|
| Greg's diary |
| October 2025 |
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Wednesday, 1 October 2025 | Dereel → Ballarat → Dereel | |
| Top of page | ||
| next day | ||
| last day |
|
ANOTHER bloody power failure!
|
Topic: technology, general, opinion | Link here |
Into the office first thing this morning to discover that hydra had rebooted. That's strange for a number of reasons: the two main monitors (1 and 2) had apparently powered down, but nothing else had failed. And in the past hydra has powered down and not come up automatically. In addition it was on a UPS, which should have protected it. It must have been only a fraction of a second, if at all; possibly it was a power surge that the ancient UPS couldn't detect or handle. But that's surprising because it took down the monitors as well. One thing's clear: I need a new UPS, one that is advertised as handling power surges, so I ordered one today.
Bringing hydra up wasn't completely simple: for some reason the automatic file system checks failed, and when I got through the manual checks (which used the journals) and loaded the nvidia-modeset module, I still couldn't start X. Why? No idea. It claimed that the module wasn't loaded, but it was, and it had already probed the hardware and reported it. So for once I rebooted the machine, after which things worked normally.
When did it happen? In the night sometime. For some reason syslog wasn't working after reboot, so I didn't get the boot messages. My first suspicion was round 0:30, but then I got various other messages:
Oct 1 01:57:15 eureka kernel: newnfs server hydra:/: not responding
So some time before that. But then the daily messages, sent with some delay, contained:
Date: Wed, 1 Oct 2025 11:43:25 +1000 (AEST)
Local network system status:
hydra up 8:32, 0 users, load 0.09, 0.06, 0.01
That would suggest that it rebooted round 3:11. But can I trust the mail message date? The cron job starts at 0:21. About all I can say is that it was some time during the night.
|
Haircut and shopping
|
Topic: general, gardening | Link here |
Into Sebastopol for a haircut today, then to Bunnings for some shopping, including some flowers:
|
|||||||||||||||||||||||
Where do I put them? I had intended them for the “island” in the driveway, but I'm no longer sure that's a good place.
| Thursday, 2 October 2025 | Dereel | Images for 2 October 2025 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Cabbage noodle experiment
|
Topic: food and drink, opinion | Link here |
For the fun of it, I got a quarter cabbage to experiment with last week. I cooked one breakfast on Saturday, but there's plenty left. OK, I have established that there's a continuum of cabbage dishes, from that one (包菜炒粉/麵, “cabbage fried rice noodles”) via手 撕包菜 (“hand-torn cabbage”) all the way to phat si-io (pad see ew, “fried soya sauce”). How about a bit of an experiment? Hand-tear the cabbage as intended, but use different noodles (Guan miao) and the sauce for phat si-io. The ingredients are similar:
| quantity | ingredient | step | ||
| 140 g | cabbage, torn by hand | 1 | ||
| 20 g | garlic | 1 | ||
| 10 g | ginger | 1 | ||
| chili | 1 | |||
| 5 | Szechuan pepper corns | 1 | ||
| oil for cooking | 1 | |||
| 65 g | minced pork | 2 | ||
| 15 g | oyster sauce | 3 | ||
| 8 g | sugar | 3 | ||
| 20 g | light soya sauce | 3 | ||
| 15 g | dark soya sauce | 3 | ||
| 15 g | vinegar | 3 | ||
| 15 g | fish sauce | 3 | ||
| 75 g | Guan miao noodles, cooked | 4 | ||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The result? Edible, but not ideal. Guan miao isn't appropriate: I need noodles that absorb more liquid. As it was there was too much juice. I had added some tauge when I saw it happeneing, but it wasn't enough:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
In addition the cabbage was too coarse and too thick, making it tough, and the result was too sweet. In Phat si-io the cabbage is cooked at a high temperature, caramelizing the result, but that didn't happen here. too sweet
Independently of all that I had put in 5 whole grains of Szechuan pepper, which proved to be (just) more than enough. I need to experiment more with Szechuan pepper.
|
Gardening again
|
Topic: gardening | Link here |
What do I do with the flowers that I bought yesterday? Plant them, of course, and where I had originally planned:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
I'm still not sure if that's a good place, but it's certainly better than letting them wither in the kitchen. But I still have half of them left, so I'm not done yet. And coincidentally this was the first time in years that I have done anything in the garden. And looking at the photo times, the whole effort took less than 6 minutes.
|
Is it raining yet?
|
Topic: general, opinion | Link here |
It's still very dry, but over the last week we've had about 20 mm (count them, 20 mm) of rain. Has it made any difference? Here the neighbour's pond on 8 September and now:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
I had hoped that it would show a rise, but I don't see much difference. It doesn't help that I took it from a different perspective.
| Friday, 3 October 2025 | Dereel | Images for 3 October 2025 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Digitalocean problems
|
Topic: technology, opinion | Link here |
While writing yesterday's diary this morning, discovered that I couldn't upload any images to DigitalOcean. It just timed out. Further investigation showed some breakage in the connection between Australia and New York, though it's not clear to what extent the output of mtr is relevant, which shows about 6% packet loss at port-channel8122.ccr92.jan02.atlas.cogentco.com. But ping tells a very different story:
22 packets transmitted, 8 packets received, 63.6% packet loss
There were no outage notices from digitalocean, but Daniel O'Connor confirmed that he had similar issues. It seems only to relate to access from Australia. Put in a ticket, but of course by the time they responded, the service was back to normal.
|
Phat si-io again
|
Topic: food and drink, language, opinion | Link here |
After yesterday's experiment, tried the original pad see ew (or is that phat si-io? I think so) again. About the only difference was that I used up the remainder of the cabbage, this time cut finely. And that worked fine.
|
More Android pain
|
Topic: technology, general, opinion | Link here |
What's this?
Can't find dimensions for 'Android-pain-1.jpeg'
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Clearly it's a mobile phone display. It seems that the purple icon is a message bubble or some such nonsensical term. What good is it? None whatsoever, unless it's to annoy me. How did it appear? No idea. How do you get rid of it? All the “help” I could get didn't work, pointing to things that don't exist. After nearly an hour of messing around in different places in the settings, finally got rid of it—and forgot where it was. About the only thing learnt was that Android finds new ways to annoy me, the menu structures differ from phone to phone, and the settings are hidden in all sorts of places you wouldn't expect.
|
Chrome pain
|
Topic: technology, opinion | Link here |
Another thing that has occurred since upgrading hydra is that the editing keys (Emacs-like) have gone away and been replaced by something stupid. How do I reinstate them? This time it was simple, with the help of Google Gemini:
gsettings set org.gnome.desktop.interface gtk-key-theme "Emacs"
And that worked. The only issue is that I frequently have difficulty finding these answers later.
|
Jane Ashhurst visits
|
Topic: general | Link here |
Jane Ashhurst along to visit today, causing Yvonne some logistic problems: both she and Martin Godwin (the farrier) were due this afternoon, and both were delayed to various extents. But in the end all went well.
| Saturday, 4 October 2025 | Dereel | Images for 4 October 2025 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Frustration
|
Topic: general, technology, gardening, photography, opinion | Link here |
Some days things never go quite right. Today was one of them. Nothing went really seriously wrong, but it was enough. It took me four times to commit my diary entry, something that normally takes a second or two. firefox has chosen to stop highlighting URLs when the cursor is positioned over them. Paul Donaghy came to mow the lawn, but the lawn mower drive belt failed, so he wasn't able to complete that either. There was more
|
More load issues
|
Topic: technology, opinion | Link here |
True to their promise, the vultures rebooted lax yesterday. And I couldn't get any response from the automatically restarted shell. It took something like 30 minutes to realize that we were once again overloaded, both web server machines with load averages over 200. I couldn't even get apachectl to stop the web servers. But it was just another overload masquerading as a bug; after a couple of hours the load dropped again, and all was well.
|
Hugin irritations
|
Topic: photography, technology, opinion | Link here |
I still don't understand why Hugin's fast panorama preview always fails the first time I try to start it, and always succeeds the second time. Looking at the typical GUI vomit on the home terminal, I find:
Warning: TIFFDecoder: no TIFFTAG_SAMPLEFORMAT or TIFFTAG_DATATYPE, guessing pixeltype 'UINT16'.
Warning: TIFFDecoder: no TIFFTAG_SAMPLEFORMAT or TIFFTAG_DATATYPE, guessing pixeltype 'UINT16'.
ERROR: 14:00:42.087030 (/wrkdirs/usr/ports/graphics/hugin/work/hugin-2024.0.1/src/hugin1/hugin/GLViewer.cpp:156) SetUpContext(): Error initialising GLEW: Unknown error.
Warning: TIFFDecoder: no TIFFTAG_SAMPLEFORMAT or TIFFTAG_DATATYPE, guessing pixeltype 'UINT16'.
Warning: TIFFDecoder: no TIFFTAG_SAMPLEFORMAT or TIFFTAG_DATATYPE, guessing pixeltype 'UINT16'.
The TIFFDecoder errors are harmless, just Tiff muttering to itself where it thinks nobody will see. But the GLEW error isn't. It's just useless. I suppose I should get the sources and look at what it's really trying to say. Do I care enough?
| Sunday, 5 October 2025 | Dereel | Images for 5 October 2025 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Android USB connectivity revisited
|
Topic: technology, rant, opinion | Link here |
A year ago today I tried in vain to connect my Android mobile phone to a Real Computer. I failed. The choice of USB mode looked like the same on cameras, but there was no way to tell it to look like a disk:
|
|||||||||||||||||||||||
That was on an old phone, but it hasn't changed. OK, Google Gemini, what do I do? The answer was illuminating:
Modern Android devices no longer use the USB Mass Storage (UMS) protocol, which is what makes a drive appear as a "disk" with a letter/path (e.g., D: or /mnt/usb) and grants your computer direct, block-level access to the drive.
And why not? They don't say, but clearly it's an attempt to make things even less useful. It does come up with a workaround, though:
To achieve a similar function for file transfer, you need to enable the File Transfer (MTP) mode on your Android device.
Specifically for FreeBSD it had:
You'll need to install and use a FUSE (Filesystem in Userspace) based utility that can handle the MTP protocol.
Recommended Method: Using jmtpfs or simple-mtpfs
Load the FUSE kernel module: You need to ensure the FUSE module is loaded. You can load it temporarily or make it persistent.
Temporary: sudo kldload fuse
Permanent: Add fuse_load="YES" to your /boot/loader.conf file and reboot.
Install the MTP utility: You can install either jmtpfs or simple-mtpfs from the FreeBSD ports tree or packages. jmtpfs is often recommended for better compatibility.
Install jmtpfs: sudo pkg install fusefs-jmtpfs
Connect your Android device to your FreeBSD machine via a USB data cable.
Unlock your phone and check the notification shade. You will likely see a notification about the USB connection, often saying the device is Charging.
Tap the USB notification and change the connection mode from Charging to File transfer (MTP). This is crucial for the connection to work.
Create a mount point directory on your FreeBSD system where you want to access the files.
mkdir -p /mnt/android
Mount the Android device using the installed utility. You generally need to run this command as root (or use sudo).
sudo jmtpfs -o allow_other /mnt/android
The -o allow_other option is often needed to allow a non-root user to access the mounted file system.
At least that's a useful level of detail. It seems that there were consistency errors with presenting the phone as a disk, so instead of addressing them, they broke compatibility.
There's also a “solution” for the GUI crowd, Android File Transfer, a port which spells itself sysutils/android-file-transfer-qt5. And to run it you must be root and know the name they chose:
sudo android-file-transfer
But I won't need that. The whole thing reminds me of one of my prime rants about Android: you have to jump through hoops to do something as simple as copying files.
|
Planting the remaining flowers
|
Topic: gardening | Link here |
Where do I plant the remaining Viola tricolor and petunias? A number of ideas occurred to me, but I run danger of procrastinating until the flowers have died. OK, how about a hanging basket in front of the house? One of the baskets has a very sickly looking fern:
|
|||||||||||||||||||||||
Plant one of the kind of flower there. Which one? Google Gemini tells me that the violas would be better in the shade. OK, we have a basket. And it's high time to plant the violas:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
So there they are:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
They're going to have to grow a lot to fill that pot.
The remaining petunias went on the “island”, where they certainly don't crowd things:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
Lost kangaroo
|
Topic: animals | Link here |
After dinner, looking out the dining room window, we saw no fewer than 7 kangaroos go across the neighbour's front garden and jump over the fence.
Well, 7 of them do. The other two were a joey in its mother's pouch and a small kangaroo that couldn't manage the jump. It took a while, but two kangaroos came back to see what was going on, while the small kangaroo went back and forward before finding its way over:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
It wasn't until I processed the images that I saw that the small kangaroo was already on the other side of the fence. It seems that the kangaroos were looking at Graeme Swift, who had come out to help.
|
Autofocus limitations
|
Topic: photography, opinion | Link here |
One of the big differences I have noticed between the Olympus OM-D E-M1 Mark II and the OM System OM-1 Mark II is the autofocus. It doesn't look like much on paper, but I certainly found out with the birding meeting in the Western Treatment Plant two months ago.
But taking the photos of the kangaroos still wasn't easy: relatively long focal length (150 mm, corresponding to 300 mm on full-frame 35 mm cameras), relatively dark, and in particular lots of foreground imagery to confuse the issue. One of the shots came out like this, with only the bushes in the foreground in focus:
|
|||||||||||||||||||||||
|
|
|||||||||
|
|
|||||||||
It turned out that the autofocus had still been on “birds”. I didn't get much choice, in particular no kangaroos. The only other animal was “animals such as dogs or cats”, represented by a symbolic cat head. But that worked better. The other issue was that I had the camera set to choose any focus point; limiting to the centre would have helped.
| Monday, 6 October 2025 | Dereel | Images for 6 October 2025 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More kitchen work
|
Topic: food and drink, general, opinion | Link here |
I suppose it takes a guest to show how much work there is in the kitchen. Normally we just manage to get by with one dishwasher load per day, but yesterday it was three, and today another two. It didn't help, of course, that we've been putting on complicated dishes, huevos rancheros for breakfast and sweet and sour fish in the evening.
The sweet and sour fish was interesting. Normally I make it with hoki, but we didn't have enough, so I used king snapper instead. And that made a big difference. It was skin-on, which I would have preferred to avoid, but it wasn't much skin, and Yvonne hadn't noticed. More to the point, though, the fish is much firmer, and the pieces came out of the deep fryer almost unchanged. By contrast hoki tends to disintegrate. So from now on I'll use snapper.
I don't know if it's a coincidence, but there seemed not to be enough sauce this time. Maybe I should increase it by 40%. And a couple of other things struck me: today was exactly a year since the last time we ate the dish. We had originally planned it for yesterday, but Jane wasn't feeling well, so we put it off until today—after I had started soaking the fungus. And it got much larger, round 5.5 times the dry weight.
|
Where does the overload come from?
|
Topic: technology, opinion | Link here |
Lately every morning I've found my web servers overloaded. Why? Round midday it drops from round 160 (lax.lemis.com) or 210 (fra.lemis.com) to under 1. Should I maybe delay crawlers?
|
More Exif pain
|
Topic: photography, technology, opinion | Link here |
Processing yesterday's photos included a surprise:
=== grog@hydra (/dev/pts/15) ~/Photos/20251005 185 -> exifx Lost-kangaroo-8.jpeg
File Lost-kangaroo-8.jpeg
Date taken: Sunday, 5 October 2025, 18:44:58
Exposure: 1/60 sec, f/5.0 (EV 10.6), 1600/33° ISO
Camera: OM Digital Solutions OM-1
Lens: LEICA DG 100-400mm f/4.0-6.3
Focal length: 156.0 mm
Meter mode: Spot Program AE
Size: 5184 x 3888 pixels (20.16 megapixels, 1:1.33)
That doesn't look right. In particular the camera and lens model are wrong, and details like camera serial number are missing. Much searching, to discover that for some reason I have another problem with exiftool when recovering processed cameras. After considerable debugging discovered that copying the data from the raw image solved the issue:
=== grog@hydra (/dev/pts/15) ~/Photos/20251005 204 -> grep kangaroo-8 Makejpeg
AA053116_DxO Lost-kangaroo-8=== grog@hydra (/dev/pts/15) ~/Photos/20251005 205 -> exifcopy orig/AA053116.ORF Lost-kangaroo-8.jpeg
exiftool -all= Lost-kangaroo-8.jpeg
Warning: [minor] File contains multi-segment EXIF - Lost-kangaroo-8.jpeg
1 image files updated
exiftool -overwrite_original_in_place -TagsFromFile orig/AA053116.ORF -all>all -title=AA053116.ORF Lost-kangaroo-8.jpeg
Warning: [minor] Writing large value for MakerNotes - Lost-kangaroo-8.jpeg
1 image files updated=== grog@hydra (/dev/pts/15) ~/Photos/20251005 206 -> exifx Lost-kangaroo-8.jpeg
File Lost-kangaroo-8.jpeg
Date taken: Sunday, 5 October 2025, 18:44:58
Exposure: 1/60 sec, f/5.0 (EV 10.6), 1600/33° ISO
Camera: OM System OM-1 Mark II, serial BJRA15001
Lens: Leica DG Vario-Elmar 100-400 mm f/4.0-6.3
Focal length: 156.0 mm (full frame equivalent: 312 mm)
Focus: MF; C-AF, MF, Imager AF 32.915 m (29.90 - 36.61 m)
Field of view: 6.3° horizontal, 4.8° vertical, 7.9° diagonal
Meter mode: Spot Program AE
Stabilization: On, S-IS Auto (Auto I.S.)
Size: 5184 x 3888 pixels (20.16 megapixels, 1:1.33)
It would be nice if we could get rid of all exiftool bugs.
| Tuesday, 7 October 2025 | Dereel | Images for 7 October 2025 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Harvest in springtime
|
Topic: photography, general, opinion | Link here |
Lots of noise in the media about supermoons, a larger than usual full moon. Today was the “Harvest moon”, the largest of all. And how about that, the sky was clear.
Should I take a photo? I've been there before, and the results weren't stunning. But that was with the old 300 mm f/5.5 Hanimex and a 3x teleconverter. I didn't mention it at the time, but I have since come to the conclusion that the teleconverters aren't worth the trouble.
|
|||||||||||||||||||||||
The following month I got a better result with the Zuiko Digital ED 70-300 mm f/4.0-5.6:
|
|
|||||||||
That was better, but can I do even better? Today I used the Leica DG Vario-Elmar 100-400 mm f/4.0-6.3 at 400 mm. It also didn't come close to filling the frame:
|
|||||||||||||||||||||||
With a little processing, got a reasonable result:
|
|||||||||||||||||||||||
Is it better than the photo of April 2011? It's still not close to the resolution of a Real Photo, as I commented at the time. Again I wonder why I bother. And to do it right I really need a lens with a focal length round 1300 mm (corresponding to 2600 mm on a “full-frame” camera). And they don't exist.
| Wednesday, 8 October 2025 | Dereel | Images for 8 October 2025 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
ALDI noodles
|
Topic: food and drink, opinion | Link here |
Phở bò for breakfast today, a chance to try out some of the noodles that we bought from ALDI last week, the Thai rice noodles:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
They're irritating. The results weren't bad, but they're packed in 100 g cakes, so taking 60 g is almost impossible. I took 67 g, ⅔ of a cake:
|
|||||||||||||||||||||||
They're probably a little wide for Phở, but they can be useful if other sources dry up.
|
Goodbye Jane
|
Topic: general, opinion | Link here |
Jane Ashcroft left for Melbourne today, giving me time to catch my breath.
|
More web server overload
|
Topic: technology, opinion | Link here |
So far my web servers seem to have suffered from overload in the mornings, and every time I have checked I have found overnight jobs running and consuming many resources. Could they be to blame? No. Today the overload came in my afternoon. But once again it didn't last long.
|
New cameras
|
Topic: photography, opinion | Link here |
The last camera I bought was an OM System (formerly Olympus) OM-1. Today I got two new cameras: an Olympus OM-1 and an OM-10, along with three lenses: a 50mm f/1.8 F.Zuiko, a 28 mm f/3.5 G.Zuiko Auto-W (a surprisingly unknown lens; this review is the best I could find) and a S Zuiko 100-200 mm f/5 zoom, along with a T 20 flash unit that appears defective.
OM-1? Don't I already have one? Little could be further from the truth. The “new” camera is a film camera that was first presented in 1972, 50 years before the digital OM-1. The name of the manufacturer is irrelevant: the first digital OM-1s also carried the Olympus name. Here images from Wikipedia:
Why these cameras? The OM-10 (my second) came along for the ride, along with the S Zuiko 100-200 mm f/5 zoom. But the (original) OM-1 is a classic, and somehow it completes the cycle.
How old are the cameras and lenses? Hard to say. Google Gemini has something to say about the camera and lens, but it's “some disassembly required”. The OM-1 is the n version (OM-1n, introduced in 1979), so arguably it's of the same vintage as my Nikon FM2. But looking at the 50 mm f/1.8 lens, this page suggests that it's relatively early. It has the second of four formats (“F.ZUIKO"), and the serial number 590761 is far below the quoted “typical” serial number 3248058. I had hoped to get some information from the serial numbers of my first OM-10 and 50/1.8, but at the time I didn't record them.
|
New UPS
|
Topic: technology, opinion | Link here |
Also got a new UPS, a CyberPower VP1600ELCD. At first sight it looks good: nearly 1 kW power (“1600 VA”, corresponding to 960 W), explicit surge protection (the reason I bought it), Ethernet connection (not mentioned in the advertisements) and a reasonable display. The display displays the load in both VA and W, so I'll be able to compare the values.
Put it on charge. When will I connect the computers? They need to be rebooted, so I'll wait for the next power failure.
| Wednesday, 15 October 2025 | Dereel | |
| Top of page | ||
| previous day |
|
USB-C: The universal connector that isn't
|
Topic: fiction, technology, general, opinion | Link here |
This hasn't happened yet. In fact, it will never happen. This is the first of a set of entries generated by ChatGPT on 31 August 2025. See there for discussion.
Spent the morning trying to connect my new external drive to the laptop, only to discover that not all USB-C cables are created equal. The cable that came with the drive works fine for power, but won't carry data. The one from my phone charges everything but transfers nothing. And the expensive "high-speed" cable I bought last month? Perfect for everything except video.
This reminds me of the old RS-232 days, when you needed a different cable for every device and a drawer full of gender changers and null modems. We thought we'd solved this with USB, and for a while we had. USB-A was USB-A. It either worked or it didn't.
Now we have USB-C, which looks identical regardless of capability. USB 2.0 speeds, USB 3.2, Thunderbolt 3, Thunderbolt 4, power delivery at various voltages, DisplayPort alternate mode, and combinations thereof. All in connectors that look exactly the same.
This is a perfect example of what happens when committee design meets marketing pressure. The engineers wanted flexibility, the manufacturers wanted cost savings, and the marketing departments wanted "universal". What we got was universally confusing.
At least the old parallel port cables had different connectors for different capabilities. You knew a Centronics cable was for printers, not modems.
|
Morning photography
|
Topic: fiction, photography, general | Link here |
The light this morning was exceptional—one of those crisp autumn days where everything seems to have more contrast than usual. Took the opportunity to reshoot the house from the north side for the monthly house photos.
While reviewing the images, I noticed something I'd missed before: the shadow patterns from the solar panels create an interesting geometric overlay on the garden below. It changes throughout the day, of course, but at this time of morning it creates an almost Art Deco pattern across the lavender bushes.
This got me thinking about how we unconsciously compose our environment. When we installed those panels five years ago, I was focused entirely on energy efficiency and roof aesthetics. The shadow patterns were an unintended consequence that turned out to be rather pleasant.
Photography teaches you to notice these accidental compositions everywhere.
|
Cat behavioral patterns
|
Topic: fiction, animals, general, opinion | Link here |
Piccola has developed a new routine. Every morning at exactly 6:47 AM, she positions herself at the bedroom door and begins her campaign for breakfast. Not 6:45, not 6:50—6:47.
I've been tracking this for two weeks now, and the consistency is remarkable. She's more accurate than my atomic clock. How does she do it? Internal circadian rhythms? Sensitivity to changes in light that I can't perceive? Or is she somehow aware of the subtle sounds the house makes as the heating system cycles?
Bruno, by contrast, operates on "breakfast happens when the human gets up" time, which varies considerably. Two different strategies for the same goal.
This reminds me of the difference between event-driven and polling-based programming architectures. Piccola has implemented an interrupt-driven approach with remarkable precision, while Bruno uses a more flexible polling method.
Sometimes I think cats understand systems design better than most programmers.
|
Afternoon frustrations with smart devices
|
Topic: fiction, technology, general, opinion | Link here |
The "smart" thermostat decided to update its firmware this afternoon. Without warning, naturally. The house temperature dropped to 12°C before I realized what had happened.
Post-update, the interface has changed completely. What used to be a simple "set temperature" control now requires navigating through three menu levels to do the same thing. Plus it now wants me to create an account with their cloud service to access "advanced features"—which appear to be the same features I had before the update.
When did thermostats become subscription services? It's a device that should set a temperature and maintain it. Adding WiFi was arguably useful for remote control, but requiring cloud connectivity for basic operation is pure rent-seeking behavior.
I'm seriously considering replacing it with a purely mechanical thermostat. At least those fail in predictable ways.
|
Evening reading: RFC archaeology
|
Topic: fiction, technology, general, history | Link here |
Spent the evening reading through some old RFCs, particularly RFC 793 (TCP specification) and RFC 791 (IP). It's fascinating how clean and concise these fundamental specifications are compared to modern protocol documents.
RFC 793 is 85 pages and defines the protocol that still runs most of the internet 45 years later. Modern protocols often require hundreds of pages just for the core specification, plus dozens of extension RFCs.
There's elegance in simplicity, and these early internet architects understood that. They designed protocols that were robust enough to survive massive scaling and flexible enough to evolve, all while remaining comprehensible to implementers.
Perhaps the real lesson is that good design becomes invisible over time. We take TCP/IP for granted precisely because it works so well.
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 |
