|
| Greg's diary |
| July 2026 |
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Wednesday, 1 July 2026 | Dereel | |
| Top of page | ||
| next day | ||
| last day |
|
More X problems
|
Topic: technology, opinion | Link here |
I've had hydra.lemis.com for nearly 3 years, but I still haven't finished configuring it. In that time various software updates have irritated me. I really need to fix things.
But today I had a new problem: I couldn't copy text from on X window to another. Spent an inordinate amount of time trying to find out why not, in which time I discovered that it only affected server :0. Window managers? Somehow fvwm3 seems to be particularly flaky, possibly because I'm using it in a manner that they didn't intend: one window manager per screen instead of one window manager per server.
OK, Google Gemini, give me a sample fvwm3 configuration. Put that in server :2, and it almost worked. Now I just need to add my add-ons, mainly menus. The only thing that didn't work was that it menu only worked on display 2:0, not 2:1 and so on. That, too can be addressed.
But it still didn't explain why my cutting and pasting didn't work on server :0. Stopped and restarted Emacs, to no avail. And when I marked something on an xterm, the highlighting disappeared as soon as I let go of the mouse button. More Google Gemini. It came up with references to xsel and xclip, neither of which I knew. Neither did hydra. A quick search showed that I needed to install the xclip and xsel-conrad ports.
The suggested invocations didn't help, but with a bit of experimentation I came across:
=== grog@hydra (/dev/pts/25) ~ 332 -> xclip -o
xclip: Error: 'x2x bde:0' (0xe800006) cannot convert PRIMARY selection to target 'STRING'
Oh. That makes sense. x2x is ancient, and nobody alive today seems to know how it works. It probably predates clipboards and things. Shoot that down and all was well. It only took me half a day.
|
New teleconverter... not
|
Topic: photography, general, opinion | Link here |
Today was Yvonne's shopping day, and on the way home she had no fewer than 3 packages to pick up at the Napoleons post office.
Three? She came back with two. According to AusPost the teleconverter had been waiting there since Monday, but they weren't able to find it. Somehow this item has been more irritating than I expected.
And the filter for the M.Zuiko Digital 150-600 mm? Despite the claims, not coated. It goes back.
Not so fast, says eBay. You're an abusive buyer!
Huh? I've seen that before, though there's no obvious reason why it happened, nor that it should happen now. This greatly annoys me. My guess is that it has triggered a false positive somewhere. Once again I'll have to talk to a Real Human, though I suspect that it's time to contact Consumer Affairs Victoria and complain.
|
Mail from Consumer Affairs Victoria
|
Topic: general, technology | Link here |
As it happened, Consumer Affairs Victoria got in touch with me first. Not about eBay, of course; they want the Annual Report that I was expecting to submit. It seems to be mainly financial, but we have until 22 November to submit it, coincidentally the day after our planned next Annual General Meeting.
|
Dog found
|
Topic: animals, general | Link here |
Call today from Rachelle, the ranger. They were able to find the owner of yesterday's dog, apparently in the neighbourhood. The dog is even older than we thought, 17 or 18 years old, blind and deaf, and she wandered off while the owners weren't looking. Not what I was expecting, but it's nice to know that she got back again.
| Thursday, 2 July 2026 | Dereel → Napoleons → Dereel | Images for 2 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Phat si-io experiment
|
Topic: food and drink, opinion | Link here |
The biggest issue with phat si-io is the slight charring that the cabbage and the noodles require. In particular the noodles can easily fall apart, and you're left with a mess.
That's done in a wok, because that's what people have. But how about deep-frying? Tried that today, with surprising results.
The cabbage gave off an amazing amount of steam, and very quickly it browned to a point that I hadn't expected:
Too much? No, it was perfectly acceptable.
And the noodles? I separated them and then deep-fried them in two batches:
|
|||||||||||||||||||||||
The result was surprising. They didn't brown much, but they blistered:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And when I removed them from the deep fryer, they had stuck together:
|
|||||||||||||||||||||||
I had to cut them with a knife.
Clearly not a success. What do I do next time? One possibility is to really fry each noodle separately, but that won't char them, and there's also the question whether it's really worthwhile. Maybe just fry them normally?
|
Google security
|
Topic: technology, opinion | Link here |
Received on my mobile phone from Google today:
|
|||||||||||||||||||||||
Now isn't that accurate? No useful information at all, and not even the correct operating system. And location resolution to the continent level, which they at least got right. If they understood DNS, they could pinpoint me to within metres via the LOC RR. And that because I logged in from a different web browser on hydra.lemis.com, the system I always use.
|
Lost data!
|
Topic: technology, opinion | Link here |
If there's one thing that horrifies me, it's losing data. 25 years ago I lost two diary pages, and it still haunts me. And then, in my overnight backups, I saw:
deleting grog/20250703/PC/Fresh-land-food-1.jpeg
deleting grog/20260103/orig/71030007.jpg
deleting grog/20260103/orig/71030007.ORF
The first one is probably OK. But why did I remove the last two? The number shows that it was taken with my Olympus E-330, and the converted image (Pantry.jpeg) was still there. But no 71030007 anywhere!
As it happened, found a CF Card with the image, but why did it go? Have I lost others without noticing?
And then, when I wrote my diary, the whole entry for yesterday was gone. That's not an issue: it's backed up and also copied to the web sites. But I had made notes after the <?php exit () ?>, and they were gone.
I was able to recover from memory, but it's unsettling. My best bet is that it had something to do with yesterday's X problems, where I started multiple Emacs. But how could I lose the updates?
|
1680 mm focal length?
|
Topic: photography, opinion | Link here |
Yvonne has been waiting for a saddle from Austria for some time now, and today it finally arrived. Time for a trip to Napoleons to pick it up.
But there's also the teleconverter that they couldn't find yesterday. So I printed out the tracking information and went there myself to pick it up, braving threats that I needed permission to pick up the saddle.
But they had already found the teleconverter, and it was waiting on top of the saddle. No issues with authentication.
The teleconverter was as advertised, along with a pouch and an oversized carabiner that could in fact be quite useful. It's interesting to think that, despite the information issues, the thing arrived almost as quickly as it could, and it spent nearly half the time from order to now waiting in Napoleons post office.
So, how well does it work? Of course I had to try it out. What lenses does it fit? The 150-600, clearly, but what others? The issue is the protruding lens element, which excludes most lenses. But how about the Zuiko Digital ED 35-100 mm f/2.0? That's a Four Thirds System lens, so the teleconverter element would protrude into the adapter, not the lens. And a 50-140 mm f/2.8 might make sense.
But I couldn't connect the teleconverter to the adapter! I'm still not 100% sure, but it seems that there is a minor difference in the flange that makes it impossible.
So back to the 150-600 (now 210-840 mm), “full frame” equivalent 420-1680 mm. Here a hand-held shot at 840 mm, taken at 1/60 second. According to my old rule of thumb, it should be hand held at not below 1/1600 s. That would require about 8.5 stops of stabilization. Is it sharp? Not completely, but I think that is at least partially to due to the fact that it was windy. It's interesting to notice what looks like scratches on “the negative”. They're relatively sharp, and I think they were raindrops.
|
|||||||||||||||||||||||
|
|
|||||||||
OK, this is only 840 mm. I already have better: my Hanimex 300 mm f/5.5 telephoto with 3× teleconverter. That gives me 900 mm. I could also put on the 2× teleconverter to give me an 1800 mm f/33, but my recollection was that the 900 mm would be bad enough. No hope of image stabilization, of course. Mount it on a tripod and try some comparison images. There's no difficulty telling these two apart:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That's not just the lens sharpness. Another comparison:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The problem with the horse image was that it was very difficult to focus, not helped by total capitulation of focus peaking, and of course there was no image stabilization on the Hanimex.
And then there's this one:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Here the leaves are moving. Maybe I shouldn't pay it much attention; about the only thing of interest is the lack of contrast in the Hanimex.
What does this tell me? The new lens is good. But I could have bought something like that (the Sigma 150-600mm F5-6.3 DG DN OS 'Sports') for little more than half the price. Only it doesn't have the same image stabilization, and the autofocus wouldn't have been the same. So I feel that I made the correct decision in this lens.
| Friday, 3 July 2026 | Dereel | Images for 3 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Filthy dishwasher
|
Topic: general, food and drink, opinion | Link here |
It's been just over a year since we got our new LG dishwasher. It seems to have been one of the worst decisions I have made: it is very difficult to load (an issue with the badly designed baskets), and maybe as a result things frequently come out still dirty. In retrospect, don't ask a dishwasher repairer for advice in choosing the machine: he will tend to recommend the one that is easiest to repair.
Today was time to clean the filter. That's always an issue, but the LG makes it very difficult to remove the filter and clean it. And maybe it's a symptom of the problem that the filter area is full of sludge:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Arguably this could be related to my choice of using very little dishwashing detergent, but it hasn't been an issue on any other dishwasher that we have had.
|
The pain of ring flash
|
Topic: photography, opinion | Link here |
Taking the photos of the dishwasher required flash, of course. What kind? Ring flash. I have four different lens-mounted “flashes”, one of which is really a toy LED device that doesn't produce enough light to be worthwhile. The other three have the issue of mounting rings, and in the end I only found one that would fit any suitable lens, the Viltrox JY-670, and then only some lenses: my standard M.Zuiko Digital ED 12-100 mm f/4.0 IS PRO has a 72 mm filter mount, too large for any of the flashes. So I had to use the Leica DG Vario-Elmarit 12-60 mm f/2.8-4 (with a , not a lens that I use very often. DxO PhotoLab tells me that it's the first time I have used it on the OM System OM-1 Mark II.
|
More lost dogs
|
Topic: animals | Link here |
Off to walk the dogs today. The gate was open because we were expecting Chris Bahlo to come along. And then Yvonne called from inside the house: “Watch out, Mona is coming”. So I went to close the door, and the dogs didn't wait. Out they went down Grassy Gully Road and disappeared.
I got into the car and went searching for them. My biggest concern was that they would get onto the Colac-Ballarat road, but I didn't see them there. About the best news was that the sheep in the paddock to the south of Grassy Gully Road were looking calm, and there were kangaroos to the north that also seemed undisturbed. Yvonne went looking down a track to the north where we sometimes go walking.
|
|||||||||||||||||||||||
Nothing. After about 40 minutes I picked up Yvonne and we went round more roads. Still nothing. I gave up, she went round Dereel, also drawing a blank.
And then they came back, from the direction where we had been looking for them. Where were they? How did we miss them? In any case, the panic was over.
| Saturday, 4 July 2026 | Dereel | Images for 4 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
The return of hope
|
Topic: animals | Link here |
Gradually there's less to do. Hope returns. Or, in Spanish, la Esperanza regresa. Today Chris and Melanie Bahlo came along and performed the latest in an over 25 year long series of horse exchanges: Carlotta went to Chris, and Esperanza came to us.
|
More bird photos
|
Topic: photography, animals, opinion | Link here |
In late afternoon, saw a couple of rosellas in the trees to the north of our property. Off to get my camera and the M.Zuiko Digital 150-600 mm. And of course, by the time I got back, they were gone. All I saw was a couple of smaller birds, and I couldn't even focus on them: the camera was set to “cat focus”, and the best it could find was this:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
In principle, a wasted shot. But it's interesting nevertheless: it was getting dark, and the fact that I got anything at all was impressive, especially such a small bird at 50 m. It was marginally underexposed, though I had set exposure compensation to +1 EV. Based on the postprocessing, probably 1⅔ EV would have been better.
| Sunday, 5 July 2026 | Dereel | Images for 5 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
What's that bird?
|
Topic: animals, photography | Link here |
Yesterday's sole bird photo wasn't very good:
|
|||||||||||||||||||||||
But is it good enough to identify? Out looking and came up with Merlin Bird ID, which also has a mobile phone app. And yes, after finding out how to use it, it identified the bird: a yellow-faced honeyeater.
Is it right? The colours don't look convincing. But then, it was a bad shot.
Also signed up on Yet Another Facebook group, Victorian Birders, who were able to give me some assistance on finding places to take photos, though it really seems as if there is little choice.
|
Cailles aux raisins secs again
|
Topic: food and drink | Link here |
Cailles aux raisins secs for dinner tonight. The last time I found (missing the case six months ago), I had noted that even 60 minutes wasn't enough, and that we should try 90 minutes next time. Today was that time.
The result? Interesting. The sauce had caramelized, with quite pleasant results:
|
|||||||||||||||||||||||
I suppose that it's symptomatic that we couldn't finish two quails between us.
| Monday, 6 July 2026 | Dereel | Images for 6 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Still more spam
|
Topic: health, technology, opinion | Link here |
Received an SMS on my mobile phone today:
Hi Gregory we act on behalf of Dorevitch Pathology and are contacting you regarding your overdue account. Please make payment of $75.00 today by clicking here https://p.armagroup.com.au/6KOHW2Q0. If paid, please ignore. Regards, ARMA
That's clearly a follow-up on other SMSs that I received, first on 18 May. It almost looks genuine. But then, so do most good scams.
The real question is: what is the legal status of an SMS? They're very prone to abuse, to the point that the government has introduced identity checks.
be highly suspicious if the message includes:
And that's exactly what they ask for. But that's only one of the alarm bells:
OK, time to enter a report. I wonder if anything will come of it.
|
Lamb's foot
|
Topic: animals | Link here |
Seen while walking the dogs today:
|
|||||||||||||||||||||||
The only remains of a lamb. Fore or hind feet? We couldn't decide. Yvonne thought that it might have been eaten by a fox, but my recollection was that foxes killed for the fun of it and then only eat small parts of the prey.
Both dogs were interested, particularly Larissa. We let her take them back home with her:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Was that a good idea? Would she get the taste and go on to chase sheep? No, I don't think so. They clearly didn't when they disappeared on Friday. But what if people saw her and came to the wrong conclusions (“vicious hunting dogs”)?
| Tuesday, 7 July 2026 | Dereel | Images for 7 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Quail nasi goreng
|
Topic: food and drink | Link here |
I had leftovers from Saturday's dinner: nearly a whole quail, raisins and rice. What kind of breakfast can I make from that? In principle it sounds like nasi goreng. Do the twins have any other ideas?
No. The recipe they came up with was almost identical to mine. The only interesting suggestion was the addition of cashew nuts. I can do that.
| quantity | ingredient | step | ||
| 72 g | Quail meat, cooked | 1 | ||
| 16 g | Duck fat | 1 | ||
| 10 g | Garlic | 1 | ||
| 5 g | Ginger | 1 | ||
| 170 g | Cooked rice | 1 | ||
| 23 g | Cashew nuts, chopped | 1 | ||
| 20 g | Soya sauce | 1 |
|
|||||||||||||||||||||||
The duck fat was an experiment: it's now considerably cheaper than lard in this bizarre culinary landscape. But it was too much. It's only really necessary for the ginger and garlic, and 5 g would be enough for that.
It's interesting how firm this quail flesh is. I had cooked it for 90 minutes, but it was still very firm, and while eating it I discovered that I had missed a number of small bones:
|
|||||||||||||||||||||||
|
Begone foul scammer
|
Topic: technology, health, opinion | Link here |
As planned, raised a scam report about these irritating SMSs with Scamwatch today. Apart from their insistence on specific non-standard date and time formats, all seemed to go well, and they responded quickly with a confirmation. The most interesting things were:
The advice: “Stop all contact with the scammer. Block the scammer from your phone, email, messaging platforms and social media sites”. OK, done. That confirms my suspicion that SMS is not an appropriate medium for sending invoices. If they really are genuine, they will have to find another way to contact me.
“If your identity credentials were compromised (like your Medicare card, Tax File Number, or driver’s licence) report it to the relevant agency”. That wasn't the case here, but it's interesting to note what they consider to be identity credentials.
I got not one, but two email confirmations:
13 N + 06-07-2026 To groggyhims@le Scam Victims Compens ( 36) N + FW: Scam Victims Compensation from Internet Crime Complaint Center NY Department (IC)
15 N 07-07-2026 To scamhelph@lem Do Not Reply Scamwat (2818) N Your report accc-scamwatch:0017561033 to the National Anti-Scam Centre’s Scamwatch service
But the first was a scam! That's clear when I look at the To: address. groggyhimself (not the real address) is my standard email address, and scam-related stuff goes to scamhelp. Another vindication of my choice of different email addresses for each contact.
|
Power advisors: new low
|
Topic: Stones Road house, general, opinion | Link here |
I followed a link somewhere offering me house electric power upgrades. One of these “you pay us your current bill for 6 years and then the system is yours” schemes. I've seen them before, but I also have the option of paying up front. In view of the upcoming free midday power, it's worth following. So yesterday I spoke with Mark Borja <mark@energyassistanceaustralia.com.au>, who arranged for a phone call from a government expert today at 14:00. To be really sure, he called me again abourt ¾ hour before and warned me that the phone call would not display a number.
Warning bells! He told me that the caller was Nathan of Rebate Assistance Victoria. Another question for Google Gemini, this time with a warning response:
No, this specific website is not an official government site, and you should treat it with a high degree of caution. While it looks professional at a glance and uses regional, official-sounding language (such as a standard Acknowledgement of Country), it is a private, third-party lead-generation site designed to gather your personal and property details and sell them to solar sales companies.
There was much more. So I sent Mark a message asking him to cancel the appointment.
Nathan didn't know about that—in fact, he didn't seem to know very much at all. He called me as planned, and since he was on the phone, I listened to what he had to say.
I first had to point him to what he should have known, and what was in his documentation: we already have a photovoltaic system with 10.8 kW of panels. His suggestion: add 6.6 kW of photovoltaic panels. This would completely eliminate bills in the summer. Part of his explanation was a choice of power suppliers who would offer me up to 12¢ per kWh for feed-in, instead of the 1¢ I'm being paid at the moment. That doesn't add up: in the summer I still consume a lot of power at night, and that starts about 1 hour after sundown. Add the 6.6 kW and it might extend that time by an hour at the most.
And what about the free power in the middle of the day? I suspect that he hadn't even considered that. When I asked him, he said that was “also an option”. And one way or another he would want to install a second inverter. I asked whether it should be in series or parallel, and I strongly suspect he didn't know the terms, and he wasn't able to give an answer.
OK, this had been going on long enough, and I hadn't expected this level of lack of knowledge. Let's finish the discussion. How much would his suggestions cost? For the 6.6 kW of panels and inverter, $13,890. For the 20 kWh battery and inverter, $16,000 odd. That's round $1000 more than Ballarat Solar Panels wanted for a 25 kWh battery. So we could forget the current offer even if they were to sound like they know what they were talking about.
It wasn't until I wrote this that I discovered that this closely resembles the nonsense I had two months ago, down to the the offer of 6.6 kW of PV panels. About the only insight of interest is from the twins: check out Solar Victoria, which is a government-run organization.
Mañana.
|
Bratwurst yet again
|
Topic: food and drink, opinion | Link here |
Bratwurst again for dinner. Two months ago I established that the sausages needed to be cooked at 210° for 20 minutes in the “hair dryer”. But this time things went much faster. Here last time after 20 minutes and today after 14:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The good news is that the sausages didn't burst.
| Wednesday, 8 July 2026 | Dereel | Images for 8 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More phone smart
|
Topic: general, technology, opinion | Link here |
Checking on something on albo.lemis.com, my mobile phone today, and discovered a missed call from Yvonne, made at 10:45. I didn't have time to call back: she had given up and come home. But why am I continually getting missed calls when the phone is in my pocket? I had another yesterday.
OK, reboot, and all was well. Or so I thought. Some time later I received another notification, this time at 12:34:
|
|||||||||||||||||||||||
Fool!. That was the time of you looked at it, not the time of the notification, which was at 11:47 (Android makes it easy to find out: just subtract 46 minutes from 12:34). But there was no evidence of anything round that time. Fool again!. That was the time of the notification, and it gave me the time: 10:45 AM. Never mind that it had already notified me of that call.
Why are these things so flaky? Is it just Xiaomi? I can't really think so. Coincidentally Telstra had a network outage today, and we use their network. Could that be the background? It wouldn't explain yesterday's outage.
Maybe that's another reason to try WhatsApp.
|
Lens database
|
Topic: technology, photography | Link here |
For the fun of it more than anything I have started to create a database of photo-related information. Last month I started collecting Exif data from my photos, though it's beginning to look as if I will have to start again on that table.
Today I started at the other end: there's a list of micro four thirds lenses on Wikipedia. Not easy to search. Another table? But how do I import it?
I know now: with difficulty, aided by Emacs. I can't just copy the columns directly: for a lens like my M.Zuiko Digital 150-600 mm there are 3 columns:
| Focal | Equivalent | Aperture | ||
| length | length | |||
| 150-600mm | 300-1200mm | f/5.0-6.3 |
That really needs to be 6 columns. And some of the entries have this stupid “equivalent aperture” squeezed in with “equivalent focal length”. All this needs to be tidied. In addition, there are multiple tables. For zoom lenses there are more columns than for fixed focal length lenses, so it's not “one size fits all”.
In the meantime I have a script 5,147 lines long, and it's not done yet. But to my surprise I have 129 entries in the table, like
mysql> SELECT * FROM lens WHERE min_focal_length = 150;
+----------------------+---------------------------------------------------+------------------+------------------+-----------+-----------+--------------+--------------+--------+----------+--------+--------+----------+--------+-----------+
| brand | name | min_focal_length | max_focal_length | min_equiv | max_equiv | min_aperture | max_aperture | filter | diameter | length | weight | elements | groups | min_focus |
+----------------------+---------------------------------------------------+------------------+------------------+-----------+-----------+--------------+--------------+--------+----------+--------+--------+----------+--------+-----------+
| Olympus | Olympus M.Zuiko Digital ED 150-400mm f/4.5 IS PRO | 150 | 400 | 300 | 800 | 4.5 | 4.5 | 95 | 115.8 | 314.3 | 1875 | 28 | 18 | 1.3 |
| OM Digital Solutions | OM M.Zuiko Digital ED 150-600mm f/5.0-6.3 IS | 150 | 600 | 300 | 1200 | 5 | 6.3 | 95 | 109.4 | 264.4 | 2065 | 25 | 15 | 0.56 |
+----------------------+---------------------------------------------------+------------------+------------------+-----------+-----------+--------------+--------------+--------+----------+--------+--------+----------+--------+-----------+
Still a lot of work to do. Even in this short example, there's obvious error: the lens names shouldn't include the maker name. And they should match the names reported by the lens. In the case of the 150-600, it's “Olympus M.Zuiko Digital ED 150-600mm F5.0-6.3”. My Exif routines sanitize this to “Olympus M.Zuiko Digital ED 150-600mm f/5.0-6.3”, or maybe “Olympus M.Zuiko Digital ED 150-600mm f/5.0-6.3 + 1.4x” with the teleconverter. Either way no “IS”. The 150-400 comes closer: after fixing some reports “Olympus M.Zuiko Digital 150-400mm f/4.5 TC1.25x IS Pro”, this time with “PRO” in mixed case and a different way to identify the teleconverter. I need to think about how to handle all that.
And then there's the question of permissions. The database is on eureka, but I want to access it on hydra. Install MySQL, of course. And gain access. Dammit, it's been years since I did that. For reference:
grant all privileges on photos.* to 'grog'@'hydra.lemis.com';
flush privileges;
The important thing is that hydra by itself isn't enough (why?). It must be the FQDN hydra.lemis.com, though I suppose I could use % instead if I trusted my firewall enough.
|
WTP key
|
Topic: animals | Link here |
In the mail today: a key to the Western Treatment Plant. That means that when I'm there at the end of the month, I don't have to stick to following Aaron.
| Thursday, 9 July 2026 | Dereel | Images for 9 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More database filling
|
Topic: technology, photography | Link here |
Spent most of the day setting up my lens database. It's slow, repetitive business, but I'm gradually making some progress. And when I'm done, I'll have to add others, at least
|
Aggressive scrapers
|
Topic: technology, opinion | Link here |
I'm still spending lots of time policing my web sites, though lately things have been much quieter. But I've been keeping a web presence for round 30 years, and in that time my own web pages run into problems, especially internal links. Once I kept my photos in a directory public_html/Images/, but that's clearly not scalable: I now have a little over a million entries. So they're now in a directory relating to the date, like ~grog/20260709 for today.
But what about references to the old locations? I have code in my 404 document to identify broken internal links, and when found it sends me an email. Some are clear:
86 N + 15-04-2026 To groggyhimself@lemis.c World Wide Web Owner ( 6) N + FAILURE: /grog/water.html <- http://www.lemis.com/grog/diary-mar2006.html
This refers to a page that was once called /grog/water.html, but now it has gone down a level to /grog/Albums/water.php. The change from .html to .php is not important—the 404 document corrects for that—but the place in the hierarchy is.
But lately I've been getting far too many messages like this:
10 N + 09-07-2026 To groggyhimself@lemis.c World Wide Web Owner ( 6) N + FAILURE: /grog/Park-Street-9.jpeg <- http://www.lemis.com/grog/diary-nov2024.php
11 N + 09-07-2026 To groggyhimself@lemis.c World Wide Web Owner ( 6) N + FAILURE: /grog/Washing-machine-6.jpeg <- http://www.lemis.com/grog/diary-nov2024.php
12 N + 09-07-2026 To groggyhimself@lemis.c World Wide Web Owner ( 6) N + FAILURE: /grog/Clematis-5.jpeg <- http://www.lemis.com/grog/diary-nov2024.php
None of these were ever in my home directory. Can I have got things wrong so often? No. None of these references exist in the file.
But then there's something like this:
144 N + 21-04-2026 To groggyhimself@lemis.c World Wide Web Owner ( 6) N + FAILURE: /Magia-im-Paso.jpeg <- http://www.lemis.com/yvonne/Magia-photos.php
That's an ancient page, but it contains what I believe to be correct:
<img src="Magia-im-Paso.jpeg">
And the crawler has just moved it up a level.
And finally, there are references that don't seem to be due to links, like
85 N + 09-07-2026 To groggyhimself@lemis.c World Wide Web Owner ( 6) N + FAILURE: /grog/Day/20220207/3D%22https://groups.google.c= <- http://www.lemis.com/grog/Day/20220207/gmail-rant.txt
I don't need to follow the link: this file has a name ending in .txt, which I only use for plain text. So somehow the crawler has invaded my plain text files and guessed at URLs in it.
What a pain!
| Friday, 10 July 2026 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
DigiDirect hits back
|
Topic: technology, photography, opinion | Link here |
Mail to my DigiDirect email address today, something like irateuser@lemis.com. Have they updated things? No, still no mention of my order, just
From: Uphold <noreply@uphold.com>
To: irateuser@lemis.com
Subject: Your Uphold account is waiting for you
Spam! Sent to an account that I had only divulged to DigiDirect! That's not the first time. I had a note in my virtual file that it had happened before. OK, a complete justification for my “one email address per company” approach. Delete it. If I ever want to buy something again from DigiDirect, it's clear that continuity is not important. I can just create a new email address.
|
Lens data import: all done bar the shouting
|
Topic: technology, photography, opinion | Link here |
More fun playing with my lens data database today. I'm pretty much done. And in the process I discovered that the Wikipedia page was incomplete: at least my 7Artisans 4 mm f/2.8 fisheye lens was not on the page. Added that, of course, in the processing realizing how painful this “simplified” Wikipedia markup is. It made the MySQL script seem easy.
What next? I need a table editor. MySQL Workbench might be a good choice. Another can of worms.
|
Null bug fix
|
Topic: technology, opinion | Link here |
I have a bug report that has been open for 2½ years: “makefs command creates UFS fs image with bad summary info when size option (-s) is specified”. And for various reasons I haven't progressed much.
But now Kirk McKusick has taken a look and come up with the obvious workaround:
It is still using layout styles that are at least a decade out of date. Rather than trying to bring all that code up to date, it should just fork off a mkfs to build the filesystem as in that way it will keep up with ongoing work to harden and improve UFS. It would require minimal work to adapt mkfs to writing a disk image to a file rather than to a disk partition.
OK, take a look at that. Oh! mkfs already can create a file system on a file rather than on a special device!
=== root@dereel (/dev/pts/4) /hydra/home/grog/work/newfs 49 -> ./newfs ~/foo
/root/foo: 0.5MB (1024 sectors) block size 32768, fragment size 4096
using 1 cylinder groups of 0.50MB, 16 blks, 128 inodes.
with soft updates
super-block backups (for fsck_ffs -b #) at:
192
But it's not quite kosher:
=== root@dereel (/dev/pts/4) /hydra/home/grog/work/newfs 51 -> fsck -y ~/foo
/root/foo is not a char device
/root/foo IS NOT A DISK DEVICE
CONTINUE? yes
** /root/foo
** Last Mounted on
** Phase 1 - Check Blocks and Sizes
** Phase 2 - Check Pathnames
** Phase 3 - Check Connectivity
** Phase 4 - Check Reference Counts
** Phase 5 - Check Cyl groups
BLK(S) MISSING IN BIT MAPS
SALVAGE? yes
2 files, 2 used, 77 free (21 frags, 7 blocks, 26.6% fragmentation)
***** FILE SYSTEM IS CLEAN *****
***** FILE SYSTEM WAS MODIFIED *****=== root@dereel (/dev/pts/4) /hydra/home/grog/work/newfs 53 -> mkdir /mnt2=== root@dereel (/dev/pts/4) /hydra/home/grog/work/newfs 54 -> mount ~/foo /mnt2
mount: /root/foo: Block device required=== root@dereel (/dev/pts/4) /hydra/home/grog/work/newfs 55 -> mdconfig -a -t vnode -f ~/foo
md0=== root@dereel (/dev/pts/4) /hydra/home/grog/work/newfs 57 -> mount /dev/md0 /mnt2/=== root@dereel (/dev/pts/4) /hydra/home/grog/work/newfs 58 -> l /mnt2
total 1
drwxrwxr-x 2 root operator 512 10 Jul 15:54 .snap=== root@dereel (/dev/pts/4) /hydra/home/grog/work/newfs 59 -> df /mnt2
Filesystem 1048576-blocks Used Avail Capacity Mounted on
/dev/md0 0 0 0 3% /mnt2=== root@dereel (/dev/pts/4) /hydra/home/grog/work/newfs 60 -> echo foo > /mnt2/bar=== root@dereel (/dev/pts/4) /hydra/home/grog/work/newfs 61 -> df /mnt2
Filesystem 1048576-blocks Used Avail Capacity Mounted on
/dev/md0 0 0 0 4% /mnt2=== root@dereel (/dev/pts/4) /hydra/home/grog/work/newfs 62 -> l /mnt2
total 1
drwxrwxr-x 2 root operator 512 10 Jul 15:54 .snap
-rw-r--r-- 1 root wheel 4 10 Jul 16:01 bar=== root@dereel (/dev/pts/4) /hydra/home/grog/work/newfs 63 -> cat /mnt2/bar
foo
It looks as if some information didn't get written. That should be easy enough to fix. I wonder if it would make sense to be able to mount files directly as well.
| Saturday, 11 July 2026 | Dereel | Images for 11 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Photo experiments
|
Topic: photography, technology, opinion | Link here |
I hadn't planned to take any photos today apart from the weekly house photos, but when I saw some kangaroos in the back yard, I dragged out my M.Zuiko Digital 150-600 mm, more as an experiment than anything else.
It worked. I now have a better understanding of the limitations than before. This photo could have been quite good:
|
|||||||||||||||||||||||
What's wrong with it? Lots, of course:
What was my autofocus setting? Not (currently) recorded. It's in the MakerNotes, and somewhere I have seen a decoding. I must add it to my Exif data function. I have a suspicion that it might have been set to “birds” for at least some of the photos.
| Sunday, 12 July 2026 | Dereel | Images for 12 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Better photos?
|
Topic: photography, animals, opinion | Link here |
Yesterday's kangaroo photos were really not spectacular. How about a better choice in better weather?
|
|||||||||||||||||||||||
That looks better, but is it really sharp? Maybe it's not such a good idea to take photos like this at full aperture. And is the blur on the fence line due to focus, camera shake or lens problems? The bark of the tree in the background at the left looks better, but still not good. And there's chromatic aberration that could be corrected in postprocessing if I had software with the right profiles. There, too, I should make comparisons. DxO PhotoLab is pricing itself increasingly out of the market, even for people like me who have purchased multiple licenses.
|
More weeding
|
Topic: gardening | Link here |
Jesse Walsh around to weed the garden today. It was necessary: it took him 4 hours.
|
Identifying focus mode
|
Topic: technology, opinion | Link here |
As planned, modified my PHP scripts to include focus mode in the Exif information. That proved to be surprisingly uncomplicated. But performing the changes wasn't. Something has changed my Emacs php-mode, and now I get // as comment delimiter instead of /* ... */. And the comment column has changed. This is presumably because of a version update, but it shouldn't change what I have set, and I wasn't able to find a way to fix it. In the end, after much searching and cursing, went back to my 11 year old eureka, where things still worked.
Why does this have to be?
|
Sensor readout times
|
Topic: photography, technology, opinion | Link here |
Today I saw an article about the best modern “mirrorless” cameras, “full frame” of course. An interesting detail was the sensor readout time, ranging from 37 ms for the Panasonic S1R II to exactly 10 times as fast for the Nikon Z8.
That's an issue that has interested me in the past. There's a big difference between the readout speed of the Olympus OM-D E-M1 Mark I and the Olympus OM-D E-M1 Mark II, as demonstrated by these rolling shutter experiments of a rotating fan:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And that's related to the sensor readout speed So what information do we have? Nobody registers the sensor readout time, but it's directly related to the maximum shutter speed for electronic flash, which is 1/13 s for the Mark I and 1/60 s for the Mark II. That means a readout time of under 77 ms for the Mark I and under 17 ms for the Mark II. How much under? Hard to say.
I haven't measured it for my current OM System OM-1 Mark II, but my guess is that it's faster: it also relates to the maximum frame rate for “continuous shooting”. And that's something that DPReview has on its comparison page .
Oh. They have “upgraded” their web site. Apart from this horrible reverse video, something that is annoying me more and more, they have removed all that information! Instead there are commercial things there that are of no interest whatsoever in this context. What a pity. I think they have just put the last nail in DPReview's coffin. And for me it means that I will no longer link to them, at least not until they fix the site.
More searching brought me to this page. Not as good, but at least I get some information. The maximum rate on the OM-1 Mark II is 120 frames per second, and on the Z8 it's 30 frames per second. That's interesting for a couple of reasons: yes, the shutter speed and the readout are sequential, but I'm sure that in each case they assume a high shutter speed (say 1/8000 s), which adds up to a total time of 3.75 ms for the Nikon or 15 ms for the OM-1. That leaves the rest of the time for the readout, which gives a readout time of less than 8.3 ms for the OM-1 or 30 ms for the Z8. I'm happy to believe that the real time is less, but why the discrepancy?
| Monday, 13 July 2026 | Dereel | Images for 13 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Scales pain
|
Topic: food and drink, technology, opinion | Link here |
Time to make a batch of Penang laksa today. Nothing unusual: take the paste, tamarind and petis udang, add water to 1600 g.
But things aren't that simple. First, I didn't have enough petis udang. I needed 40 g, but I only had 30. Use that anyway? No, I only needed one portion of laksa today, so 8 g would have done it. But for the fun of it I tried without, and in fact it didn't make that much difference.
The real problem is that my scales are very inaccurate („vage“ in German), and I know that to get 1600 g in my 578 g pot I have to weigh 2200 g, and not the expected 2178 g. And after bringing to the boil and measuring again, the displayed weight had dropped to 2167 g.
OK, measure out 5 portions of 320 g each. But after measuring 4, there was only 296 g left! That's a total inaccuracy of 46 g (2.7%)!
That's enough. Buy some new scales, braving eBay to do so. Scales measuring up to 5 kg in 0.1 g increments? Probably the last digit is inaccurate, but that's more than enough. Only $9.99! OK, follow to the detail page and have to select a “color”. The scales I was looking at cost $56.49! Only a previously unmentioned cleaning cloth costs $9.99 (a hefty price in itself). When is eBay finally going to do something about this? This is just plain misleading. Advertise scales, offer a cleaning cloth. At the very least some kind of downvote should be possible, and eBay should remove such misleading offers. For me it's a reason to avoid the sellers zeeah85 and Exotic-Treasures81.
Much searching. If I had been working at the rates I used to charge, I could have paid for a number of different scales. As it was, I finally agreed on a “postal” scale weighing up to 40 kg in 1 g increments, along with useless (to me) things like price aggregation, for $32.16:
|
|||||||||||||||||||||||
That's a bit over the top, but still cheaper than the other one I saw, and my hope is that the “postal” would signify a certain accuracy. We'll find out.
In the meantime, I choose PayPal every time because I don't trust eBay, who always recommend a different payment method. Today, for the first time in a long time, PayPal didn't recognize me on my “device” (teevee.lemis.com) and wanted to send me a confirmatory email. OK, fine.
But it didn't come. Checking my mail logs, found:
Jul 13 18:46:57 eureka postfix/smtpd[75599]: NOQUEUE: reject: RCPT from www.lemis.com[45.32.70.18]: 550 5.1.1 <pally@lemis.com>: Recipient address rejected: User unknown in local recipient table; from=<service@paypal.com.au> to=<pally@lemis.com> proto=ESMTP helo=<lax.lemis.com>
pally@lemis.com? That's an old, worn-out email address. Where did they get that from?
|
Landscape change
|
Topic: animals, general | Link here |
The new owners of the property on the corner of Bliss Road and Stones Road (the “schoolyard”, since there really was a school there in the 1880s) have done a lot of work on the property. One just occurred to us today while walking the dogs: the garden arrangement that we have admired has changed. Here last October and now:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Part of the difference is that last October some of the bushes were flowering, but more to the point, all the bushes under the central tree and a number in the foreground have been removed.
|
More Emacs pain
|
Topic: technology, opinion | Link here |
I have a number of Emacs macros that I use to write this diary. One of them creates the framework for an article:
<?php pubdate ("2026-07-14T02:26:23+00:00"); ?>
<?php texttopic ("c", ""); ?>
<p>
</p>
<?php endtopic (); ?>
But at random, the indentation is incorrect. Why? I can't find out, but I have discovered that it only happens with the latest and greatest Emacs on hydra.lemis.com, while on eureka it works correctly. Why am I having so many problems with software upgrades?
|
ALDImobile gangs up on us
|
Topic: technology, general, opinion | Link here |
Yvonne's pre-paid phone tariff needs topping up. Not a problem: go to http://www.aldimobile.com.au/, select the correct phone and “Recharge”. Select the tariff and send.
Submitting order...
And it stayed that way all evening, despite a few retries an hour or so apart. Do I have to call them on the phone? That way madness lies.
| Tuesday, 14 July 2026 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Firefox: enough!
|
Topic: technology, opinion | Link here |
Tried once again to recharge Yvonne's phone today. Once again it hung submitting the order. OK, try with chromium instead of firefox. Slow (almost certainly the site), but it worked.
This isn't the first time that I've had problems like that with firefox. I have hated it since I first used it, over 20 years ago. It shows other issues, like random loss of page navigation and misplaced button sensitivity areas. Why haven't I got rid of it long ago?
The answer is simple: I can't get chromium, the obvious alternative, to work with multiple instances. Time to ask the twins, who should have a vested interest. They came up with a number of suggestions, some contradictory, but ultimately I was able to start multiple instances. I'm still trying to get my head around it, but in essence it seems that I need to remove the singletons and use separate configuration directories for each instance. firefox does that too, but it doesn't require as much fiddling.
Done? No, I also need to invoke the browser from Mutt. And once again it's a can of worms. The twins tell me to put this in ~/.muttrc (after substitutions for my environment):
text/html; chrome --display=:0.2 --user-data-dir=~/.config/chromium-2 %s; test=test -n "$DISPLAY"; nametemplate=%s.html
Part of that (the nametemplate) is to work round breakage in chromium: it ignores the content of the file it's displaying and goes by the “extension”, which must be .html. By contrast, with firefox, my .mailcap contains
text/html; /home/local/bin/dobrowser %s; copiousoutput
OK, dobrowser is a copout, but it's just so that I can access the file elsewhere. It contains:
WEBPAGE=/eureka/home/var/tmp/mutt-firefox/`basename $1`.html
mv $1 $WEBPAGE
/usr/local/bin/firefox $WEBPAGE
But that doesn't matter. Gemini's invocation doesn't work. Instead I kept the dobrowser approach and change the last line to:
chrome --display=:0.2 --user-data-dir=/home/grog/.config/chromium-2 $WEBPAGE
And that seems to work. But what a mess by comparison. Clearly chromium is firmly entrenched in the Microsoft space, even worse than firefox.
|
Warning! Breakin!
|
Topic: technology, opinion | Link here |
Message from eBay today:
We noticed a new sign-in on your account at ebay.com.au. Here are the details:
Time of sign-in
14/07/2026 11:57 AM AEST
Device
Linux (Chrome 143.0)
Approximate location
Sydney, New South Wales, Australia
If this was you, there’s nothing you need to do. If it wasn’t, please change your password right away.
Now isn't that accurate and helpful? The “device” hadn't changed, just the browser, located 1,000 km to the south-west of Sydney, in a different state, and running on a different operating system. If they're going to warn, why can't the finally report correctly and with more details (IP address comes to mind)?
|
Clue to Emacs indentation problems
|
Topic: technology, opinion | Link here |
Reading my diary for 20 years ago, found a reference to my Emacs indentation macros. It's not clear that it's what I was using before the last update, but the clue might be in where I kept it. It still works on eureka, so I haven't lost it.
| Wednesday, 15 July 2026 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
A day spent fighting chrom*
|
Topic: technology, opinion | Link here |
My transition from firefox to Chromium continues. It's almost working, but fixing the remaining issues took up most of the day.
What's the problem? Somehow the name is symptomatic. What's it called? It seems that though the name is Chromium, it calls itself chrome, really the name of a different product. And it's firmly entrenched in the Microsoft mentality. I spent a lot of time trying to get Mutt to display on a chrom* window. Yes, I had done it yesterday, but it was less than ideal. It creates a new window rather than a new tab. I can live with that, but it holds on to the mutt session until it is done, and if I press ^C on the mutt window, the chrom* window closes. And in the process it creates an amazing amount of vomit on the mutt window, including:
[-- Autoview using /home/local/bin/dobrowser '/home/grog/mutt-hydra-1004-41029-12055955194787228941' --]
[-- Autoview stderr of /home/local/bin/dobrowser '/home/grog/mutt-hydra-1004-41029-12055955194787228941' --]
[42405:66144074215440:0715/133857.553950:ERROR:dbus/bus.cc:406] Failed to connect to the bus: Could not parse server address: Unknown address type (examples of valid types are "tcp" and on UNIX "unix")
[42405:66144074133520:0715/133857.598521:ERROR:dbus/object_proxy.cc:573] Failed to call method: org.freedesktop.DBus.NameHasOwner: object_path=/org/freedesktop/DBus: unknown error type:
[42405:66144074133520:0715/133857.607001:ERROR:dbus/object_proxy.cc:573] Failed to call method: org.freedesktop.DBus.Properties.GetAll: object_path= /org/freedesktop/UPower/devices/DisplayDevice: org.freedesktop.DBus.Error.ServiceUnknown: The name org.freedesktop.UPower was not
[42405:66144074223632:0715/133900.384970:ERROR:google_apis/gcm/engine/registration_request.cc:292] Registration response error message: DEPRECATED_ENDPOINT
OK, that's simple. Background it:
chrome --new-tab --display=:0.2 --user-data-dir=/home/grog/.config/chromium-2 $WEBPAGE &
But it didn't work. mutt was still suspended! How could that happen? Much more interaction with Google Gemini, receiving suggestions which were mostly wrong and included things like the /proc file system and interaction with systemd, which I thought was not universal even on Linux. The invalid suggestions were still interesting for the approaches that I wouldn't have considered. And somehow it's far too cosy with D-Bus and other strange things.
OK, how about the man page?
=== grog@hydra (/dev/pts/39) ~ 163 -> man chromium
No manual entry for "chromium"=== grog@hydra (/dev/pts/39) ~ 164 -> man chrome
@@PACKAGE(1) USER COMMANDS @@PACKAGE(1)
NAME
@@PACKAGE - the web browser from Google
SYNOPSIS
@@PACKAGE [OPTION] [PATH|URL]
DESCRIPTION
See the Google Chrome help center for help on using the browser.
<https://support.google.com/chrome/>
OPTIONS
@@MENUNAME has hundreds of undocumented command-line flags that are added
Now doesn't that give the impression of a polished, complete product? That's especially the case because, of all the hundreds of flags, it describes exactly nine, including --version. But the twins have help available:
It is widely used by developers because it stays continuously up to date and lists thousands of available switches, along with explanations and their origin files in the Chromium codebase
It tells me:
These are the 1,349 command line switches and 3,086 feature flags defined in Chromium as of July 15, 2026.
But it only showed four switches! Clearly others are hidden in a non-intuitive manner, but I am clearly not modern enough to understand.
In the end, I gave up. The only improvement was invoking a daemon to help me:
daemon -f chrome --new-tab --display=:0.2 --user-data-dir=/home/grog/.config/chromium-2 $WEBPAGE &
After that, things almost worked. I couldn't get it to attach to an open window. Well, maybe not. The first display always opened a new window, but after daemonizing the instance, subsequent invocations opened a new tab in that window.
|
What to do with makefs?
|
Topic: technology, opinion | Link here |
I still have this open bug report about makefs, a program that builds a file system in a file and then populates it. The delay is at least partially due to laziness, but the more I think about it, the more I wonder what use it is.
We already have a program called newfs that does the first half of the task on a device, and last week Kirk McKusick pointed that out. But what neither of us knew was that newfs will happily build a file system in a file. I tried it out then and ran into minor problems, like incorrect summary information even after fsck:
=== root@dereel (/dev/pts/5) ~ 53 -> truncate -s 512k foo=== root@dereel (/dev/pts/5) ~ 56 -> /hydra/home/grog/work/newfs/newfs.full ~/foo
/root/foo: 0.5MB (1024 sectors) block size 32768, fragment size 4096
using 1 cylinder groups of 0.50MB, 16 blks, 128 inodes.
with soft updates
super-block backups (for fsck_ffs -b #) at:
192=== root@dereel (/dev/pts/5) ~ 57 -> mdconfig -a -t vnode -f ~/foo
md0=== root@dereel (/dev/pts/5) ~ 59 -> mount /dev/md0 /mnt2=== root@dereel (/dev/pts/5) ~ 61 -> df /mnt2
Filesystem 1048576-blocks Used Avail Capacity Mounted on
/dev/md0 0 0 0 3% /mnt2
Why are all the block counts 0? Because I said so (via the BLOCKSIZE environmental variable): I'm counting in megabytes, and this file system is smaller than one megabyte. Count in kilobytes and all seems well:
=== root@dereel (/dev/pts/5) ~ 62 -> df -k /mnt2
Filesystem 1024-blocks Used Avail Capacity Mounted on
/dev/md0 316 8 284 3% /mnt2=== root@dereel (/dev/pts/5) ~ 63 -> ls -l /mnt2
total 1
drwxrwxr-x 2 root operator 512 16 Jul 12:42 .snap
So what do we need? The commands above are sufficient to create the file system. Then the data needs to be copied; we have plenty of commands for that. And then destroy the vnode device. A simple shell script should be enough. But maybe I'm missing something: there are a surprising number of reports open about makefs. Maybe I should first find out why before removing it from the tree.
| Thursday, 16 July 2026 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Chrom*: enough
|
Topic: technology, opinion | Link here |
Spent some time tidying up my transition from firefox to chromium today. It wasn't a silver bullet. All browsers seem to have great problems with text size, and I was expecting some. But chrom* seems to be even worse. I can't find a way to set a minimum text size, only sliders to set “font size” in steps of what it claims to be 2 point. In fact, the text displays is roughly half the size that it claims—clearly it doesn't take the display hardware into account. And somehow the difference in size between 22 “point” and 24 “point” is enormous. And on occasion, apparently at random, it displays text in a new tab far too small or far too large. I can fix it with the c-scroll wheel, but why should I need to?
And then there's something I've seen before: it frequently repositions the window so that it goes below the screen. Why? This seems to be a bug, maybe related to the FreeBSD port. Sometimes I wonder whether I shouldn't go back to firefox.
|
Bruno bites!
|
Topic: animals | Link here |
Phone call from Loretta Sayers today. Her daughter Remy was hospitalized for two nights after Bruno bit her!
Does she want to return Bruno? We had left it open for the first two months, and he has been gone for less than half that. But no, and apart from this one incident he's loving and affectionate. Remy still loves him and would be devastated if he were to go. But we're really puzzled: why would he do a thing like that? The only occasion that I have ever experienced a serious cat bite was decades ago while carrying an unknown cat at a cat show, and she panicked. But that's not Bruno; he knows them, and he's not the panicky type. We're still baffled.
|
Photos of Kuala Lumpur in the 1960s
|
Topic: photography, history | Link here |
For some reason, I asked Google Gemini to “Show me some sites with photos of Kuala Lumpur in the 1960s”. It showed only one photo:
|
||||||||||||||||||||||||
Jalan Tuanku Abdul Rahman (formerly Batu Road) in 1964.. Source: Saigoneer
I know that photo: I took it on 14 September 1964. And it's not Batu Road at all; it's Mountbatten Road (now Jalan Tun Perak). Where is it displayed? Saigoneer, which as it sounds is from Vietnam. And that page contains two further photos I took:
|
||||||||||||||||||||||||
|
|||||||||||||||||||||||
The first was taken from the same place on the IBM building, presumably only a few seconds apart. The second, of the railway station and flyover, was taken nearly a year later with my new Pentax SV.
So what do I do? One option is “nothing”. The twins even drafted a letter asking them to add credits. But it's not as simple as that: at the bottom of the page is a text saying that they got them from Says, and that page contains the text
The photos have been credited to the site that we found them from, R.S Murthi, as the names of the photographers could not be found. It would be great if you could drop us a comment should you happen to know the names of any of the photographers.
So maybe I should do something.
| Friday, 17 July 2026 | Dereel | Images for 17 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Experimenting with Phat Si-io
|
Topic: food and drink, opinion | Link here |
I rather like phat si-io, but it's very difficult to get right. But the real thing is the sauce. How about cheating? Today I made a couple of changes: a carrot (65 g) instead of cabbage (because that's what I had), normal rice noodles instead of kway teow, and no attempt at searing. The result:
|
|||||||||||||||||||||||
And how about that, it tasted pretty much the same, and much less work. I was surprised how the carrot seemed to disappear.
|
New kitchen scales
|
Topic: food and drink, general, opinion | Link here |
Various emails, contradictory as ever, from eBay today about the scales that I ordered on Monday. And then SMS from the courier company delivering them. They're really coming today! More SMS. Different time, 17:00-19:00 PM. More SMS. Different time, 18:41 to 20:41 PM, later than I would have liked. More SMS. Different time, 16:06 to 18:06 PM. That's (marginally) better. And then an email from eBay, dated Thu, 16 Jul 2026 20:46:33 -0700 (MST). Your item has been delivered.
Believe that who may. What time was that, anyway? The message arrived at 13:46:51, so that would be the time. Out to look in front of the front door. Nothing, of course. But then I saw:
|
|||||||||||||||||||||||
|
|
|||||||||
What's it doing there? Clearly the courier was too lazy to bring it to the door, but he put it at the wrong end of the gate. He could only have dropped it there: the pedestrian entrance is on the left. And yes, he confirmed that he put it there with an MMS:
|
|||||||||||||||||||||||
For good measure, he put it there upside-down. What a delivery! What if it had rained? Is that allowed? Checking AusPost (not the courier), I read:
Safe Drop is only available for locations not in public view, not exposed to the weather and which we regard as safe.
And that's clearly not the case. It was in public view and exposed to the weather. As it happened, the weather was clear and I found it only 5 minutes after drop-off.
But was it damaged? Opened it and found:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That's the battery cover, and it's clearly damaged. That's a mixed blessing: the listing doesn't mention the battery, but it's there, and it works, though I need to check the condition: I can't change the cover again. On the other hand, it can't be the result of the package being dropped: to close it I first need to remove the screws. So it's very likely that the item was manufactured like that.
How do I assemble it? By guesswork, mainly: the instructions were an above-average catastrophe: a single sheet of A4 paper. There were two sets of plastic spacers. One was clearly for the feet, and the others (5 items) to hold the tray into position. But the tray only has four pins to attach them, and the spacers don't fit into the hole in the beam:
|
|||||||||||||||||||||||
But it seems that they just fit loosely into the holes.
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And is it accurate? Found a 100 g weight, rather less than I needed, and compared all three scales that I have. The 1000 × 0.1 g scales read 100.2, then reconsidered and went to 100. The old, inaccurate scales (10 kg × 1 g) showed 100 g. And the new one showed 100 g. The real test will be at higher load levels, but so far, modulo adjustments, all seems well.
Can I find any better instructions? The device had a model number on the top. What do the twins have to say? No instructions themselves, but some interesting details, in particular about hidden codes (including how to change from metric to archaic units), and more interestingly about calibration. So there is hope that they'll be more accurate than the ones they're replacing.
|
Free denoise software!
|
Topic: photography, opinion | Link here |
Email today from ON1: ON1 NoNoise AI free! OK, I've tried ON1 before without being overly impressed, but at that price I'll take 2.
Oh. I already have two:
|
|
|||||||||
And the dates of the first two point to this test, where I found it more than wanting. Here my attempts and then ON1:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Run the cursor over an image to compare it with its neighbour. To be clear: the ON1 results are the second image. That's sad.
|
Trigger finger
|
Topic: health | Link here |
For some time now Yvonne has been suffering from trigger finger, where a finger tightens up and won't let go. It seems to have been getting worse, and she's planning to see Paul Smith, our GP, on Wednesday.
But today it tightened up so much that she couldn't loosen it. It came “good” after about 10 minutes. And she had this idea of putting on a “splint”:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Off to an emergency clinic tomorrow?
| Saturday, 18 July 2026 | Dereel | Images for 18 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Releasing the trigger finger
|
Topic: health | Link here |
Yvonne off to town today to visit the UFS Urgent Care Clinic about her trigger finger. Things weren't as good as when I was there last year, and she was quite upset by the perceived incompetence of the staff. In the end she left and bought a sheath at the chemists:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That seems to have done the job:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
| Sunday, 19 July 2026 | Dereel | Images for 19 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Julienne cutter
|
Topic: food and drink, opinion | Link here |
A few days ago I received advertising from Amazon offering an interesting kitchen utensil for the special sale price of only $19 or so: a cutter for automatically making Julienne strips.
OK, Amazon, I don't really trust you. What does eBay have to offer? A very similar device for only $7.99:
OK, that was worth the try. It took its time to come, and I've been playing with it for a few days. It has two sides, for very fine and medium coarse. “Very fine” didn't work at all for me on the carrot that I tried: it only scraped off the surface. The other side did up to a point:
|
|||||||||||||||||||||||
But that photo also shows the parts of the carrot that it couldn't cut, and the slices aren't as uniform as I had hoped. And it left the cutter very clogged:
|
|||||||||||||||||||||||
How do you get rid of that? The blade is sharp, of course. Scraped off what I could and put it in the dishwasher. The result:
|
|||||||||||||||||||||||
That's clearly not enough. After another round through the dishwasher, it was almost acceptable:
|
|||||||||||||||||||||||
|
|
|||||||||
|
|
|||||||||
So: carrots aren't spectacular. How about cucumbers? After a couple of attempts, got:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Not exactly what I was looking for. But with a little more effort, ended up with this:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That's almost acceptable. That's what I did with a knife. Unless I'm missing something important, the cutter isn't worth it. It wasn't until checking this article that I noticed that the device in the advertisement is different from what I got. Does that make a difference?
|
Trigger finger workaround
|
Topic: health | Link here |
For some reason Yvonne found her trigger finger splint uncomfortable. Instead she found a piece of plastic water hose:
Now why didn't I think of that? Why, didn't I think of that? It was my first suggestion when her current problems started. Now to see whether a slightly larger diameter might not be better.
|
Video downloads
|
Topic: multimedia, technology, opinion | Link here |
I've been downloading some seasons of „In aller Freundschaft“ from the ZDF. In combination with youtube-dl it's pretty straightforward:
=== grog@tiwi (/dev/pts/15) /spool/Series/In-aller-Freundschaft/27 107 -> youtube-dlp https://www.zdf.de/serien/collection-index-page-ard-collection-ard-dxjuomfyzdpzag93omzizty3nwuwmda0nwzhowm-760
And that downloads most, if not all, of the episodes on the page. It takes a while, because it's sequential, but I have time.
But today I took a look and found that it loaded even more than that. I don't know how, but it had downloaded several years' worth, a total of over 500 GB! I don't think I have ever seen as much data go across the net. 35 years ago, if it had been technically possible, it would have cost me in the order of $17,500,000. Now it's included in my monthly fee.
| Monday, 20 July 2026 | Dereel | Images for 20 July 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Replacing named
|
Topic: technology, opinion | Link here |
Once again problems with named on lax.lemis.com. Why do I have so much trouble? I got it working again, but in the meantime there were significant delays. Should I use something else? Off to ask the twins, which came up with suggestions, depending on whether I wanted an authoritative or a recursive name server.
Huh? Both. Or do I? Basically it should be authoritative, but will that serve my network? More to think about.
|
System down, won't come up, customer screaming for blood
|
Topic: technology | Link here |
At Tandem, we had three categories of support requests (“User perceived severity”): 1 was the least serious, 3 the most. Internally we had descriptions, most of which I have forgotten. But 3 was “System down, won't come up, customer screaming for blood”.
Today I had this as a customer. Vultr had a service incident in their Los Angeles centre which required modifying their host software, and so all guests had to be rebooted, including lax.lemis.com. I didn't notice until I found it was down. What went wrong this time? Off to take a look. Oh:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
I've seen that before. In fact, the images are from last month, but the boot stopped in exactly the same place. We're dead in the water!
Sent off a fault report to Vultr, of course, and then considered: fra can take much of the web load, but all my mail goes via lax (calling itself mail.lemis.com or mx1.lemis.com in this connection). OK, let's open the firewall and let eureka (mx0.lemis.com) receive the mail directly.
ipfw add 11 allow ip from any to eureka dst-port 25
But it didn't work. Why? Try allowing to any system. Still nothing. Do some network tracing. Nothing obvious. nmap? Something obvious: hydra (from which I was doing the testing) spontaneously rebooted! Horror!
When hydra came back, it was not good. No network connectivity. I didn't save the changes I made last month, so it was trying to access the net via eureka rather than interface re1. Run dhclient against re1 and I had a connection, but no routing: default gateway was still eureka. Remove that. Where do I point it? The answer in such a case is simple, as I now know: remove the default interface and then start dhclient, which will set the correct default route.
Things weren't over yet. Once again I had this horrible window manager restart syndrome, which I must finally fixed. But only on server 0. Server 1 worked normally, and I was able to establish that eureka still has a firewall, and it allowed mail only from lax. Change that and all was well.
That's enough for one day. The shock was palpable: I was trembling from the stress. Mañana.
| Tuesday, 21 July 2026 | Dereel → Sebastopol → Dereel | Images for 21 July 2026 |
| Top of page | ||
| previous day |
|
Recovering after disaster
|
Topic: technology, opinion | Link here |
After yesterday's disaster, I was worried about further problems with hydra this morning. Why did it crash in the first place? All I did was run nmap—if, indeed, that was the cause. OK, start the firewall, carefully from a vty. It hung! Why? Once again I had to reboot!
And why can't I start X server 0? Once again incomplete configuration changes, not quite the kind one would expect. It was once:
$FVWM -display $DISPLAY -f $X/.fvwm/fvwm3rc-$me$DISPLAY.0 &
$FVWM -display $DISPLAY.1 -f $X/.fvwm/fvwm3rc-$me$DISPLAY.1 &
$FVWM -display $DISPLAY.2 -f $X/.fvwm/fvwm3rc-$me$DISPLAY.2 &
$FVWM -display $DISPLAY.3 -f $X/.fvwm/fvwm3rc-$me$DISPLAY.3 &
DISPLAY=$DISPLAY xclock -digital -strftime "%A, %e %B %Y, %T" -geometry 400x29+1220-14 -update 1 &
DISPLAY=$DISPLAY.1 xclock -digital -strftime "%A, %e %B %Y, %T" -geometry 400x29+1220-14 -update 1
I had changed it to use a single window manager, and I had commented out the second xclock:
$FVWM -display $DISPLAY -f $X/.fvwm/fvwm2rc-$me$DISPLAY &
DISPLAY=$DISPLAY xclock -digital -strftime "%A, %e %B %Y, %T" -geometry 400x29+1220-14 -update 1 &
# DISPLAY=$DISPLAY.1 xclock -digital -strftime "%A, %e %B %Y, %T" -geometry 400x29+1220-14 -update 1
Both those changes were wrong: the new configuration file didn't exist, and by commenting out the second xclock, the script flow continued, in a loop. Only the last command in the flow needs to be without the all-important & at the end.
This really does my health no good. OK, for the time live without a firewall, but clearly something needs to be done, not just here but also on Los Angeles. How can I load the ipfw module without it automatically blocking everything?
Another question to Google Gemini. How about that, you can set it via a sysctl or kenv:
kenv net.inet.ip.fw.default_to_accept=1
kldload ipfw
Tried it out on dereel.lemis.com. It works! Why isn't this the default?
Will it work on lax? If not, will I be able to reboot it? Made a snapshot and created a new instance in Melbourne. Yes, it works there too. Whatever Vultr did seems to have remained.
But what happened? What did Vultr change? They responded very quickly yesterday, within half an hour, but my question this morning went unanswered until this evening. Then I received a message: “Please review when available and let us know if further assistance is needed.”. It seems that the responder didn't even read my message. I wonder if I will get a response.
|
Into town again
|
Topic: general | Link here |
I had planned to do a number of things in town today. Having my hair cut was one that had to be done, but I had also planned a number of other less important things which, under the circumstances, I left behind. But my car had a slow flat tyre on the left rear wheel, and that had to be attended to.
Surprise: damaged sidewall:
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
And that can't be repaired. $130 for a new tyre that will probably never be used in this 19 year old car.
|
Redan wetlands
|
Topic: photography, animals | Link here |
On the way home, took a look at the Redan Wetlands. With a name like that, you'd expect it to be in Redan, but in fact it's in Sebastopol. It's only 270 m straight line from the ALDI shop where we go shopping, but in my 19 years in the area, I had never seen it.
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
I was interested to see if there were any birds, but all I saw (from the vantage point of the photos) were a couple of ducks. I later discovered that there's a web page on the location. And that showed that the crow-like bird I saw mingling with the magpies were grey currawongs (which, for some reason, eBird calls “gray currawongs”). Maybe I should go there again and take a closer look.
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 |
