|
| Greg's diary |
| August 2026 |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Saturday, 1 August 2026 | Dereel | Images for 1 August 2026 |
| Top of page | ||
| next day | ||
| last day |
|
More exiftool investigations
|
Topic: technology, photography, opinion | Link here |
Followed up on yesterday's plan to isolate the exiftool bug. To install the latest version, first remove the current version, once I found the name that it had been given:
=== root@hydra (/dev/pts/20) /etc 56 -> pkg delete p5-Image-ExifTool
Checking integrity... done (0 conflicting)
Deinstallation has been requested for the following 2 packages (of 0 packages in the universe):
Installed packages to be REMOVED:
hugin: 2025.0.1_2
p5-Image-ExifTool: 13.36
How I hate the Ports Collection!
So I decided to install it on dereel, my test machine. The help from Google Gemini was less than perfect. It gave the wrong URL (https://exiftool.org/Image-ExifTool-13.10.tar.gz), and it took me a while to find the correct one. Or was it? It confused me by downloading a file called download:
=== root@dereel (/dev/pts/0) /var/db/pkg 26 -> fetch https://exiftool.org/Image-ExifTool-13.59.tar.gz
fetch: https://exiftool.org/Image-ExifTool-13.59.tar.gz: Not Found
...=== root@dereel (/dev/pts/0) /var/db/pkg 29 -> fetch https://exiftool.org/Image-ExifTool-13.10.tar.gz
fetch: https://exiftool.org/Image-ExifTool-13.10.tar.gz: Not Found=== root@dereel (/dev/pts/0) /var/db/pkg 30 -> fetch https://sourceforge.net/projects/exiftool/files/Image-ExifTool-13.59.tar.gz/download
download 7730 kB 3113 kBps 03s=== root@dereel (/dev/pts/0) /var/db/pkg 31 -> mkdir exiftool=== root@dereel (/dev/pts/0) /var/db/pkg 32 -> ls -l
total 176
-rw-r--r-- 1 root wheel 7,916,358 28 May 02:44 download
...=== root@dereel (/dev/pts/0) /var/db/pkg 34 -> cd download
su: cd: download: Not a directory=== root@dereel (/dev/pts/0) /var/db/pkg 35 -> cd exiftool/=== root@dereel (/dev/pts/0) /var/db/pkg/exiftool 36 -> tar xvf ../download
x Image-ExifTool-13.59/
...
I wasn't done yet: it extracted into a subdirectory. But then things worked as described:
=== root@dereel (/dev/pts/0) /var/db/pkg/exiftool 37 -> l
total 1
drwxr-xr-x 8 501 staff 512 28 May 02:44 Image-ExifTool-13.59=== root@dereel (/dev/pts/0) /var/db/pkg/exiftool 38 -> cd Image-ExifTool-13.59/=== root@dereel (/dev/pts/0) /var/db/pkg/exiftool/Image-ExifTool-13.59 39 -> perl Makefile.PL
Checking if your kit is complete...
Looks good
Generating a Unix-style Makefile
Writing Makefile for Image::ExifTool
Writing MYMETA.yml and MYMETA.json=== root@dereel (/dev/pts/0) /var/db/pkg/exiftool/Image-ExifTool-13.59 40 -> make
cp lib/Image/ExifTool/IPTC.pm blib/lib/Image/ExifTool/IPTC.pm
...
"/usr/local/bin/perl" -MExtUtils::MY -e 'MY->fixin(shift)' -- blib/script/exiftool
Manifying 1 pod document
Manifying 209 pod documents=== root@dereel (/dev/pts/0) /var/db/pkg/exiftool/Image-ExifTool-13.59 41 -> make install
Manifying 1 pod document
Manifying 209 pod documents
Installing /usr/local/lib/perl5/site_perl/Image/ExifTool.pod
Installing /usr/local/lib/perl5/site_perl/Image/ExifTool.pm=== root@dereel (/dev/pts/0) /var/db/pkg/exiftool/Image-ExifTool-13.59 43 -> exiftool -ver
13.59
After setting up some symlinks, I was able to confirm that the latest and greatest still has the same bug.
What next? My own version?
|
Finishing the bird netting
|
Topic: animals, general | Link here |
Bruno is due back tomorrow. Time to finish bird netting the dog run. Paul Donaghy along to do the work, while I put in a token appearance:
|
|||||||||||||||||||||
The next step was to cut the netting to shape. But we couldn't! Even the strongest cutters couldn't get through. I suppose it takes a cat to bite through it. Still, while we're investigating, the run is pretty much escape-proof. The rest is just the keep small birds out.
And then Loretta Sayers called to say that she would have to postpone: she has laryngitis. That's a valid enough reason, but we've seen how she has been agonizing about having to return him, and I wouldn't be surprised if they found a way to keep him.
| Sunday, 2 August 2026 | Dereel | Images for 2 August 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Understanding the exiftool bug
|
Topic: technology, photography, opinion | Link here |
As planned, spent some time today investigating the exiftool -geotag bug, in the process seeing a new claim for a download link: https://exiftool.org/Image-ExifTool-latest.tar.gz. Problem: this link also doesn't exist. The twins clearly have their limits.
First step was to reduce the GPS log to a readable format:
=== grog@hydra (/dev/pts/32) /src/GPS/Tracklogs/Phones/GPSLogger/albo 8 -> sed 's:.trkpt lat="::; s:" lon=": :; s:".*time.2: 2:; s:Z.*:Z:' < 20260729.gpx
-37.800248333333336 143.75107333333332 2026-07-28T20:03:38.000Z
-37.80028333333333 143.75113166666668 2026-07-28T20:03:41.000Z
-37.80030166666666 143.75116333333332 2026-07-28T20:03:44.000Z
...
The Z at the end is part of the military time zone designations, representing UTC (or possibly GMT), so they're 10 hours behind. The first entry corresponds to 6:03:38 local time, and not surprisingly the coordinates show me to be at home.
The first surprise was that the specified location of the first image I discussed on 31 July was indeed exactly an hour before the image. It was taken at 8:11:04, but the location was stored as -38.0297806,144.0656194. And looking at the sanitized log file, I see:
-38.029781666666665 144.06561833333333 2026-07-28T21:11:04.000Z
That's 7:11:04 local time, exactly an hour before I took the photo—I had thought 40 minutes. And the coordinates stored in the image are the same—almost. In fact, they show a location 15.1 cm to the north-east. Why? The time is exactly one hour before.
So, is my time conversion simply out by one hour? No. Many of the images are exactly correct. So it seems that some, and only some, of the images get dated an hour too early.
The next step is to investigate why all the image taken at Leigh Grand Junction Bridge were given the same location. But even these sanitized log files, with their precision (most definitely not accuracy) of round 1 nm, are amazingly difficult to read. Some other time.
|
A new GPS mapping app?
|
Topic: technology, opinion | Link here |
I've been using GPS Visualizer to draw maps from my GPS logs for nearly 10 years, and it works well enough. But it doesn't show where I was at a particular point on the route. Is there nothing else available? Once again a question for Google Gemini.
Yes, quite a few. Ignoring the easy drag-and-drop solutions, discovered GPXSee, which is available via the FreeBSD Ports Collection.
It installed an amazing number of dependencies, but no instructions at all. Well, not quite. It tells you how to install it (though not on FreeBSD) and how to report bugs. But not how to use it, not even on the web site.
Played around with it anyway, and it seemed straightforward enough. It was also blindingly fast compared with GPS Visualizer—clearly an advantage of running on my machine. And yes, it had popups (run the cursor over an image to compare it with its neighbour):
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The problem: they're all the same!
|
|
|||||||||
And it's wrong. It gives a time of 20:03:35, by which time I had long been home (never mind that the date is out by 2,000 years). Clearly the other information is meaningless. Quite possibly this is a result of there being no instructions. But before I move on to one of the others I should try to find out how to use it. One thing's clear: the elevation graph at the bottom shows a spike of up to 1,025 m. That's clearly wrong: nothing round here is anywhere near that high.
|
Choucroute again
|
Topic: food and drink, opinion | Link here |
Choucroute garni for dinner tonight, something that we eat from time to time. And it's still a lot of work getting the proportions right. One thing is clear: all recipes that I have ever seen want far too much meat. I'm left with the feeling that they don't really like Sauerkraut.
| Monday, 3 August 2026 | Dereel | Images for 3 August 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
DeliBen mi udang
|
Topic: food and drink, opinion | Link here |
For breakfast today tried one of the pastes that I bought in June, the DeliBen mi udang paste:
|
|||||||||||||||||||||||
The paste was in a second package inside the first. That proved to be good: it's very firm, and my guess is that I only got round 70% out before I had to rinse several times to get the rest. Even then, after cutting open the foil, there was quite a bit left over, which I was able to rinse out:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And the taste? Not bad, but not as good as the Tean's paste. It has more chili (I think Teans has become milder since I first used it), but less prawn flavour. I don't think I'll buy it again as long as I can get Tean's.
|
Tomatoes: winter produce
|
Topic: gardening, food and drink, opinion | Link here |
Tomatoes are interesting things. In general our garden is not doing well. Plants are dying for no obvious reason, including the tomatoes that we planted. But then there are the volunteers. Yvonne picked these today, in late winter:
|
|||||||||||||||||||||||
We got none worth mentioning in the summer, and few in autumn. And there are more still ripening. It's a stark contrast to the first spring flowers:
|
|||||||||||||||||||||||
|
Which aperture?
|
Topic: photography, opinion | Link here |
The background of the spring flowers was disturbing. Bring it out of focus with a large aperture, the largest I have: f/0.95:
|
|||||||||||||||||||||||
That doesn't really work. The background is still evident, but some of the flowers are also out of focus. I did a four image bracket, with apertures of f/0.95, f/2, f/4 and f/8. Probably, for this kind of image, f/4 is the best choice, but basically I need a different background.
|
More playing with GPXSee
|
Topic: technology, photography, opinion | Link here |
Spent some more time with GPXsee today, with little to show for it. Yes, there is a kind of documentation, but once again it doesn't tell me what I want to know. It did, however, give me an idea why the results were so bad last year:
Can't find dimensions for 'Track.jpeg'
|
||||||||||||||||||||||||
What went wrong there? Looking in more detail
-38.00496 144.589895 2025-07-03T00:07:21.000Z
-38.00496333333333 144.589895 2025-07-03T00:07:23.000Z
-38.00496333333333 144.589895 2025-07-03T00:07:26.000Z
...
-38.00496333333333 144.589895 2025-07-03T01:07:40.617Z
-38.02714 144.56247166666665 2025-07-03T01:07:51.505Z
-38.02713166666667 144.56248666666667 2025-07-03T01:07:54.000Z
That's the Little River Bird Hide, but according to the log I was there for almost exactly an hour. And in that time the coordinates did not diverge by 1 nm. There were more cases of the same coordinates being recorded again and again. I can only assume that the GPS receiver malfunctioned, so my idea of taking two phones was good. I need to keep a much closer eye on these things.
| Tuesday, 4 August 2026 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More map software
|
Topic: technology, opinion | Link here |
I give up with GPXsee. Based on the documentation, it's not even clear that it intends to do what I want, showing useful data for any point on the trail. But there's more, so today I tried a couple of web-based tools.
QuickMapTools GPX Viewer looked the best choice. But what a mess!
|
|||||||||||||||||||||||
Can that be improved? I didn't check. What I did find was the popups:
|
|||||||||||||||||||||||
A tastefully incomplete popup showing nothing of interest except the time, tastefully retained in UTC. You have to scroll down inside the popup to find any further information, and then it's just speed (conveniently displayed to the nearest mm per hour) and course:
|
|
|||||||||
Why no coordinates? But clearly the product is useless. How about gpx.studio? At least the map is saner, but...
|
|||||||||||||||||||||||
|
|
|||||||||
The promised popup only occurs when the cursor is set on the altitude graph at the bottom, and then it only shows, well, altitude and gradient. It does show a blue point on the map, but that's all.
I'm baffled. Both of these products have all that it takes to show the location. You'd think most people would find that more useful than hdop, vdop and pdop, whatever they may be. Why don't they show them?
More investigation of my logs from last Wednesday round the Leigh Grand Junction Bridge showed other information. I took photos between 11:43:07 and 11:47:00. What did the log file say?
-37.81048 143.91852166666666 2026-07-29T01:43:00.000Z
-37.81048 143.91852166666666 2026-07-29T01:43:03.000Z
...
-37.81048 143.91852166666666 2026-07-29T01:48:40.000Z
-37.810475 143.91855833333332 2026-07-29T01:48:43.000Z
-37.810435 143.91856166666668 2026-07-29T01:48:45.000Z
-37.81037833333333 143.918565 2026-07-29T01:48:48.000Z
-37.81028166666667 143.91855333333334 2026-07-29T01:48:51.000Z
Oh. I didn't stand still, but once again it claimed that I didn't move more than 1 nm in nearly 6 minutes. What about hirse, my backup phone? I only had it set for one observation per minute, but what I saw was:
-37.81055749 143.91838268 2026-07-29T01:43:11.718Z
-37.81050092 143.91855529 2026-07-29T01:44:13.709Z
-37.81052095 143.9185512 2026-07-29T01:45:14.941Z
-37.81061508 143.91846138 2026-07-29T01:46:17.705Z
-37.8104297 143.91848828 2026-07-29T01:47:19.702Z
That's more plausible. Clearly that's an issue with the phone or the software. Set hirse to update every 2 s. And what about enzian, the old Redmi 9A? Booted it. It seems to be even slower than I remember, but the GPS logging works. Next time I'll take all three with me and maybe have something to compare.
|
Performance issues, part one
|
Topic: technology, opinion | Link here |
Why is enzian, the old Redmi 9A, so slow? Asked Google Gemini, which came up with several suggestions, mainly using adb (“Android Debug”, nothing to do with the adb that we have known since long before Android came along. I should try that, some time.
|
Performance issues, part two
|
Topic: technology, opinion | Link here |
My external web server lax.lemis.com is overloaded again, with load averages over 100. Strangely, fra.lemis.com, which also answers to the name www.lemis.com, is unaffected. More DNS frobbing: cut out the name www for lax to watch what happened. Unlike previous occasions, the web traffic really did stop. But fra didn't get any busier! And lax didn't get much busier when I reinstated the www entry.
I really don't understand what's going on here. But one possibility remains: set up dereel.lemis.com (to my right on the desk top) as another www. That will probably overload my current network link, but we can increase that.
First, copy the web site across. That should be simple: the connection to eureka is 1 GB/s.
But I had an unexpected problem: I can't use ssh keys to connect from eureka to dereel. Yes, I know I have problems with ssh on eureka, but I didn't think that they were that bad. I don't understand the issue: eureka offers a valid key, but it doesn't get accepted:
debug1: Authentications that can continue: publickey,password,keyboard-interactive
debug1: Next authentication method: publickey
debug1: Offering RSA public key: /home/grog/.ssh/id_rsa
debug1: Authentications that can continue: publickey,password,keyboard-interactive
debug1: Trying private key: /home/grog/.ssh/id_dsa
debug1: Trying private key: /home/grog/.ssh/id_ecdsa
debug1: Offering ED25519 public key: /home/grog/.ssh/id_ed25519
debug1: Authentications that can continue: publickey,password,keyboard-interactive
debug1: Next authentication method: keyboard-interactive
Password for grog@dereel.lemis.com:
I recall that RSA keys are no longer modern, though it's still in ~/.ssh on dereel, but it should have accepted the ED25519 key. What's the issue?
The main one, of course, is that the ssh on eureka is over 10 years out of date. But the protocols haven't changed, and apart from authentication everything works. I tried running hydra's ssh on eureka, running into one missing library after another, but culminating in:
=== grog@eureka (/dev/pts/0) ~ 133 -> /hydra/usr/bin/ssh dereel
/lib/libc.so.7: version FBSD_1.5 required by /hydra/usr/bin/ssh not found
That's fatal: for reasons I don't understand, /lib/libc.so.7 appears to have been around for over 10 years, and I can't just change it. So I'm going to have to dig into the distasteful innards of ssh.
| Wednesday, 5 August 2026 | Dereel | Images for 5 August 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
A year of Yaris
|
Topic: general, opinion | Link here |
It's been just over a year since Yvonne got her new Toyota Yaris. She's very happy with it.
Not so I. Right from the start I had difficulties with it, but then I don't have to drive it. Well, not often. One of the problems is that we still have the VZ Commodore, which I now drive. And because of the photovoltaic equipment in the garage, it can only be parked on the right. And Yvonne has difficulty parking her car on the left, so yesterday, after parking my car too far to the left, I had to do it.
I understand Yvonne: it's not easy. There's a certain amount of dead play in the accelerator, and then it makes up for it by lurching forward. It's really difficult to park as accurately as I would have wanted.
But that's nothing compared to this morning. Not for the first time, I forgot that the ignition switch needs two steps to really turn it off. After turning anticlockwise as far as it will go, you need to press the key in and turn it further. If you don't, it drains the battery really quickly. This morning it was already down to 9 V. Nothing for it: send Yvonne off with my car and put hers on charge.
But once again how do you open the bonnet? Once again I had difficulty. Somehow it's really confusing. But it was nothing like the issue with the boot, which is controlled electrically. We couldn't find a way to open it. Later I asked Google Gemini, which told me to fold down the rear seats, crawl into the boot space (removing any contents if necessary), take a screwdriver and fiddle around in the dark to open it. What a mess!
|
Web server on dereel
|
Topic: technology, opinion | Link here |
How do I replicate a web server installation? It's not difficult, but there are tricks along the way. Today I installed it on dereel.lemis.com:
apache24_enable=YES
This is needed even if you just want to start with apachectl start.
After that, it worked. But: it thinks that it's local (it is in the 192.109.197.0/24 net block, after all). So it doesn't find the photo hierarchy. I could fix that with a symlink, but what I really need to do is to find a better way to decide whether I'm local or not.
|
Pizza fail
|
Topic: food and drink, opinion | Link here |
We have a new, large pizza stone, and I should give feedback on eBay. First we should check if it doesn't fall apart when heated.
Good news, bad news: the good news is that it behaves as expected. The bad news is that the big pizza bases that Yvonne found are difficult to handle, and in scraping one onto the stone I folded it:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And that left lots of mess afterwards:
|
|||||||||||||||||||||||
Not the fault of the stone, but how do I avoid a repeat?
| Friday, 7 August 2026 | Dereel | |
| Top of page | ||
| previous day |
|
Unusual spam
|
Topic: technology, opinion | Link here |
Seen in my mail today:
896 + 06-08-2026 To Lehey Greg Advice (1518) + Confirmation that you are now a Responsible Person for The Unix Heritage Society Inc. [SEC=OFFICIAL] CRM:005100123009
Now doesn't that look like spam? Whoever the spammer is, he got my name the wrong way round. And “Responsible Person”? It's no secret that I'm involved (as secretary) with The Unix Heritage Society Inc., so that doesn't mean much.
OK, open carefully. The first thing I look at in HTML attachments are the links: search for http. And there I saw:
href="https://urldefense.com/v3/__https:/www.acnc.gov.au/for-charities/manage-your-charity/charity-portal__;!!B2FmhzcL8Q!WJocNKMMVqdL3y-CvJ7hvimY6_RETOXNrzoEVhGkw14pJtWwuV8N63MwQqRK9L8xdCrXeY-vELblW7nUhIg$"
https://urldefense.com/? That doesn't look like an Australian Government site. More searching.
Oh. It's a protection service, and the following URL does look like an Australian Government site. OK, display the attachement:
Dear Lehey GregCongratulations on being appointed to the governing body of The Unix Heritage Society Inc. (47295586494) as a Responsible Person (also known as board or committee members, or directors). The Unix Heritage Society Inc. is registered as a charity with the Australian Charities and Not-for-profits Commission (ACNC).
Oh. It's an official Australian Government message masquerading as spam, and it brings good news. I suppose it's typical that it turns my name around and mentioned the real purpose of the message in a subclause. Looking in more detail, it mentions all members of the committee. It has kept David Arnold's and Warren Toomey's name the right way round, but swapped Clem Cole's and mine, and then apparently sorted the entries alphabetically, rather than the more normal sequence of President, Vice-President, Secretary and Treasurer. And of course there are syntax errors, like “Who the charity helps”, which sounds like a literal translation of German „Wer dem Verein hilft“, but apparently meaning exactly the opposite.
|
More web site performance problems
|
Topic: technology, opinion | Link here |
Continued installing dereel.lemis.com as my third web server. That went relatively smoothly, but in the process I discovered strangenesses with the other two. Until I understand them, there's no point taking dereel online.
The load on lax is relatively low, under a load average of 10, but I was getting response times of up to 25 seconds—only when I wasn't looking. Spent some time trying to understand wireshark again. I have always had difficulty, but this time I didn't get it to work at all. Instead I ran tcpdump and confirmed that everything responded immediately.
But then there are other strangenesses. Trying to access it site with lynx got errors like
HTTP/1.1 301 Moved Permanently
Is that because it's trying https first—there I do return a 301. But why should it try https when I specify http? And why does it take so long to recover?
Is it time to finally set up https?
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 | Greg's photos | Copyright information |
