Dimensions:          829 x 326, 104 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Display small version of all images on this page
All images taken on Monday, 15 April 2013, thumbnails          All images taken on Monday, 15 April 2013, small
Diary entry for Monday, 15 April 2013

 

wireshark's highlighting makes it very clear what's going on on the screen, but I find tcpdump easier to understand. This version is done with the -ttt option to show the time between each packet, and the -n option to limit the line length:

00:00:00.247630 IP 63.245.215.56.443 > 192.109.197.192.39051: Flags [.], seq 9856:11264, ack 1, win 18, options [nop,nop,TS val 2750578446 ecr 332327542], length 1408
00:00:00.001056 IP 192.109.197.192.39051 > 63.245.215.56.443: Flags [.], ack 11264, win 1034, options [nop,nop,TS val 332327812 ecr 2750578446], length 0
00:00:00.028909 IP 63.245.215.56.443 > 192.109.197.192.39051: Flags [.], seq 12672:14080, ack 1, win 18, options [nop,nop,TS val 2750578476 ecr 332327563], length 1408
00:00:00.000453 IP 192.109.197.192.39051 > 63.245.215.56.443: Flags [.], ack 11264, win 1034, options [nop,nop,TS val 332327841 ecr 2750578446,nop,nop,sack 1 {12672:14080}], length 0
00:00:00.229688 IP 63.245.215.56.443 > 192.109.197.192.39051: Flags [.], seq 14080:15488, ack 1, win 18, options [nop,nop,TS val 2750578716 ecr 332327812], length 1408
00:00:00.000437 IP 192.109.197.192.39051 > 63.245.215.56.443: Flags [.], ack 11264, win 1034, options [nop,nop,TS val 332328072 ecr 2750578446,nop,nop,sack 1 {12672:15488}], length 0
00:00:00.019564 IP 63.245.215.56.443 > 192.109.197.192.39051: Flags [.], seq 15488:16896, ack 1, win 18, options [nop,nop,TS val 2750578736 ecr 332327841], length 1408
00:00:00.000350 IP 192.109.197.192.39051 > 63.245.215.56.443: Flags [.], ack 11264, win 1034, options [nop,nop,TS val 332328092 ecr 2750578446,nop,nop,sack 1 {12672:16896}], length 0
00:00:00.229849 IP 63.245.215.56.443 > 192.109.197.192.39051: Flags [.], seq 16896:18304, ack 1, win 18, options [nop,nop,TS val 2750578966 ecr 332328072], length 1408
00:00:00.000447 IP 192.109.197.192.39051 > 63.245.215.56.443: Flags [.], ack 11264, win 1034, options [nop,nop,TS val 332328322 ecr 2750578446,nop,nop,sack 1 {12672:18304}], length 0
00:00:00.019140 IP 63.245.215.56.443 > 192.109.197.192.39051: Flags [.], seq 11264:12672, ack 1, win 18, options [nop,nop,TS val 2750578986 ecr 332328092], length 1408
00:00:00.000372 IP 192.109.197.192.39051 > 63.245.215.56.443: Flags [.], ack 18304, win 924, options [nop,nop,TS val 332328332 ecr 2750578986], length 0
00:00:00.000095 IP 192.109.197.192.39051 > 63.245.215.56.443: Flags [.], ack 18304, win 1034, options [nop,nop,TS val 332328332 ecr 2750578986], length 0
00:00:00.239836 IP 63.245.215.56.443 > 192.109.197.192.39051: Flags [.], seq 18304:19712, ack 1, win 18, options [nop,nop,TS val 2750579226 ecr 332328322], length 1408

It's interesting that there's an off-by-one difference between the sequence numbers that reported by wireshark and tcpdump

In this case I had lost the packet after sequence number 11264 or 11265, and no fewer than 4 further packets came in before the missing packet was retransmitted. All got through, of course, but in the process 0.5 s was lost—during which up to 100 kB could have come in.

Vijay had promised to call me in the afternoon, so I deferred until then and did one of the other things he had asked me to do: try it on Microsoft. Did that—once again installation went easily—and was horrified to discover that things worked at full speed! I had download data rates of up to 3 Mb/s:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130415/small/download-1.gif
Image title: download 1
Dimensions: 608 x 444, 78 kB
Dimensions of original: 755 x 551, 38 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Monday, 15 April 2013:
thumbnails    small images    diary entry

How could that happen? It defies all logic. Everything in the problem is pointing to the network, not anything local to me. Could it be the USB bus? I've had lots of trouble with USB, much of which could be due to the FreeBSD implementation, but it seems extremely unlikely. If I were losing packets in that area, I'd get error messages from the drivers, like I did a couple of years ago. It wouldn't appear as such a clear-cut TCP issue. Still, after one extreme surprise it seemed to be worth eliminating, so I changed the USB cable for a shorter one. No improvement. Reboot? That's a Microsoft “solution”. But running the modem on Microsoft seemed to fix the problem too, so I rebooted. Bingo! Things were back to normal.

Wrote a very remorseful message to Internode Support telling them to close the ticket. How I love Internode's system!

Date: Mon, 15 Apr 2013 11:42:36 +0930
From: helpdesk@ticket.internode.com.au
Subject: [ticket.internode.com.au #5593151] AutoReply: Please close ticket 5590659

This message has been automatically generated in response to the
creation of a ticket regarding:
        "Please close ticket 5590659",

There is no need to reply to this message.   You will be responded to by an
Internode representative as soon as possible.   Your ticket has been
assigned an ID of [ticket.internode.com.au #5593151].

After a while, though, things got worse again. It's not clear why, but in the hope that we'll have a Radiation Tower in the not-too-distant future, I'm even prepared to route via Microsoft in the interim. Put the modem back in dxo and tried to route through it. Nothing. OK, maybe it needs something like the FreeBSD net.inet.ip.forwarding sysctl. Did a search and came across this page, which seemed to confirm my suspicions. Set the value (is that a key?) HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters\IpEnableRouter to 1, but it still didn't work. But then, I can't even ping this box, though otherwise it works. There's clearly some strangeness with the Microsoft IP stack that I don't understand.

But when I looked back, eureka was working normally again! So for the moment at any rate I'll leave it as it is.

So what should I learn from this? I suspect that this really was a network issue that was gradually clearing up as I did my tests. It's about the only thing I can think of to explain the situation.


Computational photography
Topic: photography, technology Link here

Carlos “Cartola” Carvalho sent out a message about an online course in computational photography today. It's free, so I took a look. It requires real work, but it could be worth it. But how do you get started? There are lots of online tutorials, but they start by telling you how to install some unidentified tarball^W zip archive that so far I haven't been able to find anywhere on their site.


Tuesday, 16 April 2013 Dereel Images for 16 April 2013
Top of page
previous day
next day
last day

Monitor display scrambled again
Topic: technology, opinion Link here

I've been very happy with my new monitor, but there's one issue that is somewhat disconcerting: when I turn it on in the morning, sometimes the display is scrambled, just a row of random vertical lines. Usually power cycling helps, but on a couple of occasions I needed to do it twice. Today, though, it repeated 4 times. Discovered that switching to a vty solved it, without power cycling. So it looks like a misinterpretation of the incoming data stream.


Long-nosed animals
Topic: animals, opinion Link here

Zhivago has a very long nose. Inside is a very long tongue, and they don't always agree:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130416/small/Zhivago-2.jpeg
Image title: Zhivago 2
Complete exposure details
Dimensions: 600 x 450, 84 kB
Dimensions of original: 4032 x 3024, 1504 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Tuesday, 16 April 2013:
thumbnails    small images    diary entry


Garden flowers in mid-autumn
Topic: gardening Link here

It's the middle of autumn, and I have the impression that the garden still hasn't recovered from the hot summer. But already it seems that autumn is coming, seemingly far too soon:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130416/small/Autumn-leaves.jpeg
Image title: Autumn leaves
Complete exposure details
Dimensions: 450 x 600, 144 kB
Dimensions of original: 3024 x 4032, 2160 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Tuesday, 16 April 2013:
thumbnails    small images    diary entry

More photos here.


Computational photography comprehended
Topic: photography, technology Link here

After some searching, discovered that there was a good reason why I couldn't find yesterday's zip archive for the computational photography course: it doesn't exist. Somehow the layout of this course is not what I expected. Downloaded the archive for the first assignment and took a look at it. The most obvious thing is that I needed software that I didn't have: numpy and cv2. Are they in the Ports Collection? Maybe. Went looking for graphics/numpy and found that Thomas Gellekum had removed it without further comment about 11 years ago. After further investigation discovered that it had been replaced with a port with the intuitive name graphics/p5-numpy, so installed that. And cv2? It seems it's part of graphics/opencv, but I had that installed already. Removed it and reinstalled, but I didn't have time to see whether that solved the problem.


Another house candidate
Topic: general Link here

Another house in Dereel has just been listed. It's a brick veneer house on 2 ha, and probably smaller than what we want. But the price looks right: $295,000, compared to Bernie Massey's $145,000 for a naked block of land without facilities. It's in Snowgum Road, a road that Google Maps still calls “Dereel-Berringa Road”. It's not far from where we live now, so round to take a look at it. At first sight it doesn't look at all bad: I like the block better than the ones in Enfield, and it's well installed, has sheds, bore, fences. The only issue is whether the house is big enough. Yvonne wanted to visit immediately, but of course we had to make an appointment, and tomorrow we're off to Adelaide, so we'll be looking at it on Monday.


Wednesday, 17 April 2013 Dereel → Meadows Images for 17 April 2013
Top of page
previous day
next day
last day

Out of the state again
Topic: general Link here

Until we moved to Dereel six years ago, I had spent my life almost continually travelling. Things have changed. I first left Victoria with my grandmother shortly after Christmas 1951, at the tender age of 3, bound for Adelaide. Today I left Victoria for the first time in nearly 6 years, coincidentally also for Adelaide. I've never been in any one place for such a long time before.

My daughter Yana is graduating from the University of Adelaide tomorrow, so we're off to see the fun. I had forgotten a lot about the traffic on the Western Highway: I had set the wrong parameters in the GPS navigator, and it wanted to take me across a stretch of unknown dirt road, and when I changed that, it was too late for my normal short cut from Skipton to Ararat, and we had to go via Beaufort, where the traffic is pretty bad. And the road works! We must have been through 10 of them, one of which required stopping for several minutes. But even without stopping, each set of road works limits the speed to 80 km/h, or even 40 km/h, for kilometres on end, though the real road works are only 100 m or so long, and in some cases there was no work to be seen at all.

All that doesn't explain why it took us 7 hours to do the 623 km. I had recalled a time of 6½ hours, but I can't see how I could do that in the time. On the other had, we had a record low fuel consumption: 53.43 litres for 655.9 km, or a little over 8.1 l/100 km. I thought that was a record, but it doesn't even come close to the 7.4 l that Yana achieved 10 years ago in the same car.

Arrived at Di Saunders' place at about 16:30, and somehow spent the evening doing little. Our bœuf bourguignon came across very well—we've decided to keep a record of what we serve people after we managed to serve Nele Koemle Pavlova three times in a row, though she doesn't like it.


Entering the digital generation
Topic: technology, opinion Link here

Di Saunders isn't what you'd call a tech-head. She's more interested in horses, and she only uses computers for communication. But she has an ADSL connection, something we can't get, and in recent times she has acquired an iPhone and a Samsung Galaxy Tab 2, the latter for the express purpose of reading eBooks. I've looked at that option 10 months ago and came to the conclusion that it wasn't for me. One of the reasons was the low resolution of the tablet, which didn't seem to be the case for the Samsung, though on comparison the Samsung has only 1024×600, in fact lower than the 1024×768 of the ALDI tablet. The difference must be in the rendering.

Di has tethered the thing to her iPhone, rather than using the ADSL connection, so I went looking for the article. It's so easy with a real keyboard, but once again this horrible keyboard substitute made life hell, and I couldn't work out how to search through web pages, so in the end gave it up.

The moral of the experience? I still don't like tablets. But as an eBook reader this could be an option: it renders text considerably better than the ALDI tablet, and it's also bigger, so it can display more.


Thursday, 18 April 2013 Meadows → Adelaide → Meadows → Adelaide → Meadows Images for 18 April 2013
Top of page
previous day
next day
last day

Tablets: practical example
Topic: technology, opinion Link here

Yana has invited us to dinner at the Red Ochre Grill to celebrate her graduation. That's another place without an address suitable for a GPS navigator. I had been there before (well, same building), but the address “War Memorial Drive” doesn't really help: War Memorial Drive is about 5 km long. How do I tell my GPS that? Enter the coordinates, of course. How do I find them? Under the circumstances they should be on their web page, but they're not. Using Google Maps I can find the place and then press the shift key to display the coordinates (31.974° S, 138.5875° E).

But I didn't have a computer connected to the Internet, just Di's Samsung Tablet. That has access to Google Maps too, but it doesn't have a real shift key, let alone one that I can press while looking at the map. How did they solve that problem? The easy way, it seems: they ignored it. I went through all the help I could get, but I couldn't find a way to get it. In the end I set the street midpoint and gave up. Score 1 for desktops, 0 for tablets.


Wantadilla revisited
Topic: general, opinion Link here

I haven't seen Wantadilla since we left nearly 6 years ago. Di lives just round the corner (that's how we know her), so this morning went by to see what it looks like now. It's barely recognizable:


https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Wantadilla-4.jpeg
Image title: Wantadilla 4          Dimensions:          928 x 291, 170 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Display small version of all images on this page
All images taken on Thursday, 18 April 2013, thumbnails          All images taken on Thursday, 18 April 2013, small
Diary entry for Thursday, 18 April 2013 Complete exposure details

   
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Wantadilla-5.jpeg
Image title: Wantadilla 5
Complete exposure details
Dimensions: 600 x 450, 145 kB
Dimensions of original: 4032 x 3024, 4527 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry

They had wanted to keep the name “Wantadilla”, but to have exclusive rights to the name, going back before their purchase. Google put paid to that, so they chose the name “Chiltern Park”, which I find completely out of place. They have also built the biggest covered riding arena I can recall having ever seen; it must be 54×40 m, or about 2200 m²:

One good idea they had was to cut off the top north-west corner of the property, which had a gate with access to Kuitpo Forest, but which meant that you had to stand outside with horses on the side of busy Battunga Road. Now the gate is further inside, and it's easier to close without horses being frightened off:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Wantadilla-1.jpeg
Image title: Wantadilla 1
Complete exposure details
Dimensions: 450 x 600, 182 kB
Dimensions of original: 3024 x 4032, 5531 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry

The fence posts in the foreground show the location of the old boundary fence.


Yana graduates
Topic: general Link here

Into town this morning primarily for Yana's graduation, but there was plenty else to do. She's currently working in Waymouth St, where there are lots of parking spaces—at unbelievable prices. One was day-long parking for a fixed price of $12.80 (“Only $12”), and next door was a house with prices starting round $5 for an hour. And across the road were 3 hour parking spaces for free, one of which we found.

Good thing too: as soon as we found Yana, she wanted to go off and see some people, but we couldn't find a parking place there at all, so more looking for parking places in Rundle St. $6 for up to 3 hours, which didn't sound too bad. But Yvonne wasn't feeling up to much walking, so we went round the other side of the university, where parking was $9 for the maximum 3 hours. And the parking meter hung validating Yvonne's credit card. So I tried another one, with the result that we ended up with two—$18 for what we could have got for $6 in Rundle St.

As if that wasn't enough, we ended up walking through to Charles St, much closer to the cheaper parking place, and had lunch there:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Charles-St.jpeg
Image title: Charles St
Complete exposure details
Dimensions: 450 x 600, 91 kB
Dimensions of original: 3024 x 4032, 1425 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry

We had a bit of time before the graduation, and there was an exhibition of Turner's paintings on loan from the Tate Gallery (now apparently called Tate Britain to make it clearer what it is) at the Art Gallery of South Australia, so went in to take a look at that—a total of $65 for the three of us! It wouldn't have been so bad if we had had the time to go through it, but as it was, we were through in 20 minutes. Somehow we're bleeding money today.

At the end of the exhibition, Yana learnt how to be a graduate:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Yana-3.jpeg
Image title: Yana 3
Complete exposure details
Dimensions: 450 x 600, 75 kB
Dimensions of original: 3024 x 4032, 2073 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Yana-4.jpeg
Image title: Yana 4
Complete exposure details
Dimensions: 450 x 600, 80 kB
Dimensions of original: 3024 x 4032, 2508 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Yana-7.jpeg
Image title: Yana 7
Complete exposure details
Dimensions: 450 x 600, 82 kB
Dimensions of original: 3024 x 4032, 2673 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Yana-8.jpeg
Image title: Yana 8
Complete exposure details
Dimensions: 450 x 600, 82 kB
Dimensions of original: 3024 x 4032, 2668 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry

The graduation was in Bonython Hall, and lots of pomp and ceremony:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Graduation-2.jpeg
Image title: Graduation 2
Complete exposure details
Dimensions: 600 x 450, 157 kB
Dimensions of original: 4032 x 3024, 3788 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry

https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Graduation-47.jpeg
Image title: Graduation 47          Dimensions:          851 x 318, 157 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Display small version of all images on this page
All images taken on Thursday, 18 April 2013, thumbnails          All images taken on Thursday, 18 April 2013, small
Diary entry for Thursday, 18 April 2013 Complete exposure details

 

Unfortunately, the light wasn't good enough for me to get good photos of Yana receiving her handshake:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Graduation-27.jpeg
Image title: Graduation 27
Complete exposure details
Dimensions: 600 x 450, 132 kB
Dimensions of original: 4032 x 3024, 2460 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Graduation-28.jpeg
Image title: Graduation 28
Complete exposure details
Dimensions: 600 x 450, 134 kB
Dimensions of original: 4032 x 3024, 2528 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Graduation-29.jpeg
Image title: Graduation 29
Complete exposure details
Dimensions: 600 x 450, 134 kB
Dimensions of original: 4032 x 3024, 2382 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Graduation-30.jpeg
Image title: Graduation 30
Complete exposure details
Dimensions: 600 x 450, 133 kB
Dimensions of original: 4032 x 3024, 2513 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry

Outside, the crowd was overpowering. Not for normal people, maybe, but for us:


https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Graduation-48.jpeg
Image title: Graduation 48          Dimensions:          1002 x 270, 173 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Display small version of all images on this page
All images taken on Thursday, 18 April 2013, thumbnails          All images taken on Thursday, 18 April 2013, small
Diary entry for Thursday, 18 April 2013 Complete exposure details

 

Sundance showed up to congratulate Yana. Yana split up with him about a year ago, but they're still friends, and I thought it rather nice of him to come along. I suspect that he was instrumental in convincing her to complete her degree.

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Yana-Sundance-2.jpeg
Image title: Yana Sundance 2
Complete exposure details
Dimensions: 450 x 600, 144 kB
Dimensions of original: 3024 x 4032, 2510 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Greg-Yana-Yvonne-4.jpeg
Image title: Greg Yana Yvonne 4
Complete exposure details
Dimensions: 450 x 600, 134 kB
Dimensions of original: 3024 x 4032, 2291 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry

He didn't stay, though, and we didn't either. There was a reception with lots of people we didn't know, so we decided to go and do something useful instead. In our case it was going shopping at Gaganis, which took a while because I had never been there before, we didn't have an address, and Yvonne couldn't remember where it was (9-13 Bacon St, Hindmarsh).

Bought lots of stuff there, and then back to Meadows to pick up Diane, who was also invited to dinner. That was non-trivial: in Dereel I had got used to being about ½ hour from town, but Meadows is more like 45 minutes—in good traffic. We were in the rush hour on the wrong side of town, and decided that going down Glen Osmond Road in that traffic would be suicide, so diverted by Cross Road, several kilometres longer and probably 10 minutes shorter. Even so it took us 55 minutes. Barely had time to take the long-suffering Zhivago for a walk—he had been locked in all day, and was locked in again—before heading off back to Adelaide, arriving only about 10 minutes late.

Food and ambience were both very good:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Dinner-3.jpeg
Image title: Dinner 3
Complete exposure details
Dimensions: 600 x 450, 121 kB
Dimensions of original: 4032 x 3024, 2588 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Dinner-5.jpeg
Image title: Dinner 5
Complete exposure details
Dimensions: 600 x 450, 115 kB
Dimensions of original: 4032 x 3024, 2434 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Dinner-7.jpeg
Image title: Dinner 7
Complete exposure details
Dimensions: 600 x 450, 83 kB
Dimensions of original: 4032 x 3024, 1487 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Thursday, 18 April 2013:
thumbnails    small images    diary entry

The name “Grill” doesn't really do it justice: they've made some very interesting dishes (not too many) with Bush Tucker, and it works well. The only thing that irritates me is that the flavours are things I've never encountered before. It's certainly interesting enough to investigate further.

Red Ochre is located almost directly on the River Torrens, and the view to the east is spectacular in the daytime. In the evening, it presented an excellent opportunity for a tone-mapped panorama. Here a first cut:


https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130418/small/Torrens.jpeg
Image title: Torrens          Dimensions:          924 x 292, 108 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Display small version of all images on this page
All images taken on Thursday, 18 April 2013, thumbnails          All images taken on Thursday, 18 April 2013, small
Diary entry for Thursday, 18 April 2013 Complete exposure details

 


Friday, 19 April 2013 Meadows → Adelaide → Dereel Images for 19 April 2013
Top of page
previous day
next day
last day

The long drive home
Topic: general, food and drink Link here

Into town this morning to the Central Market for the rest of our purchases. Things have got expensive! I suppose that's normal in 6 years, but paying up to $90 a kilogram for cheese goes against the grain. We'll have to go to the Victoria Market again soon to compare prices. Found some cheese for fondue de fromage and a quantity of fish, then off back home. Another 7 hour drive, though we made better time in South Australia. It would have been even longer if we hadn't taken the short cut from Ararat to Skipton. It saves only 4.2 km over the normal route, but there's no traffic. In the 70 km we saw only 5 other cars, and so made a much better average than we would have done on the Western Highway.

Finally back at 19:10. We drove 1601 km in the last 3 days, and I was quite tired as a result. By comparison, on 9 February 2004 I drove 1407 km and felt refreshed. The difference? Traffic, of course.


Time zones interpreted correctly
Topic: technology Link here

I've complained about my GPS navigator on frequent occasions, but yesterday I noticed something interesting: South Australian time is half an hour behind Victorian time, and it seemed to have adapted automatically. On the way back, I checked. It really did switch time zones exactly at the border:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130419/small/SA-time.jpeg
Image title: SA time
Complete exposure details
Dimensions: 694 x 390, 112 kB
Dimensions of original: 3520 x 1976, 2627 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Friday, 19 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130419/small/VIC-time.jpeg
Image title: VIC time
Complete exposure details
Dimensions: 691 x 391, 106 kB
Dimensions of original: 3847 x 2175, 2393 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Friday, 19 April 2013:
thumbnails    small images    diary entry

This image was somewhat spoilt both by the difficulty of getting a good photo of the navigator in position, and the fact that I had to turn back into the rest area, so the second image shows us pointing in the wrong direction and 12 km further from our destination. In fact they were taken about 1 km and 71 seconds apart. The time is at bottom left on the display and also in the Exif popup; 14:26:51 in SA, 14:58:00 in Victoria.


Observing road works
Topic: general, opinion Link here

Our return journey today was a total of 641 km, of which 516 km were on highways. During this time we had:

Why do they do this? If the remaining road works were a potential danger to traffic, I could understand it. But things on the side of the road are no more dangerous than roads with no verge in the first place. For that, they limit about 2% of the total distance and greatly irritate people.


Saturday, 20 April 2013 Dereel Images for 20 April 2013
Top of page
previous day
next day
last day

Trip report day
Topic: general, photography Link here

In the Good Old Days when I did a lot of travelling for business, the return required a trip report, which typically took up the whole of the next working day. That much hasn't changed. Along with the photos, it really took me all day. Apart from this diary and the photos there's not much to show.


How much fondue?
Topic: food and drink, opinion Link here

In Adelaide yesterday we bought enough cheese for 6 portions of fondue de fromage—I thought. For some reason I was going by 150 g per portion, but looking at the recipe it should have been 200 g. Chris Bahlo came for dinner tonight, and we served a fondue, but decided to see just how hungry would be after the 150 g per person. Result: not at all. We couldn't even find space for the dessert. I'm surprised how much less we're eating—30 years ago we would have eaten 300 g per person (and been correspondingly satiated, of course). Time to update the recipe.


Sunday, 21 April 2013 Dereel Images for 21 April 2013
Top of page
previous day
next day
last day

Another house inspection
Topic: general Link here

Round to Ballarat-Colac road this afternoon to meet Liam Crowley of Blue Ribbon and inspect another house. What a paradox! It's a brick veneer house that looks quite new from the outside, but some of the fittings suggest that it's at least 40 years old. In particular no ensuite bathroom: there's only a single bathroom for the 5 bedrooms, and there's not even a shower cabin. On the other hand, it's very big, with a good separate kitchen and a conservatory that would make a good replacement for our verandah. But it would need a lot of renovation to be useful to us, and Yvonne really doesn't want a two-storey house.

The real issue, though, is the land. First, where is it? Google Maps shows it to be one block down from Swanson Road (which it doesn't identify; the road to the north), but it seems it's the corner block. Comparing with the official site shows that this is an error in Google. There are far too many trees, so the 3.5 ha (“9 acres” in Australian Metric) are only partially usable. And the soil is the same very poor sandy kind that we have in the eastern part of our current property.

Liam came along later to assess the sale value of our property. He was clearly impressed, and looking at it through his eyes makes me wonder why we should want to leave. It certainly puts the other place into perspective: no, thanks.


More computational photography
Topic: photography, technology, opinion Link here

Finally got round to looking at the computational photography course again today, in particular the software needed for it. I had reinstalled graphics/opencv, but it didn't seem to make any difference. In particular, the file cv2.py didn't get installed.

I was wrong, though. Reinstalling opencv did have an effect: it installed a new version of perl, with the result that most of my perl modules disappeared. In particular, spamassassin had died. And it wouldn't reinstall:

REQUIRED module missing: HTML::Parser
REQUIRED module missing: Net::DNS
REQUIRED module missing: NetAddr::IP
optional module missing: Digest::SHA1
...
warning: some functionality may not be available,
please read the above report before continuing!

Can't open Makefile: No such file or directory.

Why doesn't it just install them? That's the point of the Ports Collection, after all. And how do you install them? Once again there's a disconnect between the module and the name of the port. After lots of searching, established the following relationships:

Module       Port       Package name
HTML::Parser       www/p5-HTML-Parser       p5-HTML-Parser-3.69
Net::DNS       dns/p5-Net-DNS       p5-Net-DNS-0.72
NetAddr::IP       net-mgmt/p5-NetAddr-IP       p5-NetAddr-IP-4.066

Why does this have to be so complicated? There should be an easy lookup between the three forms.

Back to Computational Photography, further investigation (RTFM) shows that you have to generate the file cv2.py:

=== grog@eureka (/dev/pts/6) ~/Photography/Computational-Photography/hw0 8 -> python
Python 2.7.3 (default, Jun 13 2012, 14:22:22)
[GCC 4.2.1 20070831 patched [FreeBSD]] on freebsd9
>>> import numpy
>>> import scipy
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ImportError: No module named scipy
>>> import cv2
>>>

That also showed that I needed to install another port, scipy (which the Ports Collection spells science/py-scipy). Started working on that, greatly hindered by a renewed bout of heavy packet loss. Internode's speed test reported 35 kb/s downstream and 80 kb/s upstream, roughly that of an asynchronous modem. Downloading the tarball timed out after about a third, so I loaded it to the external server, where I could use rsync to optimize the download:

=== grog@eureka (/dev/pts/12) /usr/ports/distfiles 5 -> rsync -av www:scipy-0.11.0.tar.gz scipy-0.11.0.tar.gz
scipy-0.11.0.tar.gz

sent 9102 bytes  received 4523240 bytes  4608.38 bytes/sec
=== grog@eureka (/dev/pts/12) /usr/ports/distfiles 29 -> rsync -va w3:swig-1.3.40.tar.gz swig-1.3.40.tar.gz
swig-1.3.40.tar.gz

sent 6612 bytes  received 3694741 bytes  2775.67 bytes/sec

That meant that, once again, I couldn't do anything more with it today.


How many enchiladas?
Topic: food and drink, opinion Link here

We're trying to clear out old food from our deep freeze at the moment, and one of the things we had there was 250 g of enchilada filling. Ended up making a salsa roja to go with them, and made 4 enchiladas. And we only managed one each! Is that a sign that our appetites are dwindling further, that enchiladas are very filling, or both?

The other thing that surprised me was how hot the chilis were. I had considered both Chile pasilla and Chile guajillo to be relatively mild, but the result was quite powerful. I can't recall this from last time.

On the definitely positive side: I've solved the problem with the tortillas sticking. Simply dust a bit of flour on them before cooking.


Monday, 22 April 2013 Dereel Images for 22 April 2013
Top of page
previous day
next day
last day

Another house inspection
Topic: general Link here

Off to Snowgum road today to view another house for sale. It's pretty much what I expected: a very nicely presented property with a house that was too small for our needs:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130422/small/Entrance.jpeg
Image title: Entrance
Complete exposure details
Dimensions: 600 x 450, 98 kB
Dimensions of original: 4032 x 3024, 2160 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Monday, 22 April 2013:
thumbnails    small images    diary entry

I liked the shade area on the south side:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130422/small/Shade-1.jpeg
Image title: Shade 1
Complete exposure details
Dimensions: 600 x 450, 120 kB
Dimensions of original: 4032 x 3024, 1872 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Monday, 22 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130422/small/Shade-2.jpeg
Image title: Shade 2
Complete exposure details
Dimensions: 600 x 450, 112 kB
Dimensions of original: 4032 x 3024, 2368 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Monday, 22 April 2013:
thumbnails    small images    diary entry

But the house has only two bedrooms and a small study that the agent (Jarrod Hodgson) claims can be turned into another bedroom. Not easily. And the house is quite old, once again only one bathroom (no ensuite bathroom). Yvonne was taken by it anyway, and spent much of the afternoon thinking of how to work around the deficiencies.


Ports hell
Topic: technology, opinion Link here

It's been nearly 2 weeks since I started upgrading my ports on my build machine. They were only 3 months out of date, but it took forever, what with slow downloads, ports waiting for configuration input, trips to Adelaide and conflicts. Today, finally, I got a shortlist of still-failed ports:

** Listing the failed packages (-:ignored / *:skipped / !:failed)
        - multimedia/ffmpeg-011 (port deleted)
        - lang/tcl-modules (port deleted)
        * bsdpan-Image-Magick-6.83 (bsdpan-Image-Magick-6.83)
        ! ftp/wget (wget-1.13.4_1)      (unknown build error)
        * lang/tcl85 (tcl-8.5.12_2)
        * x11-toolkits/tk85 (tk-8.5.12)
        * x11-toolkits/py-tkinter (py27-tkinter-2.7.3_3)
        * graphics/py-imaging (py27-imaging-1.1.7_1)
        * multimedia/mlt (mlt-0.8.2_1)
        ! www/chromium (chromium-24.0.1312.56)  (unknown build error)
        ! devel/qt4-assistant (qt4-assistant-4.8.2)     (linker error)
        * devel/qt4-designer (qt4-designer-4.8.2)
        * devel/qt4-linguist (qt4-linguist-4.8.2)
        * x11/kdelibs4 (kdelibs-4.8.4_1)
        * multimedia/xbmc (xbmc-11.0_6)

I could go back to the log files and find why each of these failed, but who cares? This is a build machine. Just remove the flaky ports and reinstall. But that didn't quite work either:

=== root@stable-amd64 (/dev/pts/0) /usr/ports 5 -> pkg_delete /var/db/pkg/ffmpeg-011-0.11.1_1/
pkg_delete: package 'ffmpeg-011-0.11.1_1' is required by these other packages
and may not be deinstalled:
kdenlive-0.9.2_1
mlt-0.8.2_1

Once again things seem to be tied in knots. It's easier just to start again from scratch. I have a list of ports I want to install, so I can do it relatively simply—I hope. In any case, doing this in a virtual machine means that I can go back to the current state again if I want to. Spent the afternoon building a new world for a version dating back to January.


More network hell
Topic: technology, photography, opinion Link here

Returned to the computational photography course today. Started watching a lecture, but the connection was so bad that I couldn't watch it at all. Discovered, though, that there are MPEG-4 versions of the lectures for download, so started that. What a catastrophe! My packet loss rates are as bad as I've seen them, and some of the downloads came over at about 1.5 kB/s, slower than a steam modem. And after an hour, they timed out.

Used the same workaround as yesterday: load the files on my external server and rsync them here. It's rather disappointing to see them transfer to the external server in 4 seconds and then take over an hour to get here.

But what if it's a FreeBSD problem after all? Decided it was worth another try in the Microsoft box, and sure enough, this time it was just as bad there—I couldn't even load Internode's web pages. Trying their test file download gave me an error message that, for Microsoft, wasn't too bad, even if it's a little contradictory:

Error 111 (net::ERR_TUNNEL_CONNECTION_FAILED): Unknown error.

Sent off a new problem report (ticket 5623229) with some traces showing packet loss. And then, suddenly, all was right again. It was so sudden that it couldn't have been a congestion issue, which would tend to change more slowly. What is it? Another strangeness is that it's only a download issue. After processing today's photos, I uploaded them to the external server:

sent 6769692 bytes  received 290 bytes  73188.99 bytes/sec

That's quite a respectable uplink speed.


Tuesday, 23 April 2013 Dereel
Top of page
previous day
next day
last day

Computational photography: finally
Topic: photography, technology, opinion Link here

Despite severe ongoing network problems, I've managed to download enough of the computational photography course to finally do something. It's more difficult than I expected: the course leaves a number of issues to the student, such as learning the software libraries that it uses (notably numpy and opencv), not to mention the python on which the whole thing builds. Somehow it's an island in a large and for me uncharted sea of new software.

Still, once I found my way round the island, it was quite interesting. The first assignment was image manipulation: split an image into its component colours, interlace two half images, and convert an image to greyscale. I'm beginning to understand why so much photographic software uses python: the libraries make it a good tool for this kind of work. Each assignment was only a couple of lines of code. Hopefully by the end of the course I'll have a better overview of the software, and have the advantage of knowing enough python to do other things with it as well.


Wednesday, 24 April 2013 Dereel → Ballarat → Dereel
Top of page
previous day
next day
last day

VCAT hears Radiation Tower complaint
Topic: general, opinion, technology Link here

Into Ballarat this morning to the VCAT hearing of Wendy McClelland's objection to the Dereel radiation tower. They had set aside 3 hours (10:00—13:00) for the hearing. I had left early, and that's a good thing too: it took me over 20 minutes from arriving in front of the Magistrate's court to getting into the correct court room. The parking place across the road is only good for 2 hours, so I had to go to Central Square and park there. And for that I needed coins, which I didn't have, and when I got back to the courts I was sent round the corner: VCAT has its own part of the building. But they weren't there: they were back in the main building.

Got there just before the chairman, John Bennett, came in. There were three groups represented: Golden Plains Shire, represented by Amy Boyd, the McClellands, and NBN, represented by a Miss/Mrs Brennan. They spoke in that order.

Before starting, the chairman referred to the total of five practice day hearings and reminded Wendy that many of the things she had talked about in those hearings were out of order, notably issues relating to radiation and codes of practice. The only issues that VCAT could consider would be visual impact and possibly things like soil erosion.

At this point the woman sitting next to me said to me, “This radiation is a worry, isn't it?”. She wasn't very positively impressed when I told her the whole thing is nonsense. It seems that she and two other women, all apparently considerably older than Wendy, are True Believers.

At this point Wendy asked if all phones were turned off. She brought out some electronic device, apparently a field strength measurement device, and said that they were not. Many phones turned off. Still too much radiation. She said that she would have to leave. The chairman asked if Stewart (her husband) could present the case. No, he could not. A friend of hers then said that she had a smart phone in her pocket, and though it was turned off, it might still emanate radiation. She left the room. Yes, now it was better, but not good enough. And it stayed that way when her friend came back into the room. Stewart went off to get her “tinfoil hat”, and shortly later Wendy left the room. She returned after a while with her tinfoil hat, but without chain-mesh gloves. The getup looks like this (photo from a TV programme in January):

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130123/small/tinfoil-hat.gif
Image title: tinfoil hat
Dimensions: 679 x 398, 184 kB
Dimensions of original: 864 x 506, 224 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Wednesday, 23 January 2013:
thumbnails    small images    diary entry

Finally Amy Boyd came to word, referring to the six grounds for Wendy's appeal, unfortunately without stating what they were. In essence it's the same thing that she said last year: they considered the contradictory requirements of infrastructure and visual impact and struck a balance. The issue of whether satellite or fibre would be a better alternative is not for them to decide, nor are codes of practice, and the Council will not comment on inaccurate and inflammatory statements (presumably a repeat of her claim of corruption in the council). She asked the court to dismiss the appeal.

Next came Wendy, starting at 10:30. She spoke until 12:10, saying the same things over and over again, mainly things that the chairman had told her not to talk about. I'm amazed how patient he was with her. She presented a wad of paper about as big as the one she produced last year, but I couldn't get a copy of any of the documents, and she doesn't seem to have put them on her web site. She started on the right foot by saying that the tower was intended primarily for smart meters, and that decisions were made based on misleading, inaccurate and outdated (“genetic”) information. She produced enough herself: a reference to a “fibre optic wireless satellite”, a reference to some document that the WHO had apparently changed in June 2011, a violation of the zoning rules by taking a 1.1 ha parcel out of an area in the farming zone (which should be at least 100 ha in size). She repeatedly referred to the “enormous amounts of electricity” that the facility would use (my recollection when talking to Optus RF engineers was that a 16 A circuit would be enough, about what an electric heater uses). This one she repeated many times.

Other “issues” were that it would destroy the quality of life in the area, kill tourism, kill the birds that people have desperately tried to lure back to the swamp, somehow affect walking trails, that it was only 400 m from the heart of Dereel, and that they had moved the town sign to ensure that the tower was outside of the town limits. Instead it should be in an industrial zone, and since there are none here it should be in Rokewood, that it would be a bushfire hazard if it gets hit by lightning and all the sparks get out. And that would endanger Enfield State Forest to the north (not the direction that bushfires normally take). Even Wendy's friends found this one a bit far-fetched. In addition it's in a vegetation protection overlay, a claim that raised the chairman's eyebrows. He asked Amy: no, it's not, just the sides of the road.

She even suggested that it was suspicious last month's bushfire was lit suspiciously close to one of her practice day hearings. The chairman said that this was not appropriate and that there was no evidence for any such connection. Of course, maybe Wendy knows more than she was prepared to say.

Other claimed dangers were that once one tower was built, no further approval was necessary to erect further towers, that the radiation limits (0.0041% of ARPANSA limits) were only that low because ARPANSA had increased the acceptable limits, and that other countries, such as Austria, had limits as much as 4000 times lower. She didn't address the fact that the tower would easily fulfil those requirements too; she just later claimed that the limits in Austria were 4000 times less than what the tower would emanate. Given that this was a level of 41 µW/m², this would suggest that the Austrian limits were only 10 nW/m², which is ridiculous. Instead we should be using Ka band satellite, which is much faster than wireless. It's not surprising that she has no idea about latency.

At this point the chairman interrupted and said that this was a matter for RF engineers, but Wendy wanted the opportunity to challenge their viewpoints. She mentioned—again—that idiot John Patterson, the loony who stole a tank and drove around destroying mobile phone towers some years ago, as an expert.

The chairman then interrupted the proceedings and—again—reminded her that the only potential grounds for objection would be visual impact, and that radiation didn't count unless it was close to the limit. 0.0041% did not qualify as close to the limit.

Stewart then asked Wendy to read her conclusions. I don't know if she did or not; they were no more coherent than the rest. We should use Ka band satellite and satellite phones, which are much more reliable (and, which she didn't say, emanate much more radiation from the handsets). Chairman: “That's a question of the NBN philosophy”. The ARPANSA guidelines are out of date because they were set up in 1987, and the first microwave towers were erected in 1997 (in fact, the first microwave towers were TV towers, starting no later than 1980).

Once again she mentioned a claimed proximity to the middle of town. The closest dwellings are now only 300 m from the tower, it seems. This was later refuted, though I appear to have forgotten to note who said it: the closest house is a little over 500 m away, which agrees with my measurements.

At this point, Amy suggested to the Chairman that they skip the next few pages, which covered personal attacks on people in the room; I assume she meant Scott Weston and myself. It seems, though, that somebody else, not present, had made a submission to VCAT asking them to dismiss the case on grounds of insanity. Amy's concern, which the chairman shared, was that Wendy shouldn't be allowed to make that kind of accusation when we would not be allowed to respond. I don't really follow that logic: if we hadn't been there, it seems, there wouldn't have been a problem.

In any case, round about there Wendy's diatribe drew to an end. Mrs/Miss Brennan of the NBN pointed out that she wouldn't have time to call on her three witnesses before 13:00, and suggested adjournment. Fortunately that didn't happen: instead, the chairman agreed to carry on in the afternoon. After a short break she came back with rebuttals of a number of Wendy's statements. To the chairman she pointed out that this issue was to be treated de novo (a term which he asked her to explain, hopefully only for the benefit of the audience), that the land would not be subdivided, that the amount taken from the area barely diminishes agricultural use.

As a guideline, yes, this kind of tower should be built in an industrial zone, but most rural towns don't have industrial zones, so that wouldn't work. The location of the town sign is irrelevant for the choice of location; that's what zoning is for.

It seems that there is a provision for erecting a new tower without a further permit. It's only to replace the existing tower, and can be no more than 20 m from the original. It's clear that two towers that close together would cause significant interference, so it wouldn't make sense anyway. Fibre rollout, something that I personally would have liked to have had, is only for new developments, not for existing ones (a real pity).

Wendy's concerns about the environment and tourism go back to the core of Wendy's complaint, that the ARPANSA regulations are wrong. And that wasn't part of the discussion. Friends next to me clucked indignantly, and one said “People can die”. Wendy's objection relates to the use of the facility, not the development, but VCAT can only address the development. Paragraph 52.19 of the ARPANSA regulations was introduced in 2006, not in 1987 as Wendy claimed. It was updated to refer to wireless broadband applications in 2010.

And then it was time for lunch. I decided not to stay, but Scott Weston did. He published a much shorter account noting that they did, in fact, finish round 16:30, and that one of the experts stated “that the location of the Dereel tower is “one of the best locations for a telecommunications tower” he has ever seen and would consider it to be the perfect example for NBN and other councils to use when looking for further locations in the future.”

The other, less pleasant thing about the meeting is that the decision will take another 6 weeks. We had been hoping for two.


Still more ports pain
Topic: technology, opinion Link here

Continued with my ports build today. I have Makefile targets to fetch all tarballs and configure them, so did that. Configuration (hitting Return most of the time) took 1½ hours. And since so many ports depend on more recent versions of gcc, I decided to build it first. A good thing too: it died on me. Why am I having so much trouble with ports built on a clean install?


Whitebait quantities
Topic: food and drink, opinion Link here

Yvonne bought some Whitebait in Adelaide last week, and we decided to eat it today. We had 500 g, which she thought we would get through in one meal. Not so. We managed about 200 g, maybe less.


Thursday, 25 April 2013 Dereel
Top of page
previous day
next day
last day

VCAT followup
Topic: general, opinion, technology Link here

Spent much of today writing up yesterday's VCAT hearing. One thing in particular interested me: are the limits in Australia so much higher than elsewhere? Wendy claimed that the limits in Austria are 1/4000 of the Australian limits, and later that they were 1/4000 of the actual expected emission of the tower.

It's not easy to find this kind of information. Wikipedia took me round in circles, and even on the ARPANSA web site I had my difficulties. And of course they said nothing about Austria. Finally, though, I found some information:


More blocks of land in Dereel
Topic: general Link here

Another block of land appeared on the estate agent web sites yesterday. It proved to be one that we had seen before. At the time they were asking $150,000 for it, but now they've changed their agent and dropped the price to $125,000. Given what people are asking in Enfield, that's an interesting price for 20 ha. Back to take another look, this time walking all over it. It's on a slope, and at the bottom is a firebreak that was obviously made during last month's bushfire: the fire made it just inside the property, but not across the firebreak. I suppose that could explain why they've gone down in price. In fact, I don't see very much fire danger there, since the property is pretty open, and the prevailing winds come from the opposite direction.

That's still the biggest issue: the prevailing winds. I still don't think it's for us.


Building ports, continued
Topic: technology, opinion Link here

Yesterday my build of gcc 4.9 failed. That's a beta or similar version, I think, so today I tried gcc 4.8. It, too, failed. I had already previously built gcc 4.7, so decided to fall back to that. And it, too, failed!

There's something basically flawed here. This was a fresh install of a virgin system. Why did it fail? Had I somehow managed to make a mess of other ports? One thing I did do wrong was not to make a snapshot of a recent version of FreeBSD STABLE without ports, so went back to do that. Then I can install the ports freshly on something that hopefully doesn't have any inconsistencies. But building a new world took all day.


Internode: we give up
Topic: technology, opinion Link here

My ongoing network issues have received some strange responses from Internode support. Yesterday I received a message telling me to change my settings to limit the frequency range to 900 MHz (in other words, eliminate 2100 MHz). That makes no sense for a number of reasons: there is no 2100 MHz service in this area, I have an antenna that only does 900 MHz, there's no way in the software to limit the frequency to 900 MHz, and it's fairly clear that it's not a connection problem anyway.

Today I got a message from another person, which didn't exactly give me the feeling that he knew what he was talking about:

It seems there is continued correspondence ongoing relating to the latency issues you are experiencing. I have had a look over all the history of this issue and further reviewed this by a support specialists. I have also looked at the tracert you provided for Wireshark and can see the dilemma you are being randomly faced with.

What I'm seeing is packet loss, not latency, as I have told them a couple of times. And under normal circumstances I don't get any packet loss, not even when the ping times go as high as 7 minutes. And what's a “tracert”? Mcrsft's mutilated name for traceroute, which of course has nothing to do with packet traces. If he can confuse output of traceroute with packet traces, is he even in a position to analyse the packet traces?

Apparently not: he continues with “We have exhausted all possible angles of troubleshooting”, and suggests two alternatives: a once-off payment of $80 to make me shut up about the problems, or terminate the service with no early termination fees, which means nothing, since there are none. I refused both, since I have no alternative. But I'm disappointed: once Internode was a company you could rely on for technical excellence.


Afghan Chana Dal
Topic: food and drink, opinion Link here

Watching Afghan Food Safari on SBS TV today. I've commented on its inaccuracies before, but that hasn't stopped them. In the programme somebody shows Maeve O'Meara (the moderator) a bag of split peas and claims that they're Chana Dal. Maeve confirms that they're called split peas in Australia, but doesn't comment on the Indian name (for chick peas). But on the Key Ingredients page they write:

Chana is a small Indian chickpea and dal is the Indian term for dried, split and hulled pulses, which include peas, beans and lentils. Sweet, nutty chana dal is an excellent source of protein and a favourite in soups, stews and curries. Rinse and soak before cooking.

No idea where they get the idea that they're small, but at least they seem to have recognized their mistake—too late, and inadequately corrected. To be fair to them, this mistake appears to be more widespread: at Gaganis last week I saw split peas labeled “Chana Dahl”. I wonder how this error occurs, and also why Afghans use the Indian term.


Friday, 26 April 2013 Dereel Images for 26 April 2013
Top of page
previous day
next day
last day

Still more ports and network pain
Topic: technology Link here

My ports build is still not done. This morning I once again had a build breakage of gcc. Started trying to download the binary package from http://ftp.freebsd.org/, which came across at a snail's pace—it took 8 hours to get it here. And when it did, it wanted a second package, which I didn't have time to download. Once again I noticed that the uplink speed seems to be relatively unaffected by the problems. While downloading gcc at about 3 kB/s, I uploaded some photos:

sent 1062018 bytes  received 144 bytes  28324.32 bytes/sec
total size is 1332763  speedup is 1.25

I'm not sure how rsync calculates the speed; presumably it's total data transferred divided by time. But that's nearly 10 times the downlink speed I was experiencing at the same time. The network quality isn't just terrible, it's getting worse. Today there were no less than 8 disconnects. Here's a summary of reconnections over the last couple of weeks:

=== grog@eureka (/dev/pts/18) ~ 2 -> grep hisaddr /var/log/ppp.log
Apr 13 06:25:39 eureka ppp[3021]: tun0: IPCP: myaddr 118.209.85.56 hisaddr = 10.1.0.1
Apr 13 14:34:59 eureka ppp[3021]: tun0: IPCP: myaddr 118.209.22.252 hisaddr = 10.1.0.1
Apr 14 14:59:18 eureka ppp[3021]: tun0: IPCP: myaddr 118.209.109.64 hisaddr = 10.1.0.1
Apr 15 11:53:25 eureka ppp[3021]: tun0: IPCP: myaddr 118.209.72.209 hisaddr = 10.1.0.1
Apr 15 12:21:18 eureka ppp[3021]: tun0: IPCP: myaddr 118.209.33.206 hisaddr = 10.1.0.1
Apr 15 12:34:42 eureka ppp[2121]: tun0: IPCP: myaddr 121.44.74.228 hisaddr = 10.1.0.1
Apr 15 12:39:54 eureka ppp[2121]: tun0: IPCP: myaddr 118.209.83.136 hisaddr = 10.1.0.1
Apr 15 13:50:38 eureka ppp[2121]: tun0: IPCP: myaddr 118.209.112.248 hisaddr = 10.1.0.1
Apr 15 14:19:53 eureka ppp[2121]: tun0: IPCP: myaddr 121.45.169.81 hisaddr = 10.1.0.1
Apr 18 18:01:10 eureka ppp[2121]: tun0: IPCP: myaddr 118.209.122.198 hisaddr = 10.1.0.1
Apr 20 17:19:48 eureka ppp[2560]: tun0: IPCP: myaddr 121.44.109.238 hisaddr = 10.1.0.1
Apr 21 14:12:37 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.53.121 hisaddr = 10.1.0.1
Apr 21 19:04:06 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.51.150 hisaddr = 10.1.0.1
Apr 22 02:19:30 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.96.4 hisaddr = 10.1.0.1
Apr 22 12:01:54 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.82.201 hisaddr = 10.1.0.1
Apr 22 12:09:53 eureka ppp[2560]: tun0: IPCP: myaddr 121.44.121.48 hisaddr = 10.1.0.1
Apr 22 12:55:44 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.120.159 hisaddr = 10.1.0.1
Apr 22 13:57:45 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.19.158 hisaddr = 10.1.0.1
Apr 22 15:16:46 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.89.241 hisaddr = 10.1.0.1
Apr 22 16:58:45 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.111.38 hisaddr = 10.1.0.1
Apr 22 18:07:49 eureka ppp[2560]: tun0: IPCP: myaddr 121.44.55.97 hisaddr = 10.1.0.1
Apr 23 10:05:04 eureka ppp[2560]: tun0: IPCP: myaddr 121.44.106.64 hisaddr = 10.1.0.1
Apr 23 13:58:24 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.23.250 hisaddr = 10.1.0.1
Apr 23 17:54:57 eureka ppp[2560]: tun0: IPCP: myaddr 121.45.162.103 hisaddr = 10.1.0.1
Apr 24 13:04:02 eureka ppp[2560]: tun0: IPCP: myaddr 121.44.81.198 hisaddr = 10.1.0.1
Apr 24 16:54:54 eureka ppp[2560]: tun0: IPCP: myaddr 121.44.80.124 hisaddr = 10.1.0.1
Apr 25 10:17:24 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.33.28 hisaddr = 10.1.0.1
Apr 25 11:14:40 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.68.141 hisaddr = 10.1.0.1
Apr 25 12:03:22 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.104.60 hisaddr = 10.1.0.1
Apr 25 13:47:59 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.46.134 hisaddr = 10.1.0.1
Apr 25 17:06:19 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.8.191 hisaddr = 10.1.0.1
Apr 26 08:02:59 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.32.219 hisaddr = 10.1.0.1
Apr 26 10:01:24 eureka ppp[2560]: tun0: IPCP: myaddr 118.209.90.162 hisaddr = 10.1.0.1
Apr 26 10:43:04 eureka ppp[18142]: tun0: IPCP: myaddr 118.209.60.11 hisaddr = 10.1.0.1
Apr 26 11:45:36 eureka ppp[18142]: tun0: IPCP: myaddr 118.209.8.121 hisaddr = 10.1.0.1
Apr 26 12:57:50 eureka ppp[18142]: tun0: IPCP: myaddr 118.209.79.247 hisaddr = 10.1.0.1
Apr 26 13:46:36 eureka ppp[18142]: tun0: IPCP: myaddr 118.209.101.82 hisaddr = 10.1.0.1
Apr 26 14:59:31 eureka ppp[18142]: tun0: IPCP: myaddr 118.209.17.177 hisaddr = 10.1.0.1
Apr 26 16:17:27 eureka ppp[18142]: tun0: IPCP: myaddr 121.44.116.179 hisaddr = 10.1.0.1

It's interesting that nearly all of them are during the time 8:00 to 18:00. Put out a query on the Dereel Facebook page and of course got replies that lots of people have network problems. But which? The obvious thing to do is to is to go round with a laptop and monitor things. But for that I need to upgrade the system on my 10-year-old Dell Inspiron 5100 , which was still running FreeBSD 7.1-STABLE, and that doesn't support my modem. Started a rather ambitious attempt to upgrade to 9-STABLE, which failed:

cc  -O2 -pipe  -I. -I/src/FreeBSD/svn/stable/9/lib/libthread_db -std=gnu99 -fstack-protector -Wsystem-headers -Werror -Wall -Wno-format-y2k -W -Wno-unused-parameter -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Wreturn-type -Wcast-qual -Wwrite-strings -Wswitch -Wshadow -Wunused-parameter -Wcast-align -Wchar-subscripts -Winline -Wnested-externs -Wredundant-decls -Wold-style-definition -Wno-pointer-sign -c /src/FreeBSD/svn/stable/9/lib/libthread_db/libthr_db.c -o libthr_db.o
building static thread_db library
ar: fatal: Unrecognized archive format: Inappropriate file type or format
*** Error code 65

Stop in /src/FreeBSD/svn/stable/9/lib/libthread_db.
*** Error code 1

So restarted upgrading via 8-STABLE, which kept the machine busy for the rest of the day.

On the ports build side, continued without the gcc installation. The next surprise was:

Writing Makefile for XML::Parser
Writing MYMETA.yml
==> Your Makefile has been rebuilt. <==
==> Please rerun the make command.  <==
false
*** [Makefile] Error code 1

What kind of nonsense is that? If you rebuild the Makefile, you should then use it, not die. Still, it sounded like a repeat would do the job. It didn't. More assembly needed. In the meantime, though, I have a Makefile target ports-try that just continues with the next port if one dies, so ran that, keeping the system happy for the rest of the day.


First narcissus of autumn
Topic: gardening Link here

I've already noticed that the unseasonal summer weather has confused the plants in the gardens. Today I found a Narcissus, which should flower in spring, not mid-autumn:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130426/small/Daffodil.jpeg
Image title: Daffodil
Complete exposure details
Dimensions: 600 x 450, 120 kB
Dimensions of original: 4032 x 3024, 7197 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Friday, 26 April 2013:
thumbnails    small images    diary entry


Saturday, 27 April 2013 Dereel Images for 27 April 2013
Top of page
previous day
next day
last day

ALDI Networking Gear
Topic: technology, opinion Link here

Into Sebastopol this morning to pick up some of ALDI's weekly specials: 4 TP-Link TL-PA411 powerline Ethernet adapters and two TP-Link TL-WR841N 802.11n wireless routers. The main reason was to replace the Ethernet cable between cvr2 (the TV recording computer) and teevee (the playback computer) that has been lying in the hallway for two years.

Getting the goods was difficult in itself. They weren't with the other specials; instead I had to find then up the front by the cashiers, and even then they wouldn't give them to me! Instead they brought the goods to the cash register when I was about to pay. It seems that a lot of this kind of goods walks, though the boxes are big enough that I can't see how that could be. And of course I was given the wrong goods—laptop power supplies instead of the powerline Ethernet adapters. Fortunately I noticed before I returned home.

Tried the powerline adapters first, in the assumption that they would be easier to install. That was the case, though it's clear that people at TP-Link haven't paid much attention to packaging. They state that it's better to connect them directly to a power point rather than an extension board, since the latter might filter the signal. But how do you plug it into a standard Australian power point?

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130427/small/Adapter-3.jpeg
Image title: Adapter 3
Complete exposure details
Dimensions: 600 x 450, 62 kB
Dimensions of original: 4032 x 3024, 960 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 27 April 2013:
thumbnails    small images    diary entry

And how did it work? Not at all. There are three LEDs supposed to indicate status, but they're so badly shielded from each other that they all appear to be on. And there was no indication of any connection. The one at the cvr2 end was on a power board, so I moved it to the office. Then I got a connection:

=== grog@eureka (/dev/pts/20) ~ 13 -> ping teevee
PING teevee.lemis.com (192.109.197.158): 56 data bytes
64 bytes from 192.109.197.158: icmp_seq=0 ttl=64 time=50.513 ms
64 bytes from 192.109.197.158: icmp_seq=1 ttl=64 time=4.362 ms
64 bytes from 192.109.197.158: icmp_seq=2 ttl=64 time=4.400 ms

That's doesn't exactly suggest the 500 Mb/s that they claim, and the transfer times confirm it:

2032_20130426212700.mpg                                 100% 1922MB   5.8MB/s   05:33

Edwin Groothuis has similar adapters, and he confirms that this is about the speed, roughly a factor 10 slower than advertised. But that's the great thing about ALDI: you can take things back with no questions asked if you're not satisfied. That's also the reason I bought the routers, in the hope of using them as a bridge. They proudly boast: “300 Mbps Wireless N Router”. Jürgen Lock found the specifications and discovered that the switch is 100 Mb/s. So much for those claims. It's only the 802.11n component that could potentially be that fast. It has a typical switch configuration: 4 sockets for “LAN” and one for “WAN”:

Once there was an image here, from http://www.tp-link.com.au/resources/images/products/gallery/TL-WR841ND_8.0-04.jpg, but TPG have seen fit to remove it from their site.

It came with the LAN sockets covered over with a sticker reading “Run CD Before Connecting Cables”. Well, that's for Microsoft, no doubt, so I plugged a cable into the WAN socket and connected it into my network, in the process discovering that I have DHCP running after all, which gave it a local IP address. Tried to talk to it, but it proved to have no ports at all open. How's that supposed to work? Clearly RTFM time.

The Quick Installation Guide confirmed my suspicions that the CD was only for Microsoft—even for Apple, you have to use the alternative installation method: connect a computer to the LAN socket, the WAN to the WAN socket, and access http://tplinklogin.net/. Well, I have a WAN connection anyway, so I don't need to do that. But http://tplinklogin.net/ (IP address 69.43.161.161) redirected to some random textile sales web site:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130427/small/tplinklogin.net.gif
Image title: tplinklogin.net
Dimensions: 565 x 477, 114 kB
Dimensions of original: 1626 x 1373, 464 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 27 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130427/small/tplinklogin.net-detail.gif
Image title: tplinklogin.net detail
Dimensions: 600 x 450, 122 kB
Dimensions of original: 716 x 537, 116 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 27 April 2013:
thumbnails    small images    diary entry

“Viva 1000 Thread Count Cotton Quilt Set in Choice of Three Colours – Queen ($109) or King Size ($119), Includes Nationwide Delivery (Up to $229 Value)”. How did I get there?

OK, I have a Microsoft box, so tried it that way. Once again I had to connect the router between LAN and “WAN” (the rest of my network). And of course I couldn't access it via rdesktop, because the router wasn't routing. And my dxo machine decided not to play nicely with the monitor; once again I got this out-of-range display that I've had intermittently since I got it, so I couldn't use that. Instead shut down eucla, the laptop on which I'm still building FreeBSD 9-STABLE, and booted it as pain, the old Microsoft XP installation, and ran the application from the CD.

And how about that, it worked! Admittedly, it took forever to do anything, but at the end I was able to set a bare-bones configuration, though not the way I wanted. My default router is in the same network, of course, but this box wants to be the default router, and it wouldn't allow WAN and LAN to be on the same subnet. And although I set the correct local address (in subnet 192.109.197.0/24), when I tried to access http://tplinklogin.net/, I got the router. I also discovered that it had completely overwritten my network configuration and given me a DHCP address, presumably from the router. But it did show that the router had the RFC 1918 IP address 192.168.0.1.

With that information and an IP alias I was able to access it from my FreeBSD machine. But what amazing stupidity:

So finally I had access to the web server in the box. Given the experience to date, I was expecting something like the horribly broken ASUS RT-N13U that I used briefly 2 years ago. But surprise, surprise! It is well laid out and has just about every function I could desire, including bridging. But that, it seems, is another can of worms. There's this term WDS, which I haven't heard before. Found a couple of descriptions online, notably this one and this one, but it looks like a lot of reading is necessary before I continue.


Build: one done, one continues
Topic: technology Link here

I had to shut down stable-amd64 virtual machine today while processing panoramas, which need all available memory, and I shut down eucla (the laptop) while trying to configure the wireless router, but after that I continued and finally got FreeBSD 9-STABLE running on it. Time to try a wireless card—I have an ancient Lucent/Orinoco PCMCIA card that fits. And of course it didn't associate. Powered down the router, and it took notice: it panicked out of wi_intr. And then I noted that I had forgotten to build a kernel with debugger, so there wasn't much I could do about it.

Things didn't stop there, of course: I could no longer log in:

login: in openpam_dispatch(): pam_nologin.so: no pam_sm_autheticate().

I've seen this before, and noted the solution, but I still don't understand why it doesn't get updated automatically.


Explosive food
Topic: food and drink, opinion Link here

Chris Bahlo over for dinner tonight as usual. A very fishy meal: ceviche followed by the remaining whitebait, with which we served deep-fried croquettes. While removing them from the deep fryer, one of them exploded all over me:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130427/small/Explosion-1.jpeg
Image title: Explosion 1
Complete exposure details
Dimensions: 600 x 450, 74 kB
Dimensions of original: 4032 x 3024, 1008 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 27 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130427/small/Explosion-3.jpeg
Image title: Explosion 3
Complete exposure details
Dimensions: 600 x 450, 98 kB
Dimensions of original: 4032 x 3024, 1664 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 27 April 2013:
thumbnails    small images    diary entry

Also Rødgrød med fløde, sort of—really a Rote Grütze with custard, and quite consistent:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130427/small/Rødgrød-med-fløde.jpeg
Image title: Rødgrød med fløde
Complete exposure details
Dimensions: 600 x 450, 72 kB
Dimensions of original: 4032 x 3024, 1056 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Saturday, 27 April 2013:
thumbnails    small images    diary entry


Sunday, 28 April 2013 Dereel Images for 28 April 2013
Top of page
previous day
next day
last day

TP-Link setup, the real way
Topic: technology, opinion Link here

I couldn't really be bothered to work my way through the web links about WDS that I had found yesterday, so today I put the other router where I wanted it, at teevee, and used the powerline Ethernet connection to access it. As I suspected, the stupid installation instructions aren't just plain wrong some of the time, they're much more complicated than the real thing. To configure a TP-Link TL-WR841N 802.11n wireless router, do this:

  1. Ensure that you have an address in the 192.168.0.0/24 address range.

  2. Connect to one of the “LAN” connectors.

  3. Access http://192.168.0.1/ with a browser.

  4. Log in as admin, password admin. This information is also written on the underside of the router.

  5. Select Quick Start and follow the prompts.

And that's it! No long, strange and easily misspelt names, no disruption of your network topography, no changes in your Microsoft IP settings. Of course, there are many other things you can do, notably disable DHCP, but with this you can do everything that the SimplisticEasy Setup does.

We did some further investigation of http://tplinklogin.net/ on IRC today, and came to the surprising discovery that the domain doesn't even belong to TP-Link! The whois information shows:

    Domain Name: TPLINKLOGIN.NET

    Registrant:
        Above.com Domain Privacy
        8 East concourse
        Beaumaris
        VIC
        3193
        AU
        tplinklogin.net@privacy.above.com
        Tel. +61.390057904

Now doubtless this is a domain name squatter, but what a stupid thing for TP-Link to do: require specific topology for configuration, use a name instead of a (shorter) IP address for the device, and then not even own the domain! I'm amazed.

You set up WDS bridging via the “Wireless” menu. At the bottom select “Enable WDS Bridging” and extra fields appear. On each router set the SSID and BSSID to the SSID and BSSID of the other router. That's even easier than it seems: the button with the unlikely name “Survey” scans for SSIDs and presents them for selection:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130428/small/AP-List.gif
Image title: AP List
Dimensions: 529 x 172, 4 kB
Dimensions of original: 529 x 172, 4 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 28 April 2013:
thumbnails    small images    diary entry

After selection, the configuration is complete:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130428/small/Office-config.gif
Image title: Office config
Dimensions: 573 x 471, 58 kB
Dimensions of original: 839 x 689, 116 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 28 April 2013:
thumbnails    small images    diary entry
 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130428/small/Teevee-config.gif
Image title: Teevee config
Dimensions: 573 x 471, 58 kB
Dimensions of original: 839 x 689, 116 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 28 April 2013:
thumbnails    small images    diary entry

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130428/small/Office-config-detail.gif
Image title: Office config detail
Dimensions: 638 x 85, 10 kB
Dimensions of original: 638 x 85, 10 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 28 April 2013:
thumbnails    small images    diary entry

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130428/small/Teevee-config-detail.gif
Image title: Teevee config detail
Dimensions: 638 x 85, 12 kB
Dimensions of original: 638 x 85, 12 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 28 April 2013:
thumbnails    small images    diary entry

And how does it work? Even slower than the powerline adapters! Here a comparison for moving the same file, first with powerline, then with wireless:

Seven_News-2013-03-28-1757              100% 2152MB   6.9MB/s   05:14
Seven_News-2013-03-28-1757              100% 2152MB   6.2MB/s   05:46

Why so slow? One of the restrictions of WDS is that it's half-duplex, but this is highly one-sided traffic. But in each case the speeds look like half the speed of a 100 Mb/s Ethernet: over real Ethernet I get speeds of about 11.2 MB/s. Certainly the wireless adapters are limited by their 100 Mb/s switch component, so I suppose they'll go back. TP-Link also has a model with a 1 Gb/s switch, which coincidentally I found online for the same price as these ones. In general it looks as if the prices for wireless devices have plummeted. Currently Officeworks has a clearance item: two Netcomm Bp121 routers, which look to have pretty much the same specs as the TP-Link, for $29 the pair! That's a clearance item, but even so it's amazing. Of course, maybe it's not an item at all:

 
https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20130428/small/out-of-stock.gif
Image title: out of stock
Dimensions: 735 x 315, 24 kB
Dimensions of original: 735 x 315, 24 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Sunday, 28 April 2013:
thumbnails    small images    diary entry

No stock within 500 km of Melbourne or Sydney suggest that there aren't many anywhere.

Before I go down that route, there's one more thing to try: the powerline adapters have a speed of “up to” 500 Mb/s, but teevee only has a 100 Mb/s network card. I should put in a 1 Gb/s card and see if there's any improvement.


Monday, 29 April 2013 Dereel
Top of page
previous day
next day
last day

Power line Ethernet: slow
Topic: technology, opinion Link here

One potential reason for my slow transmissions with the power line Ethernet adapters was that the interface at one end was only 100 Mb/s. As planned, today I put a 1 Gb/s adapter in that machine, not without difficulty: it is a PCIe card, and the motherboard had only one PCIe slot, already occupied by the graphics card. But for a test I removed the graphics card.

The results? No improvement; in fact it was marginally slower. So much for that. “Up to” speeds seem to require amazingly good conditions. I'll probably still keep them because they remove the cable from the hallway, unless I can find a better wireless solution.


Still more ports pain
Topic: technology Link here

Finally my ports-try has completed,

   122813.19 real     60857.07 user     18081.91 sys

That's a total of 34 hours, and 370 MB of build logs. Did things work? No, not even remotely. X didn't get built, and so many dependent ports didn't either:

checking whether to rebuild gperf header files... checking for POLKIT... no
configure: error: PolicyKit not explicitly disabled and no PolicyKit found
===>  Script "configure" failed unexpectedly.
Please run the gnomelogalyzer, available from
"http://www.freebsd.org/gnome/gnomelogalyzer.sh", which will diagnose the
problem and suggest a solution. If - and only if - the gnomelogalyzer cannot
solve the problem, report the build failure to the FreeBSD GNOME team at
gnome@FreeBSD.org, and attach (a) ...

What's that? Did a bit of looking around and found a port with an identity crisis: is it POLKIT, PolicyKit, policykit or something else? Looking in the installed ports on eureka, I found policykit-0.9_6, policykit-gnome-0.9.2_6, polkit-0.99 and polkit-qt-0.103.0_1. Which was it? And where is the port directory? Once again the Ports Collection is showing itself from its ugly side. And why doesn't a dependency get installed? After some searching found /usr/ports/sysutils/policykit and installed that. That proved to be the correct one; but that's the second dependency that I have to install manually to be able to build X.

Things didn't stop there, of course. The next issue was ImageMagick, which came up with all sorts of problems, including dependencies on files that didn't get installed. After some considerable time discovered that the files I had on my machine didn't match the files in the repository; it took much more time to discover that a couple of months ago I had intended to take over maintainership, something that didn't happen. But presumably I had started to update to the newest version, but not finished.

By comparison, the crash installing the Linux base was relatively benign: it only installs if the emulator is loaded. Once again, it's not clear why the port can't load it if it needs it.


Computatational photography revisited
Topic: photography, technology Link here

I've been making slow progress with the computational photography course. Not that slow, roughly the speed that it's supposed to be taken at, but I started something like 3 weeks after the course started, and it finishes in a week. As a result spent most of the day going through the videos; there are still three assignments with a total of 11 programs to go, so it'll keep me busy.


Tuesday, 30 April 2013 Dereel
Top of page
previous day

Throughput of alternative networks
Topic: technology Link here

I had really wrapped up my investigation of the network equipment I bought last week: the wireless adapters worked, but not fast enough to justify using them. I'll return them. The powerline adapters work too, also not fast enough. But they're both cheaper and marginally faster, and if I were to return them too, the alternative would be to run CAT6 along the hallway again. So I'll keep one pair.

And that would have been that, except that Edwin Groothuis was interested in more testing. He wanted to know what the performance was like if both adapters were next to each other on the same board. So I tried that—not the easiest thing, since to do it right I needed to find two FreeBSD machines with 1 Gb/s network cards. The obvious choice was monorchid, the physical test box with a damaged onboard 1 Gb/s adapter. Put my PCIe card in there, it was recognized—and once again didn't work.

I've had trouble with that card in the past, and so I didn't try very hard. But dxo, the Microsoft box, is also nearby and had a 1 Gb/s adapter. Connected that to the adapters and—nothing happened. No communication. ARP showed that something was happening, but I couldn't connect with rdesktop, my normal way of accessing it, and I couldn't ping either. And it was still giving me out-of-range video signals, so I couldn't talk to it locally. So here, too, I gave up and went back to conventional Ethernet.

But that no longer worked either! Much cursing and swearing, and in the end, with the horror of reinstalling a Microsoft box in my mind's eye, took the disk out of monorchid to see what was going on. It worked fine: clearly not the hardware. Horror, here I come. But while I had the disk in the machine, I finally had the right configuration to do the tests.

=== grog@eureka (/dev/pts/4) /src/Images 17 -> scp Seven_News-2013-03-28-1757 monorchid:/var/tmp/
Seven_News-2013-03-28-1757                 100% 2152MB 5.6MB/s   06:24

In other words, no obvious difference from previous attempts. But scp has many overheads. Edwin asked me to run iperf. The results for powerline and normal Ethernet: 95 Mb/s for powerline, 860 Mb/s for Ethernet.

So how can the manufacturers claim speeds of “up to 500 Mbps”? Under what circumstances do they even come close to this figure? Once again, it sounds like dishonest advertising to me.


Microsoft network pain
Topic: technology, opinion Link here

Things didn't stop there, though. I still had to get dxo back and running. But I hadn't been able to shut it down: the power key just hibernates it, and when I replaced the disk, it came back up again as before, still with scrambled display and unpingable. Somehow managed to get it into safe mode and to display correctly—and the network interface worked! Further investigation showed that the machine was now blocking ICMP, something it didn't do before. And that was presumably due to the firewall, which was enabled. I can't even recall whether it was before or not, but I'm sure I was once able to ping it.

Spent some time looking through the firewall configuration (Control Panel/Network/Windows Firewall), but couldn't see much that I could do there. Then Andrew Perry came up with these instructions. To configure your firewall, you don't go where I had assumed. Instead it's the much more obvious Administrative Tools/Windows Firewall with En... (hanced security carefully hidden by too-narrow columns)/Inbound Rules, from whence you have to do lots more just to enable ICMP. But it worked.

So: the problem was at least scrambled display and unexpected enabling of the firewall. That didn't explain the lack of communication with the powerline adapters, but finally things were finished.

That's what I thought. I replaced the powerline adapters for communication with teevee, booted the machine and... no communication! And one of the adapters was spontaneously powering down and showing no Ethernet connection. Defective? To be on the safe side, I replaced both adapters with the other pair I had bought. Still no communications. Much investigation and cursing later I discovered I had not one, but two defective CAT6 cables, one which I used to connect to dxo, and the other which I used to connect to teevee. After replacing them, to my surprise discovered that the real-world throughput had significantly increased:

2002_20130430023300.mpg                   100% 1583MB   7.8MB/s   03:24
2008_20130430015700.mpg                   100% 3415MB   7.8MB/s   07:19
2002_20130430023300.mpg                   100% 1583MB   7.8MB/s   03:24

That compares with a previous best of 6.9 MB/s. Why? Running iperf showed a throughput of 84 Mb/s, lower than the tests with monorchid. But monorchid didn't get that file transfer speed. There seem to be factors involved that I don't understand.

In any case, what a pain! Maybe I should take time off and do something else.


Ports pain
Topic: technology, opinion Link here

My ports build still isn't finished. The latest error, while building vlc, was one that I've seen before:

(CDPATH="${ZSH_VERSION+.}:" && cd .. && /bin/sh /src/FreeBSD/svn/ports/graphics/frei0r/work/frei0r-1.3/missing --run autoheader)
autom4te-2.69: cannot lock autom4te.cache/requests with mode 2: Operation not supported
autom4te-2.69: forgo "make -j" or use a file system that supports locks
autoheader-2.69: '/usr/local/bin/autom4te-2.69' failed with exit status: 1
*** [./config.h.in] Error code 1

This is the result of using NFS without locking. In the past I've chickened out and installed the ports tree locally. But the real answer is to set up locking on NFS. How hard can it be? Harder than I want to think about today.


Computational Photography or Python tutorial?
Topic: photography, technology, opinion Link here

On with the computational photography assignments today. They're not really difficult; I did one before the network blew up, Convolution. This involves running a multi-pixel window (confusingly called a kernel) over an image and producing a new image where each pixel is the sum of the products of the kernel element with the corresponding pixel covered by the kernel. It would be easy enough in C, but I had to do it in Python, and that required learning still more functions. And when I thought I had it working, the unit test failed: one of the issues was converting floats back to uint8. Python is quite simplistic here: it just truncates. And the kernel had been written to deliberately underflow (negative coefficients, which can't have any useful purpose apart to catch code bugs).

OK, that's simple: limit the result to the range 0 to 255. How? In fact, one of the videos told how to do it. But how do you grep a video? Spent some time going through looking for it, but after about 20 minutes I gave up and RTFMed. The answer is straightforward enough: np.clip(). But it's indicative of why I no longer like watching videos.

On the positive side, the resultant code is much smaller than it would have been in C.


Do you have a comment about something I have written? This is a diary, not a “blog”, and there is deliberately no provision for directly adding comments. It's also not a vehicle for third-party content. But I welcome feedback and try to reply to all messages I receive. See the diary overview for more details. If you do send me a message relating to something I have written, please indicate whether you'd prefer me not to mention your name. Otherwise I'll assume that it's OK to do so.


Top of page Previous month Greg's home page Today's diary entry Next month Greg's photos Copyright information

Valid XHTML 1.0!

$Id: diary-apr2013.php,v 1.49 2023/04/30 04:08:19 grog Exp $