|
| Greg's diary |
| July 2024 |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Monday, 1 July 2024 | Dereel | Images for 1 July 2024 |
| Top of page | ||
| next day | ||
| last day |
|
Cold weather
|
Topic: general | Link here |
We're going through an unusually cool period of weather at the moment:
mysql> select year(date), min(outside_temp), avg(outside_temp), max(outside_temp)
from observations
where month(date) = 6
group by year(date);
+------------+-------------------+--------------------+-------------------+
| year(date) | min(outside_temp) | avg(outside_temp) | max(outside_temp) |
+------------+-------------------+--------------------+-------------------+
| 2017 | 0.1 | 8.583251664357402 | 17.2 |
| 2018 | -0.2 | 8.587153978918375 | 18.9 |
| 2019 | -0.9 | 9.446074306448931 | 17.6 |
| 2020 | -0.7 | 8.782459793250906 | 16.6 |
| 2021 | 0.8 | 9.77173985397815 | 18.2 |
| 2022 | 0.6 | 8.86966862322499 | 19.6 |
| 2023 | -0.4 | 10.222608372489137 | 20.5 |
| 2024 | -2.4 | 8.394851808112373 | 16.5 |
+------------+-------------------+--------------------+-------------------+
And of course we had to discover that our new Mitsubishi Electric MSZ-AP60VG(D) air conditioner isn't really up to the task. It takes hours in the morning to bring the lounge room up to temperature:
I turned it on shortly after 5:00, and it took until 10:00 to come up to temperature (top line). I'm going to have to leave it running overnight.
|
Lies, damn lies and statistics
|
Topic: technology, general, opinion | Link here |
Last month was cold? How about this month?
mysql> select year(date), min(outside_temp), avg(outside_temp), max(outside_temp)
from observations
where month(date) = 7
group by year(date);
+------------+-------------------+--------------------+-------------------+
| year(date) | min(outside_temp) | avg(outside_temp) | max(outside_temp) |
+------------+-------------------+--------------------+-------------------+
| 2017 | -1.8 | 8.386610996751084 | 17.2 |
| 2018 | -0.5 | 8.84761882739429 | 19.7 |
| 2019 | 1 | 9.319298495442224 | 18.5 |
| 2020 | -0.8 | 8.394708139534231 | 16.7 |
| 2021 | -0.6 | 8.415084340357575 | 17.2 |
| 2022 | -2.6 | 8.445030549610115 | 20.4 |
| 2023 | -0.5 | 9.574544398292241 | 17.9 |
| 2024 | 0.9 | 2.9052224463604865 | 6.8 |
+------------+-------------------+--------------------+-------------------+
The maximum temperature is much lower than the average of any previous year! Of course, that's what you get with 10 hours of overnight data.
|
Constructive translations
|
Topic: language, technology, food and drink, opinion | Link here |
Google Translate is quite useful for reading labels on East Asian food, but it has its limits. Here's a particularly amusing one:
|
|||||||||||||||||||||||
|
|
|||||||||
|
|
|||||||||
Somehow Google Translate is just too eager to please.
|
More garden work
|
Topic: gardening, animals | Link here |
Jesse Walsh along to do some more garden work today, notably the “island” in the driveway:
|
|||||||||||||||||||||||
The Leucospermum had been so covered in Carpobrotus that there's a big bald patch:
|
|||||||||||||||||||||||
At the other end also planted some small Carpobrotus and some nameless plants that Yvonne got from Kelly-ann Nyari:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And while tidying up, he found a particularly bloated frog:
|
|||||||||||||||||||||||
Also planted a long-suffering Buddleja_×_weyeriana in front of the “shade area”:
|
|||||||||||||||||||||||
|
X configuration: RTFM!
|
Topic: technology, opinion | Link here |
More searching for answers to the .Xdefaults issue today. What does the xterm(1) man page say? Much more than it did 30 years ago. Now there are 7,758 lines, about 130 pages. When I learnt X it was 253 lines, less than 5 pages. And much of this new version describes exactly what I'm looking for: a line like xterm-r*faceName: DejaVu Sans Mono refers to a name xterm-r, which is described in X(1):
-nameThis option specifies the name under which resources for the application should be found. This option is useful in shell aliases to distinguish between invocations of an application, without resorting to creating links to alter the executable file name.
And it's used like this:
/usr/local/bin/uxterm -name "xterm-r" -s -sl 2048 -sb -ls -j -rw -display :0.3 -geometry 90x50-53+0 -e /usr/local/bin/bash
And faceName is a resource described in the man page.
So now I have the background to actually do something. Looking through the configuration file, I find that most of these names simply refer to colours. They're relatively arbitrary, but they also describe the location of the icons. Here a “before” and “after” converting to Color
Style "xterm-r" Icon xterm.xpm,Color yellow/black,IconBox -40 300 -1 -140
Style "xterm-r" Icon xterm.xpm,Colorset 21,IconBox -40 300 -1 -140
So now I need to create entries for each of these names. Currently I have xterm-lx xterm-rx, xterm-r, xterm-rsxa and xterm-rsxd. Do I need all of them? Probably not.
And then there are the options passed to the xterm:
/usr/local/bin/uxterm -name "xterm-r" -s -sl 2048 -sb -ls -j -rw -display :0.3 -geometry 90x50-53+0 -e /usr/local/bin/bash
So after all that, my invocation reduces to:
/usr/local/bin/uxterm -name "xterm-r" -display :0.3 -e /usr/local/bin/bash
Do I need to specify the display? Probably not. But there's no resource to specify the shell; the best choice there is the SHELL environment variable. If that works, I can reduce things to
/usr/local/bin/uxterm -name "xterm-r"
So do I need to generate individual files? Yes, unfortunately. I still have different invocations for local and remote machines, which could potentially reduce to something like this on hydra:
+ "eureka" Exec ssh -A eureka /usr/local/bin/uxterm -name "xterm-r" &
+ "hydra" Exec /usr/local/bin/uxterm -name "xterm-r" &
And on eureka it would be the other way round. Still, things look a lot simpler like that. The next step is to decide what names to use, and how to allocate them.
|
20 years of wrong directions
|
Topic: history, technology | Link here |
Twenty years ago today I started this idea of saving all my RCS files in a separate hierarchy. Bad idea, and it's still biting me.
|
bhyve: out
|
Topic: technology, opinion | Link here |
I've already had my concerns about using bhyve instead of VirtualBox, but another thing that concerned me was how to save a running instance. Once again Google Gemini to my aid. You don't:
Unfortunately, as of July 1, 2024, bhyve itself doesn't offer a built-in feature to save the state of a running virtual machine. This functionality is under development, but there's no official release yet.
| Tuesday, 2 July 2024 | Dereel | Images for 2 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
The end of US democracy?
|
Topic: history, politics, opinion | Link here |
The Supreme Court of the USA has made an amazing decision: the President is above the law.
I needed to catch my breath. This special treatment of the ruler was one of the differences between the USA and Great Britain. It seems that the concept of equality under the law is not stated as clearly in US law as it is elsewhere, but the Fourteenth Amendment to the United States Constitution implies it, and it also states:
Section 3. No person shall be a Senator or Representative in Congress, or elector of President and Vice President, or hold any office, civil or military, under the United States, or under any State, who, having previously taken an oath, as a member of Congress, or as an officer of the United States, or as a member of any State legislature, or as an executive or judicial officer of any State, to support the Constitution of the United States, shall have engaged in insurrection or rebellion against the same, or given aid or comfort to the enemies thereof.
And, it seems, this decision reverses exactly the issue of the attack on the US Capitol on 6 January 2021. How can that happen? My understanding is that the only way to reverse this decision is for the Supreme Court to change its mind.
The consequences are staggering. The Shovel put it into perspective. Biden Responds to Supreme Court Ruling by Immediately Assassinating Donald Trump. Part of the article is worth reading:
The Biden team used an unconventional method to carry out the assassination – an actor posing as Vladimir Putin was hired to tell Trump that he doesn’t like him. Trump immediately died of sadness.
But somehow that's the death knell of US democracy. The 2020s have not been a good decade so far. I see even worse times ahead.
|
Fixing the gate
|
Topic: gardening | Link here |
Our front gate does not shut very accurately, and lately it has been overshooting the “closed” position. A minor adjustment, maybe?
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
It doesn't look like much, but I think it should be enough. I didn't want to overdo it and risk breaking off the hook.
|
ANZ: You typed the wrong password
|
Topic: general, technology, opinion | Link here |
Time for my monthly check on my bank accounts today. But no, my ANZ web page told me that my password was wrong.
What? It's stored in the web browser. I also have it written down, so I tried that. No, wrong. Please reset.
How did that happen? It's definitely not my doing. Called up and finally, after nearly a minute of announcements, I was asked the reason for my call. “I think my password has been breached”.
WRONG. Another two minutes explaining to me how to use the ANZ App to reset my password. And then I was disconnected. Nearly 3 minutes for nothing.
Tried again, this time telling the non-recognition that I had a security breach. OK, 5 to 10 minute wait, which proved to be closer to 15. Connected to Gracia, who told me that no special characters were allowed in the password, and then helped me reset it after a third time also failed and locked me out of my account.
When was my password last changed? 2 May, she said. OK, I've logged in since then, so if their site didn't let me log in, there must be something seriously wrong. Please inform your security team. ”I am the security team”. No escalation needed: I typed my password with an special character or some other error.
Please take a complaint. After three attempts and a threat to involve the regulatory bodies, she agreed to do so. Give me the complaint number, please. Can't do that yet, will send a “text message”, presumably an SMS. No, I'd like to have an email, please. Sorry, we can't send email. Wait on the line.
Finally she came back with the number. And where do we go from here? She seemed to think that that was the last I would hear of it, and maybe she's right. And after all, she said, I just typed my password in wrong.
I exploded, being rather rude and calling her an idiot. I regret that, but I've seldom had such a frustrating and stupid interaction.
And after it was all over and done, I got an email with the complaint number.
Somehow this is so like the interaction with Aussie Broadband a few months ago. We're being held to ransom by the “help desk” people.
| Wednesday, 3 July 2024 | Dereel | Images for 3 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Repairing the bore pump
|
Topic: Stones Road house, gardening, general | Link here |
Steven from UPI along today with an apprentice (Riley?) to look at the wiring of the bore pump. Irritatingly, it worked for him, sort of. At least the pump ran. But it seems that these junction boxes are really pressure switches, one for the pump, one for the pressure cell. And both of them were badly corroded by the bore water:
|
|||||||||||||||||||||||
So he had to replace both of them, along with the pressure gauge:
|
|||||||||||||||||||||||
Why are they not made of a material that doesn't corrode? They're very expensive—the parts alone came to a total of $450.
|
Preserved lemons, status
|
Topic: food and drink, opinion | Link here |
It's been a little over a week since I started on my preserved lemons:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Now they look like this:
|
|||||||||||||||||||||||
One of the big differences between the recipes was whether to add lemon juice or not. The rationale is that the lemons will go mouldy if they're not covered. No recipe specifically wants them uncovered, so my guess is that the ones that don't add juice expect there to be enough juice in the lemons themselves. Clearly that's not the case here, so today I added what proved to be 100 ml of lemon juice:
|
|||||||||||||||||||||||
So far there's not much evidence of softening.
| Thursday, 4 July 2024 | Dereel → Ballarat → Dereel | Images for 4 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Shopping again
|
Topic: general, food and drink, health | Link here |
Somehow a number of things have cropped up that required me to go shopping: Mona has learnt how to open doors, so we need knobs instead of levers. As often, I needed a visit to the Fruit Shack, and my glasses still needed adjustment. In addition, ALDI had an air compressor on offer, and it was too heavy for Yvonne to pick up yesterday, so did that too.
And somehow the day was done. The good news is that everything went smoothly, not a foregone conclusion.
|
ANZ password: Groggy's fault
|
Topic: technology, opinion | Link here |
Checking my local web pages, discovered this:
-r--r--r-- 1 grog lemis 6462 2 May 14:32 bank.php
That's the page I use to access financial web sites. And it had a new password, one that I hadn't transferred to my password list. So the issues that I had on Tuesday were my fault after all.
Did Gracia do the right thing then? No. she couldn't have known that, and her utter conviction that I had typed a special character in the password shows a lack of will to investigate the real issues. But I do feel silly.
|
Problems with NBN scheduled outages
|
Topic: technology, general, opinion | Link here |
It seems that I'm not the only person to get annoyed by National Broadband Network scheduled outages.
| Friday, 5 July 2024 | Dereel | Images for 5 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
두부부침양념장
|
Topic: food and drink, opinion | Link here |
The (Korean) title 두부부침양념장 transliterates as dububuchim-yangnyeomjang, a tongue-twister that even reflects in the Korean spelling. It translates as “Pan fried tofu with spicy sauce”, but it seems easier to distinguish it by calling it Korean fried dofu.
The photo looked good:
So yesterday I bought some dofu, not quite the same as used in the recipe:
|
|||||||||||||||||||||||
In particular, the pieces are smaller.
After a bit of recipe translation, came up with this:
|
|||||||||||||||||||||||
The rice is mentioned in the original recipe, but not illustrated.
Per serving
| quantity | ingredient | step | ||
| 150 g | firm dofu | 1 | ||
| 12 g | soya sauce | 2 | ||
| 5 g | gochugaru (Korean chili flakes, 고춧가루) | 2 | ||
| 15 g | garlic | 2 | ||
| 30 g | onion, chopped finely | 2 | ||
| 12 ml | sesame oil | 2 | ||
| sesame seeds | 2 | |||
| 15 g (1) | spring onion | 3 | ||
Cut the dofu into slices about 6-7 mm thick:
|
|||||||||||||||||||||||
Fry until crispy:
|
|||||||||||||||||||||||
Mix the ingredients together:
|
|||||||||||||||||||||||
Cut spring onion and add to the sauce:
|
|||||||||||||||||||||||
Serve with rice:
|
|||||||||||||||||||||||
|
Britain votes and gets on with it
|
Topic: politics, opinion | Link here |
A number of national elections are creating interest at the moment: the USA, Iran, France and now the United Kingdom. The USA elections have been number one topic for months, and there are another 4 months to go (until Melbourne Cup day). France started its election campaign at the beginning of last month and had elections on Sunday, only the results were a tie, so they need a second round come Sunday. In Iran it was similar: they called elections at the beginning of last month, they were held last Friday, and they, too, need a run-off that will take place today.
Meanwhile, in the UK, they also called elections at the beginning of last month. They took place yesterday. No run-off: polling ended at 22:00 BST (7:00 this morning here), and by the evening the King had invited Sir Keir Starmer to form a government. No messing around like in the USA, where the incumbent has 2½ months to subvert the election before the president-elect actually takes office.
|
New doorknobs
|
Topic: Stones Road house, animals, general, opinion | Link here |
I really don't like the idea of round doorknobs: they're significantly more difficult to handle than the lever type. But Mona agrees, so I have to do something.
Then Yvonne came up with the obvious answer: reposition the levers so that they're pointing down. Now why didn't I come up with that idea? But it turned out that the maker of the levers had found an asnwer: you can't. So I had to mount the knobs after all:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Mona was not amused. In the evening we heard her trying to get into my office and failing. She didn't exactly growl, but her tail was twice the normal thickness. Somehow I feel that I have let her down: here she had the great success of finding a new way through doors, and already it stops working.
| Saturday, 6 July 2024 | Dereel | Images for 6 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Breakfast experiment fail
|
Topic: food and drink, opinion | Link here |
I have a surprising amount of dofu and almost no recipes for it. How about using some in my Pad Thai recipe? Fried the slices a little hotter (7 on the induction cooker), which resulted in much faster browning, but despite all attempts they still stuck a little.
|
|||||||||||||||||||||||
And while I was at it, I used a different ALDI paste (first image) instead of their “Pad Thai” paste:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The result?
|
|||||||||||||||||||||||
Boring! And the “stirfry” sauce tasted of almost nothing.
|
A mouse!
|
Topic: technology, history, opinion | Link here |
I'm still planning to put in the new teevee.lemis.com in the lounge room Real Soon Now. The intention is to run it in parallel with tiwi until I'm happy. Until then, all the videos will be on tiwi. Things that are missing are: keyboard, mouse, display card (maybe).
Surely I have lots of keyboards and mice? Yes, I do, but for some reason they “don't work”, at least partially because all the dongles are unmarked, and none of my experiments in pairing them have worked. So when I was in town on Thursday, I picked up an el-cheapo wireless keyboard and mouse from Officeworks, total price $33.
They had others on offer, including things aimed at Apple for round 10 times the price, so I wasn't expecting very much. But what I found was much better than I dared to hope.
Backspace 35 years to my first graphics environments. Mice had three buttons and nothing else, and of course they were wired. Since then a number of things have happened: wireless mice appeared (very good; I had to use cable mice again recently, and they're a real pain), scroll wheels appeared (also useful), and the second button went away (bad). To make up for it, some high-end mice ended up with additional buttons for the (right) thumb, and I was able to remap them to button 2. But the cheap mice didn't.
But this one does! In fact, the button layout is the same as my Logitech M705: two main buttons, scroll wheel in between, which can also be pressed for a button press, and two buttons for the thumb. xev tells me:
| left | right | scroll | scroll | scroll | thumb | thumb | ||||||
| forward | back | press | forward | back | ||||||||
| 1 | 3 | 4 | 5 | 2 | 9 | 8 | ||||||
That's the same for the Logitech mouse, except that the thumb buttons have been remapped.
The probe messages are:
ugen0.3: <Telink 2.4G Mouse> at usbus0
ums1 on uhub4
ums1: <Telink 2.4G Mouse, class 0/0, rev 1.10/1.00, addr 34> on usbus0
ums1: 5 buttons and [XYZ] coordinates ID=1
That's much more than I had hoped for. And to make things even better, the mouse has a storage place for the dongle, so if I do take it out of service, at least I won't have to look for the dongle.
| Sunday, 7 July 2024 | Dereel | Images for 7 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Olympus E-30 auto-exposure
|
Topic: photography | Link here |
I've been using my Olympus E-30 for photos of a still-planned Analemma series for almost 5 years. I use a 10 stop (99.9%) neutral density filter and take two images, one at 1/4000 s and f/5.6 for the sun, and the other at 2 s and f/5.6 for the surroundings:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The intention is to align the surrounding images, and then use the same alignment to join the sun images. I must now really have images for every day of the year, and at some point I should really get round to doing it.
But today it occurred to me that I have almost never taken photos with the camera and auto-exposure. Tried that today. Big fail:
|
|||||||||||||||||||||||
It must have been 6 EV underexposed. Even attempts to correct it didn't help much.
|
|||||||||||||||||||||||
What went wrong there? Yes, I still had the ND filter on, but auto-exposure should have compensated for that.
|
Understanding mice
|
Topic: technology, opinion | Link here |
More playing around with my new mouse today, on tiwi. Yes, it works, but how do I remap the side buttons? moused is started with
root 36620 0,0 0,0 13112 1720 - Is 22:13 0:00,00 /usr/sbin/moused -3 -p /dev/ums1 -t auto -I /var/run/moused.ums1.pid
That doesn't even match the entries in /etc/rc.conf.d:
moused:moused_enable="YES"
moused:moused_flags="-m 2=4" # Remap rear side button to 2.
And if it did, it wouldn't help: the side buttons are 8 and 9, not 4. Is there something in the X configuration file? No:
Section "InputDevice"
# generated from default
Identifier "Mouse0"
Driver "mouse"
Option "Protocol" "auto"
Option "Device" "/dev/sysmouse"
Option "Emulate3Buttons" "yes"
Option "ZAxisMapping" "4 5"
EndSection
OK, shoot down the moused instance and start it again with the correct option -m 2=8. But when I shot it down, the mouse carried on working! /dev/sysmouse must bypass the driver daemon.
I've been playing with X configuration for over 30 years, but they keep changing the goalposts.
|
Fish gratin
|
Topic: food and drink, opinion | Link here |
Yvonne made dinner tonight, a fish gratin:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
| Monday, 8 July 2024 | Dereel | Images for 8 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
The people have spoken—some of them
|
Topic: politics, opinion | Link here |
There were three important elections in the last few days: the United Kingdom general election, the French legislative election and the Iranian presidential election. All of them were interesting: arguably the results in the UK and Iran show positive tendencies, and in France they were able to limit the power of the Rassemblement National.
But the details are interesting: in France and Iran the elections were second rounds. In Iran only 27.2% of registered voters voted for the winner Masoud Pezeshkian. The other candidate, Saeed Jalili, received 22.5%. And over 50% of registered voters didn't vote at all. What would have happened if they had voted?
But what if there are more than two candidates? Take the situation of the first French election round, where most constituencies had multiple candidates. And that was to the advantage of the RN. So in the second round the other two main parties, the Nouveau Front populaire and the Ensemble pour la République (the party of Emmanuel Macron), decided to only field one candidate for many of the constituencies. The result: RN got 37.06% of the votes, NFP got 25.8%, and Ensemble got only 24.53%. So the RN won, right? Wrong. They ended up with only 142 seats, while the NFP got 180 and Ensemble got 159. That only makes sense (and then only just) because of the way the votes were distributed.
And then there's the UK, where the Labour Party defeated the Conservative Party by a “landslide”. Labour got 211 more seats, the Conservatives lost 251 seats for their worst-ever result. But looking at the vote count, Labour got 33.69% of all votes cast, only 1.6% more than in 2019. And that, along with a more significant loss by the Conservatives (19.9%) made the difference between 200 and 411 seats out of 650.
What's wrong with this picture? In Iran, low voter turnout means that the results are only marginally representative of what people want. In France, and arguably in the UK, the number of candidates determine who wins. And in Australia? I'm coming to the conclusion that the Australian system is better: it requires compulsory voting, which must result in a more accurate representation of what the voters want, and if your candidate doesn't win, you specify who should get the vote instead. I wonder how that would have affected the results in the three countries.
But once again I'm reminded of Jawaharlal Nehru's comment “Democracy is good. I say this because other systems are worse”.
|
Understanding mice
|
Topic: technology, opinion | Link here |
Yesterday's confusion with mice puzzled me. Wasn't there a way to set a button map?
Yes, and I found it not so long ago: there's a button map that xinput is too polite to display with xinput list-props. Instead I need xinput get-button-map
With that information, I've added this to my HOWTO page:
Current (2024) versions of X don't seem to use moused any more. Use /dev/sysmouse:
Section "InputDevice"
# generated from default
Identifier "Mouse0"
Driver "mouse"
Option "Protocol" "auto"
Option "Device" "/dev/sysmouse"
Option "ZAxisMapping" "4 5"
EndSection
It's not clear if ZAxisMapping is of any use.
I used to include this:
Option "Emulate3Buttons" "yes"
But there's a better way to do this. First find the mouse or mice:
=== grog@hydra (/dev/pts/23) ~ 369 -> xinput
⎡ Virtual core pointer id=2 [master pointer (3)]
⎜ ↳ Virtual core XTEST pointer id=4 [slave pointer (2)]
⎜ ↳ System mouse id=6 [slave pointer (2)]
⎜ ↳ Logitech USB Receiver id=11 [slave pointer (2)]
⎜ ↳ Telink 2.4G Mouse id=12 [slave pointer (2)]
⎣ Virtual core keyboard id=3 [master keyboard (2)]
↳ Virtual core XTEST keyboard id=5 [slave keyboard (3)]
↳ System keyboard multiplexer id=7 [slave keyboard (3)]
↳ Power Button id=8 [slave keyboard (3)]
↳ Logitech USB Receiver id=9 [slave keyboard (3)]
↳ vendor 0x0430 Sun USB Keyboard id=10 [slave keyboard (3)]
Yes, those are two mice. Next,
=== grog@hydra (/dev/pts/26) ~ 9 -> xinput list-props 11
Device 'Logitech USB Receiver':
Device Enabled (155): 1
...
libinput Middle Emulation Enabled (287): 0
libinput Middle Emulation Enabled Default (288): 0
The important thing here is the number 287 in that output. Next, note the name in the initial output and enter:
# Logitech mouse: find the ID
MOUSEID=$(xinput | grep -m 1 "Logitech USB" | sed 's/^.*id=\([0-9]*\)[ \t].*$/\1/')
xinput set-prop $MOUSEID 287 1 # Logitech
If the mouse has other buttons that can be remapped, do something like:
# Logitech mouse: find the ID
MOUSEID=$(xinput | grep -m 1 "Logitech USB" | sed 's/^.*id=\([0-9]*\)[ \t].*$/\1/')
# And set middle button to 8 and 9
xinput set-button-map $MOUSEID 1 2 3 4 5 6 7 2 2 10 11 12 13 14 15 16
That's the same assignment to MOUSEID. This sets buttons 8 and 9 to return button 2.
| Tuesday, 9 July 2024 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Quiet day
|
Topic: general, technology, multimedia, opinion | Link here |
Today was one of those days where I didn't do anything worth mentioning, quite a difference from the last few weeks. When am I going to continue with upgrading tiwi (or teevee)? Somehow the issue with VirtualBox has thrown a spanner in the works. An obvious attempt to localize the problem would be to try the same scenario running Linux. And for that I need a machine for the installation—the same one that I need for teevee.
What do I do? So far, nothing. If I can speed up the migration from tiwi to teevee, I will have a machine for the VirtualBox testing.
| Wednesday, 10 July 2024 | Dereel | Images for 10 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Scam from Google Play
|
Topic: technology, opinion | Link here |
While looking for that most elusive of things, a good Android app, got this unsolicited popup:
|
|||||||||||||||||||||||
221.236.211.162? That's not an IP that I use. A bit of checking showed:
=== grog@eureka (/dev/pts/6) ~ 204 -> host 221.236.211.162
162.211.236.221.in-addr.arpa domain name pointer 162.211.236.221.broad.cd.sc.dynamic.163data.com.cn.
A traceroute failed after 13 hops, so it's clearly not nearby.
What does that mean? Possibly the advertisement can't determine the IP addresses in use on my phone, but why use a fake address? Is this some kind of scam? And in passing, it suggests that the IP addresses can be hidden by encrypting the content. Whoever wrote that doesn't understand the Internet Protocol, or thinks that the user is stupid enough not to know it: encryption doesn't change the IP addresses.
|
More work on teevee
|
Topic: multimedia, technology, opinion | Link here |
Finally got round to doing some more “work” on teevee. Somehow this cable mess really gets me down. Spent a lot of time trying to get the correct cables in the right place, in principle only monitor and audio. And apart from minor frobs to get the mouse right and to mount the correct file systems, all worked pretty much out of the box. But the next step is more cabling in the lounge room, which is another horror.
|
Executing wills
|
Topic: history, general, opinion | Link here |
Unexpected mail from Benoît Nogues today:
Following the death of Mr Louis Binos de Pombarat , the notary in charge of the estate wants to contact Yvonne , please contact the notary office :
What's that? It was too late for us today, but it's intriguing. It's unlikely that Louis left anything to Yvonne: they clarified their relationship with the divorce in 1980. What else might be the reason?
| Thursday, 11 July 2024 | Dereel | Images for 11 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
What use fridge magnets?
|
Topic: animals, general, opinion | Link here |
In the last millennium an interesting combination of advertising and information arose: fridge magnets, small magnetic cards that stuck to the side of a refrigerator and contained address information for companies. Our most exposed fridge contained a whole lot of them.
But clearly the cats think that that's a bad idea:
|
|||||||||||||||||||||||
And do we really use them? I think not.
|
How much sugar?
|
Topic: food and drink, health, opinion | Link here |
From time to time, like here, I comment on the impression that almost every US American recipe includes sugar, and I've taken to ignoring it. But today I came across this map of daily sugar intake by state in the USA:
The measurements are in teaspoons per day. Interestingly, though, it seems that the weight of a teaspoon of sugar has been standardized in the USA: 4.2 g (or 4 g between friends). That's particularly interesting because of the unit: 4.2 g, not 0.148 ounces.
So that means that the average US American eats between 56 and 85 g of added sugar every day! That's quite amazing. What does “added” mean? Beyond what is naturally present in foodstuffs, it seems. Clearly there are borderline situations. Does it include the sugar in orange juice? In honey? I would guess no and yes, respectively.
The other thing is in the fine print: the American Heart Association recommends between 6 to 9 tsp/day (25 to 38 g). While it's considerably less than actual consumption, it still seems to be amazingly high. I've compared my consumption: 0, assuming the single glass of orange juice for breakfast doesn't count. That's not a reaction to my diabetes: I just don't want any more.
| Friday, 12 July 2024 | Dereel | Images for 12 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
¡Hola Estíbaliz!
|
Topic: general | Link here |
Estíbaliz Harrison (a name I hadn't heard before) along today to do some cleaning. She's the first cleaner we have had who does it on a professional basis, and she did a good job. The most interesting thing is her name, which I had never heard before. But she calls herself Steph, a lot easier to pronounce and remember.
|
VirtualBox on Ubuntu
|
Topic: technology, opinion | Link here |
My continuing pain with VirtualBox had established that the problem was not (only?) related to the hardware on which it ran: it also happened on quartet, a test ThinkCentre. So if it's a single point of failure, it's probably either FreeBSD or the specific VirtualBox port.
OK, before I move quartet (as teevee) to the lounge room, how about installing Ubuntu on it and trying like that?
I've been through the installation a couple of weeks ago, and I still have the install media. How hard can it be? Excruciating! It started with a broken copy to the USB card, but after I had done the install, I still needed to integrate it into the home LAN. For NFS I need a user grog and a UID that matches the other systems. How do you do that? The “Settings” menus don't seem to cater for it. And when I entered my password, it decided that it didn't like it. Not “please choose another password”, just blanked-out buttons and input fields (“I implied that I didn't like the password. Don't try to enter it again!”).
OK, at least I can change my IP address. Yes, it allows me. But I can't specify a domain name. OK, fix it in /etc/resolv.conf. But no, you can't do that!
# This is /run/systemd/resolve/stub-resolv.conf managed by man:systemd-resolved(8).
# Do not edit.
#
# ...
nameserver 127.0.0.53
options edns0 trust-ad
search .
root@echuca:/etc# ls -l resolv.conf
lrwxrwxrwx 1 root root 39 Apr 24 20:47 resolv.conf -> ../run/systemd/resolve/stub-resolv.conf
But that's all wrong! And the “Settings” don't give me the option of changing it, though they know that the name server address is 192.109.197.137. There's this horrible systemd thing. What can I do? Rename it and create a new resolv.conf like God intended. Who knows how long it will stay there.
Ultimately installed a number of apts, many of which I would have expected to be
present in the base
system: rwho, openssh-server, xterm, ifconfig, emacs. In
passing, one of the dependencies was Postfix, which was also
configured for me. I should take a look at what it has done.
And with that I was able to fix up /etc/passwd and friends.
Next, mount NFS file systems. No, no, you can't do that! First you need a mount.<something> (the same message applies to other file systems). OK, Ubuntu, where do I get it? 389 lines of output that all looked like this:
|
|
|||||||||
This HORRIBLE half-coloured text! In this case I found that putting the output through less sanitized it, showing that the vast majority of the entries have nothing obvious to do with NFS. But WHY is it so difficult to get Linux to shut up? Off on a tangent to find ways, but there seem to be none. This answer suggests two ways, both of which look like they should work, but don't:
Even the second solution has a caveat: “The presence of this file disables color output for compatible utilities”. Clearly everything I use is incompatible.
People, this is STUPID. If you really, really want to splash colour over a screen, you should at least set the background as well.
Back to the case in hand: I found the package for mount.nfs, but forgot to write it down, possibly nfs-common. It wasn't immediately obvious. Next time will be the same pain all over again. But after that, with a marginal environment set up, I was able to install VirtualBox. It looks—again—completely different. It's version 7.0.16, while on FreeBSD it's 6.1.18. That's not as out-of-date as it might seem: both seem to be maintained in parallel, and both date from this year, though VirtualBox 7.0.0 was released on 10 October 2022. But it would be interesting to see whether my network problems would occur under 7.0. Can I face the prospect of updating the port?
Lots of messing round, copying hundreds of gigabytes of files, but kept running into problems starting the VM. I really need to read up on this and find the correct way to start it.
| Saturday, 13 July 2024 | Dereel | Images for 13 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More garden work
|
Topic: gardening | Link here |
Jesse Walsh along today to do some more work in the garden.
|
More VirtualBox fun
|
Topic: technology, opinion | Link here |
Spent a large part of the day setting up VirtualBox on Ubuntu. I have resolved at least one thing: to move a VM from one instance of VirtualBox to another, I need to press the “Add” button, not the “New” button. And that uses the configuration file (like despise.vbox), and it works up to a point. Clearly a saved image isn't going to work well, especially as in this case I had to reduce the memory from 24 to 16 GB and the number of CPUs from 8 to 4.
But it's so slow! Well, at least the disks were. There was a difference: this setup uses normal magnetic disks, while my other instances have used SSDs. But I used magnetic disks on eureka, and they were nothing like as slow as on echuca, the Ubuntu “machine”. After a cold boot it took two hours for disk activity to die down.
Why is it so slow? Is this Ubuntu, or is it VirtualBox 7? While I was pondering that, took my weekly house photos, which I had wanted to process on this instance of despise. That was clearly not the optimum way, so I did them on distress, a real machine, instead.
Later, though, things had calmed down, so I tried again. Surprisingly, there was little difference. distress and echuca (the Ubuntu host) are both ThinkCentre M93ps, so the results should be comparable. Based on events, I had expected the VirtualBox instance to be an order of magnitude slower, but in fact there wasn't much difference: the hardware version converted 129 images in 17:37 minutes, while the VM took 18:55, about 93% of the throughput of the naked machine. That's quite acceptable. The real question is just why there's so much disk activity.
|
The pain of cables
|
Topic: technology, general, food and drink, opinion | Link here |
Osso buco for dinner tonight. For that, we need risotto alla milanese. And how about that, according to our freezer database, we have some in basket 9.
Do I believe that? Not overly. But it's a good chance to bring the database up to date with reality. Print out the contents of 9 and compare.
Oh.
=== grog@eureka (/dev/pts/7) ~ 109 -> echo 'select * from freezer where basket="9" order by description ;' | mysql household | a2ps
ERROR 1064 (42000) at line 1: You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near '0~select * from freezer where basket="9" order by description' at line 1
Huh? That's exactly the query I submitted directly. Why doesn't it like it? It took me a while to realize that the mouse was adding invisible information. Pasted directly into mysql, it looked like:
=== grog@eureka (/dev/pts/7) ~ 110 -> mysql household
mysql> 0~select * from freezer where basket="9" order by description ;1~
Why? A sick mouse? Yet Another thing to investigate. Prepared the query in Emacs and it worked. Well, no error messages. But no output on the printer either.
What does lpr say? “Can't communicate with lp”. Took a look: no network cable! I must have “borrowed” it for something else. Off looking for a relatively long cable, and finally found a particularly nasty-looking one that must once have done duty in Wantadilla, where it had been stuck to the floor with masking tape. One end had the fixing tab broken off, but that shouldn't be a problem. But the office switch (16 ports) didn't have a spare port! Surely I don't have that many devices connected.
No, at least one cable had nothing on the other end. OK, connected it up. On the second attempt I was able to talk to it. But still no output!
Is that something with a2ps? I had some distant memory of having it set up to automatically send to lpr. But I can do that directly: pipe to lpr instead of a2ps. And that worked, sort of, merrily going off the right margin: no ^M characters.
Finally chased down where I use a2ps. It wants to be invoked a a2ps -M A4. And then—finally—I got my single page of output.
And the risotto? No, none there, but a couple of other things that I had forgotten to enter. Databases are fine for this sort of thing, if you keep them up to date. But that's such a pain.
|
Thoughts on risotto alla milanese
|
Topic: food and drink, opinion | Link here |
So I had to make a new risotto alla milanese. I never cease to marvel at the amount of broth it needs: 4 times the weight of the rice, where I normally use about 1.8 times as much water to cook normal rice.
And it wants beef broth, not water. Do we have that? Looked in the pantry and found a particularly dusty container with something claiming to be beef broth, suitable for vegetarians. That sounds wrong, probably the reason that the container is almost full, although it past its “best by” date in April 2002. Found another one that was only 9 years out of date, mixed it up and... it was far too dark. So in the end I used chicken broth powder.
First insight: yes, I knew that the risotto takes a long time. I had given it 30 minutes, but it was not really enough. 45 minutes would be better. Second insight: it really needs to be done slowly. After 30 minutes and ridiculous quantities of liquid, it still wasn't cooked through. I really hadn't expected that.
| Sunday, 14 July 2024 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More noodle cooking times
|
Topic: food and drink, opinion | Link here |
Started a new pack of “Chef Lo” noodles today to go with mi udang. I had assumed without checking that they would need 2 minutes, like nearly all noodles of this nature. But no, after 2 minutes the basket had barely opened. In the end I gave them 3, though they were still relatively firm. After being in the soup for a while they softened, so maybe 3 to 4 minutes is correct.
But then I checked the cooking times page. 2 minutes! No, not this time round. On the other hand, they didn't stick together like they used to. Have they changed the composition?
|
More VirtualBox pain
|
Topic: technology, opinion | Link here |
More messing around with VirtualBox today. Why can't I save the state of a virtual machine? More experimentation showed that I couldn't do anything to stop it. No suspend. No ACPI shutdown request (was just ignored). No power off. All I could do was to shut down the VM from within (with the SHUTDOWN command).
Clearly something's wrong there. Searching the web, found a reference to the user belonging to the vboxusers group. I knew that, of course, but I had expected a clean install to set it, and it didn't. But it didn't help. Tried starting as root, and that didn't help either.
Clearly the package for Ubuntu isn't all that it's cut out to be. OK, I never really intended to use it. Try a different tack: port version 7(.0.18) to FreeBSD. Off to do that, tripping over a strangeness in the Makefile:
MASTER_SITES= https://download.virtualbox.org/virtualbox/${PORTVERSION}/:src \
LOCAL/bofh/emulators/virtualbox-ose:docs
What's this LOCAL? I couldn't find the docs on the web site, and what I see suggests that they have been created locally. OK, I can fake that: just link the docs for version 6.1.50.
But things still weren't done. The patch files failed. That's not surprising, but I was unprepared for the number of patches: 95 of them! Why are they all needed? But that kills it for now. I should check with others what efforts have been made before going to what could be much more trouble.
|
A new monitor
|
Topic: technology, health, opinion | Link here |
It's been nearly 9 months since I got hydra.lemis.com, my new machine. And I still haven't completed the migration.
One of the issues was the X configuration, of course, and I had run into trouble with my old Matrix monitor, which is probably on its last legs. I had kept it because of the lower resolution, only 2560×1440, while the new LG 27UP850 monitor is 3840×2160. But since my cataract operation, it seems that 3840×2160 is perfectly readable.
So: another 27UP850. I know my way round the thing, it works well, and now it's over 25% cheaper, since it's last year's model.
Will I ever complete this changeover?
| Monday, 15 July 2024 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
VirtualBox: enough!
|
Topic: technology, general, opinion | Link here |
So where do I go from here with VirtualBox? Discuss the matter on the FreeBSD mailing list. But I'm not a member, and something seems to have gone wrong with the signup.
Enough! This is really very frustrating. I'll leave it for a while; after all, I do have Microsoft boxen that will do the job. Instead I'll turn my attention back to the new teevee. And for that I need a better display card than the GeForce 710 that I have been using. Apart from potential performance issues, the fan is noisy, and blowing it out with compressed air doesn't help. So I've ordered a GeForce 1030 to replace it.
|
Aussie does it again!
|
Topic: technology, opinion | Link here |
Mail from Aussie Broadband today with their inimitable asterisks:
Subject: It's time to review your emergency contact details
We’re just checking in to make sure all of your **emergency** details are correct.
...
Address: **29 STONES RD DEREEL VIC 3352**
Contact number: **0353184210**
Well, no. They include the numbers of the four VoIP lines that they have half provisioned, but not the mobile phone that they also have on file. And two of those VoIP lines are not connected. Time to change things.
Log in. “Invalid user or password”. On three different browsers with saved passwords. Clearly they've changed things at their end. Finally got it changed, after this horrible 2FA, requiring the “security” of a mobile phone. No obvious way to supply the contact information.
OK, try the MyAussie app on my phone. Wrong password! OK, change that too. Another 2FA, this time sent to the same phone! What kind of security is that?
And then I found that the fingerprint access that I had admired a couple of months ago was no longer available. OK, I have the documentation online. How do I set it up?
Back to the web site. “Self-help articles on Internet, mobile, our app and more”. OK. No mention of the app at all, just an offer of a “chat now”. Tried that, but they lied:
Can't open /home/grog/www.lemis.com/grog/Photos/20240715/photolist.php: fopen(/home/grog/www.lemis.com/grog/Photos/20240715/photolist.php): failed to open stream: No such file or directory
|
|
||||||||
Four minutes wait! I've never seen that. When I finally got a response, it seemed like he was multitasking. “Where's the docco?” First he wanted my name, address and date of birth. Just for supplying a pointer to the docco. No, “Can you try reinstalling the app?” No, that's not an answer to my question. In the end I asked him to find out and send me email. I didn't get any, of course. But while we were talking, I got a prompt on my phone: enter fingerprint. And that worked. I wonder if they even have documentation for the app.
How Aussie have deteriorated!
|
Catching the Trump bullet
|
Topic: photography, politics, opinion | Link here |
It seems that news has stood still since Donald Trump was hit in the ear by a bullet at a rally yesterday. As one person said, “Never since Stormy Daniels has anybody been so disappointed by two inches”.
But then this photo surfaced:
That's from this article in the New York Times. Could it really be a bullet? Yes, I think so. I'm sure Doug Mills, the photographer, is over the moon.
Is it fake? The article doesn't think so. Neither do I. I could have taken it too. I would just have needed to be there, point my camera in the right direction and take continuous photos at 60 fps. And that's effectively what the article says. But some of the details are interesting:
Mr. Mills was using a Sony digital camera capable of capturing images at up to 30 frames per second. He took these photos with a shutter speed of 1/8,000th of a second — extremely fast by industry standards.
1/8000 s is no longer “extremely fast”. My Olympus E-30, now 15 years old, has that speed. So do our three OM-D cameras. But in fact it's not even overly fast for modern cameras. If he was taking 30 fps, he was probably using an electronic shutter. On my Olympus OM-D E-M1 Mark II and Yvonne's E-M5 Mark III that's 1/32000 s. And of course the 30 fps isn't overly fast either: my E-M1 Mark II, now 7 years old, can do double that, and the latest OM System cameras can do 120 fps. Much more important is how long they can keep it up.
And then there's the question of accuracy. They go on to say:
“If the gunman was firing an AR-15-style rifle, the .223-caliber or 5.56-millimeter bullets they use travel at roughly 3,200 feet per second when they leave the weapon’s muzzle,’’ Mr. Harrigan said. “And with a 1/8,000th of a second shutter speed, this would allow the bullet to travel approximately four-tenths of a foot while the shutter is open.”
OK, is that really 0.223"? I recall from my schooldays playing around with “.22” rifles. But yes, that seems to be the calibre. Only: 0.223" is 5.66[42] mm. Where do they get 5.56 mm? Wikipedia agrees with 5.56 mm. Is it a different way of measuring them?
And then the longitudinal measurements. It's interesting that they divide feet by 10: “four-tenths” of a foot is 4.8". Assuming that the speed is correct, that converts to 975 m/s. In 1/8000 s that would be 12.2 cm, pretty much 5", a little more the width of a hand. What the photo shows is much longer than that. Maybe the shutter speed was wrong: looking at that photo, I'd assume 1/2000 s.
And then there's the question whether it's even genuine. If it was taken with a 30 fps mode, there would have been more images showing the path of the bullet. But they say “capable” of taking up to 30 fps. What's the reality? Later they show three images taken, at a guess, about 0.5 s apart. And normal logic suggests that Mills wouldn't have taken that many photos of what was essentially a static subject.
Surely I'm not the only person to be puzzled by these details. Yes, it's close to my own interests, but this is a photo that is going to go down in the annals. Millions of people have looked at it. Has nobody thought about these details?
| Tuesday, 16 July 2024 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Back to X configuration
|
Topic: technology, opinion | Link here |
Lately I've been doing very little, and the wait for getting VirtualBox running correctly—if at all—means that I have time to look at other painful things.
So, today fvwm3. The more I look at it, the more it's a can of worms. There are two major problems:
MenuStyle black gold brown -*-*-medium-r-*-sans-12-*-*-*-p-*-*-* mwm
But now it complains about the keyword gold. The documentation starts with
MenuStyle stylename [options]Sets a new menu style or changes a previously defined style. The stylename is the style name...
options is a comma separated list containing some of the keywords Fvwm / Mwm / Win, BorderWidth, HilightBack / !HilightBack, HilightTitleBack, ActiveFore / !ActiveFore, MenuColorset, ActiveColorset, GreyedColorset, TitleColorset, Hilight3DThick / Hilight3DThin / Hilight3DOff, Hilight3DThickness, Animation / !Animation, Font, TitleFont, PopupDelay, PopupOffset, TitleWarp ...
It goes on for pages, with a few examples that don't really point at what I'm doing. About the best thing that I can glean is that the black in the original entry has become the name of an unused MenuStyle.
But one thing is becoming clearer: I can move a lot of this stuff into the .Xdefaults file, and I spent a fair amount of time playing with that, not without some success.
| Wednesday, 17 July 2024 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
How to sharpen knives
|
Topic: food and drink | Link here |
A few months back I went to some trouble to sharpen my kitchen knives. Yes, they got sharper, but it was much more work than I had expected.
Today I read this article in Quora, which still isn't dead. The article explains that cheap knives are much harder to sharpen than more expensive ones, but the expensive ones need sharpening more frequently. Specifically, he mentions molybdenum alloy or carbon steel for the expensive knives and 440 stainless steel for the cheap ones. Should I try something a bit more up-market?
|
fvwm3: enough?
|
Topic: technology, opinion | Link here |
More playing around with fvwm3 today, and at least partially cracked the MenuStyle entry. Instead of this,
MenuStyle black gold brown -*-*-medium-r-*-sans-12-*-*-*-p-*-*-* mwm
I tried this:
MenuStyle * mwm
It didn't complain, but the colours were wrong. OK, we know that: Colorset. But no, it didn't want to know. It wants MenuColorset:
MenuStyle * MenuColorset 5, mwm
But with that, I finally have my configuration the way I want it. And moving the specifications from the window manager to .Xdefaults has changed a number of things. Here an example of “before” and “after”:
+ "eureka" Exec ssh -A eureka LC_ALL=en_AU.UTF-8 /usr/local/bin/uxterm -name "xterm" -bg BlanchedAlmond -s -sl 2048 -sb -ls -j -rw -display hydra:0.3 -geometry 90x50+53+0 -e /usr/local/bin/bash &
+ "eureka" Exec ssh -A eureka LC_ALL=en_AU.UTF-8 /usr/local/bin/uxterm -name "xterm-r" -s -sl 2048 -sb -ls -j -rw -display hydra:0.3 -e /usr/local/bin/bash &
The down side is in the .Xdefaults file, of course. I have 13 “names” for my xterms, all differing only in geometry and font size. But for that I need 26 entries for each screen. There should be a way to inherit things from a parent. Maybe there is, but I haven't found it.
|
VicRoads insult to injury
|
Topic: general, opinion | Link here |
A while back we got a registration renewal notice for Yvonne's car from VicRoads. And almost immediately another letter telling us that the sum was wrong. OK, pay the new sum. No, can't do that, the sum is encoded in this silly BPAY system. Pay the sum indicated and we'll contact you about the rest.
That happened today, with a letter that suggested that it was all our fault that we hadn't payed the final $12.65. Never mind, you can call 13 11 71 and pay by credit card (card payment fee applies) or go personally to VicRoads in Ballarat (and wait in line).
Both are seriously inconvenient and costly. But there's no mention of a way to transfer the money. That's really offensive.
|
Septic tank pump fail
|
Topic: Stones Road house, general, opinion | Link here |
A couple of days ago I had another greeting from Jim Lannen, my least favourite sparky, whom by chance I first met ten years ago today. It has been raining a lot lately, and the power connection to the septic tank pump got wet, causing the RCD to trip. It was late at night, so I just disconnected it.
But clearly the tank needs to be emptied, and this evening the warning light went on. The problem is that the power point is in the middle of nowhere, so even though it's an outdoor power point, it's not adequately protected:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
This has happened in the past, and drying the connections has worked round the problem. But try as I might, I couldn't get the damn thing to work today. Another thing to be looked at tomorrow, this time more urgently. Is it even the same problem, or has the pump given up? It ran for about 20 seconds before tripping the RCD. Whom do I call? A sparky? UPI? Mari Hendriks, the plumber? This has the potential to—once again—become expensive.
| Thursday, 18 July 2024 | Dereel | Images for 18 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Septic tank pump success
|
Topic: Stones Road house, general | Link here |
Outside first thing this morning and connected the septic tank pump to a different circuit with an extension lead. Success! So at least the pump itself is still working.
What causes are left? The most obvious would be the presence of moisture in the power point. But there's also the possibility that there are a number of connections with slight current leaks, like to the bore pump, which is on the same circuit. Finding that could take time. But at least we now have time.
|
More fvwm3 conversions
|
Topic: technology | Link here |
Continued with my X reconfiguration today, and discovered that I had forgot the details that I had investigated at the beginning of the month. Put them in, and they worked. But now I have even more multiple entries. Spent some time looking for ways to inherit resources, but there doesn't seem to be one. Now I have 163 mainly duplicate resources in my .Xdefaults file, and it looks like it might increase.
|
Bad day for kangaroos
|
Topic: animals, opinion | Link here |
While walking the dogs today, came across this sight:
|
||||||||||||||||||||||||
What's that? Looks like blood. And sure enough,
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
What happened there? Yes, kangaroos get hit by cars and die. But this one didn't seem to have anything like that. Instead, its stomach was torn open, and internal organs (lungs?) had been removed:
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
How did that happen?
| Friday, 19 July 2024 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Mona acclimatization
|
Topic: animals | Link here |
Yvonne found this picture today:
|
|||||||||||||||||||||
Mona is still not really happy with the dogs, but she's coming to terms with them, even when they come closer:
|
|||||||||||||||||||||
Not before time. The red eyes are a reflection of the blanket she's sitting on, not her mood.
|
|
|||||||||
|
Emacs resources
|
Topic: technology, opinion | Link here |
So gradually my xterm configuration is close to where I want it, though the effect on the .Xdefaults file is clear:
=== grog@hydra (/dev/pts/23) ~ 762 -> l /eureka/home/grog/.Xdefaults .Xdefaults-hydra
-rw-r--r-- 1 grog lemis 2,477 30 Jun 14:38 /eureka/home/grog/.Xdefaults
-r--r--r-- 1 grog lemis 7,644 19 Jul 13:30 .Xdefaults-hydra
Next, Emacs. And somehow things started all over again. Somehow the specified resources don't seem to have any effect. They do, however, have a couple of strangenesses. For example, I have:
emacs*Geometry: 110x75
That's the xrdb syntax that I know. But an example in the Emacs manual uses a dot instead of an asterisk:
emacs.Foreground: dark green
Does that work? Does it make a difference? It seems that emacs and Emacs do make a difference:
The program name is the name of the executable file to which the resource applies. For Emacs, this is normally ‘emacs’. To specify a definition that applies to all instances of Emacs, regardless of the name of the Emacs executable, use ‘Emacs’.
I need to understand that. Could it be a hint at some kind of hierarchy? Courtesy of emacs or of X?
One thing did become increasingly irritating, though: this popup telling me that “pure-GTK” doesn't work with X. Somehow, over 6 months later, it still grates that a port should be configured to work with Wayland but not with X. OK, time to fix it.
=== root@hydra (/dev/pts/32) /usr/ports/graphics/hugin 81 -> git diff /usr/ports/editors/emacs-devel/Makefile diff
diff --git a/editors/emacs-devel/Makefile b/editors/emacs-devel/Makefile
--- a/editors/emacs-devel/Makefile
+++ b/editors/emacs-devel/Makefile
@@ -87,7 +87,7 @@ OPTIONS_DEFINE= ACL DBUS GNUTLS GSETTINGS HARFBUZZ JSON LCMS2 \
XFT XIM XML XWIDGETS
OPTIONS_DEFAULT= ACL CAIRO DBUS GIF GNUTLS GSETTINGS GTK3 \
HARFBUZZ JPEG JSON KQUEUE LCMS2 MAILUTILS \
- MODULES NATIVECOMP OSS PGTK PNG SCROLLBARS \
+ MODULES NATIVECOMP OSS PNG SCROLLBARS \
SOURCES SQLITE3 SVG THREADS TIFF TREESITTER \
WEBP XML
OPTIONS_GROUP= GRAPHICS
@@ -118,7 +118,7 @@ MAILUTILS_DESC= Mail framework via GNU Mailutils
MODULES_DESC= Dynamic modules support
NATIVECOMP_DESC= Emacs Lisp native compiler support
OTF_DESC= Opentype fonts
-PGTK_DESC= Pure GTK GUI [implies CAIRO, GTK3]
+PGTK_DESC= Pure GTK GUI [implies CAIRO, GTK3, wayland only]
SCROLLBARS_DESC= Toolkit scroll-bars
SOURCES_DESC= Install sources
SVG_DESC= Scalable vector graphics support [implies PNG]
How I love Linux half-coloured output! How I wish I could turn it off globally! At least my choice of a not-quite-white background at least gives a clue that there's something there, and I can “highlight” the white on pale output on a web browser. Even that doesn't work in an xterm. And where does this diff --git come from? It's not in FreeBSD diff.
But somehow that didn't work. When I ran config, I still had PGTK set. Saved configuration? Couldn't find it. I had to run ktrace to find where it was hidden: /var/db/ports/editors_emacs-devel/options. Now why did they do that?
Things still weren't done, of course. It wanted to build gcc 13, and for some reason it failed. OK, install the package. And then some random dependency took exception to my autoconf. Another package, and then the thing went off on its merry way.
People, we've had the Ports Collection for over 30 years. When is it going to grow up?
|
xterm clipboard access
|
Topic: technology | Link here |
A while back I found suggestions that X, and in particular xterm, can access the clipboard directly. I made a note of it in my diary, but not clearly enough: it took me 20 minutes to chase down this reference, to which I referred less than 2 months ago. It wasn't without issues, which I should chase up.
| Saturday, 20 July 2024 | Dereel | Images for 20 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
New Emacs
|
Topic: technology | Link here |
My Emacs build finished yesterday without complications, and this morning I installed it, also almost without complications. Only this message:
Emacs is now built with native compilation enabled by default, which requires GCC version 11 or later. However, it is known that some elisp applications don't work correctly when compiled. You can work around any such issues by disabling native compilation using the following steps...
That's in /usr/ports/editors/emacs-devel/pkg-message. Is it something to worry about? Only, I suppose, if I get an “application” that doesn't work when being compiled. Apart from that, it started, worked correctly, and respected my X resources. Can it be that the “Pure GTK disables X resources? Since it's intended for Wayland, that could be the case.
|
More ports pain
|
Topic: technology, photography, opinion | Link here |
House photo day today, made more difficult by wind and threatening rain. But I got the photos done. The fun started with the processing:
/usr/local/share/hugin/data/plugins/top_five.py
CAT:Control Points
NAM:keep 5 CPs per image pair
fails @api-max
What does that mean? Hugin complains about things all the time, but this one was fatal. My best bet is that it was a knotted python. OK, reinstall Hugin.
===> Registering installation for hugin-2023.0.0_4
pkg-static: Unable to access file /home/src/FreeBSD/git/ports/graphics/hugin/work/stage/usr/local/lib/python3.11/site-packages/_hsi.so:No such file or directory
pkg-static: Unable to access file /home/src/FreeBSD/git/ports/graphics/hugin/work/stage/usr/local/lib/python3.11/site-packages/hpi.py:No such file or directory
pkg-static: Unable to access file /home/src/FreeBSD/git/ports/graphics/hugin/work/stage/usr/local/lib/python3.11/site-packages/hsi.py:No such file or directory
More ports agony! But I should be able to install the package:
=== root@hydra (/dev/pts/32) /usr/ports/graphics/hugin 74 -> pkg install hugin
The following 32 package(s) will be affected (of 0 checked):
Installed packages to be REMOVED:
ImageMagick7: 7.1.0.62_8
imlib2: 1.12.2_3,2
ja-tex-ptex: 3.6_4
poppler: 23.05.0_2
tex-dvipdfmx: 20230313
... many more tex
xteddy: 2.2_3
What's that? Once again it wants to completely update my system. What a pain! And why the choice of what to remove? /usr/ports/games/xteddy is a game that I found from Thomas Adam of fvwm fame. It's not my thing, and I was just too lazy to remove it, but how can it possibly conflict with Hugin?
So what do I do? Could it be that the package list is not up to date? Still more work to do for something that shouldn't have changed. But since I still have the issues with enblend, it was simpler to do everything on eureka. Somehow this migration will never end.
|
Here's a basket for all your eggs
|
Topic: technology, opinion | Link here |
I continue to get popups like this on random web sites:
|
|
|||||||||
What a good idea! It makes it so much easier to compromise multiple sites if somebody gets your Google password.
The real issue is, of course, that creating and maintaining passwords is too complicated, especially if you're using a mobile phone. I keep a non-standard list of passwords and associated email addresses, of which I have several hundred. Crack one of those and you have access to one site only. Spam me on it and I can remove it. Begone foul spammer!
|
World network crash
|
Topic: technology, general, opinion | Link here |
Email from the Ballarat Courier today:
Date: Fri, 19 Jul 2024 00:47:00 -0600
Subject: Breaking: Major IT outage impacting banks, media, airlines and supermarkets
Why should the Ballarat Courier report that? Nobody else did.
Well, for about 1¼ hours. Then the New York Times came up with this:
date: Fri, 19 Jul 2024 04:01:28 -0400
Subject: Breaking news: Global tech outage grounds flights and hits businesses
Reading headers is frequently interesting. The Ballarat Courier seems to think that it's in the US Mountain time zone—maybe its server is. And the Times uses lower case for its Date header.
How did the Ballarat Courier get there first? Part was because at the time it seemed to be limited to Australia—not the first time that our geographical position has caused us to find a global problem first. But somehow the Courier often seems to be the first to report something global.
So, what was it? Lots of news about it, of course, but it took forever to find out what had happened. A botched update from CrowdStrike caused a BSOD on Microsoft machines. And after rebooting it got loaded before anything could be done, so the machines went into a reboot loop. The solution required physical access to the machines in question.
Could this happen to FreeBSD? A good operating system can't be shot down by an application program. But yes! It seems that the module in question corresponds to a kernel module, so protection from application programs isn't enough. Only the FreeBSD solution is simpler: boot single user, remove the module, continue with boot. Does it still require physical access?
The real issue seems to be inadequate testing and too-fast rollout. Kernel modules can cause crashes. Don't spread the same module around the world all at once.
|
Mona at three months
|
Topic: animals, opinion | Link here |
We've had Mona for three months now. Things haven't developed quite as we expected. In particular, she still doesn't sleep with Bruno, though there was a time when I thought that they would get there. My guess is that they never will.
On the other hand, she's getting closer and closer to Yvonne and myself. Her only issue is with the dogs, and though that's getting better—see yesterday—there's still a way to go.
|
New Kirsch
|
Topic: food and drink, opinion | Link here |
I've received a bottle of „Schwarzwälder Kirschwasser“ (“Black Forest cherry water”) from Andrea & Daniel Schöpf in Steinach, apparently a very small producer. What does it taste like? This evening we had fondue for dinner, time for a comparison with the other two that I have:
|
|||||||||||||||||||||
|
|||||||||||||||||||||
And how was it? I don't know. I seem to have lost my sense of taste. They all tasted pretty much the same. At least it's not like this horrible Baitz Kirsch.
| Sunday, 21 July 2024 | Dereel | Images for 21 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Mona in bedroom
|
Topic: animals | Link here |
Seen this afternoon.
|
|||||||||||||||||||||||
The dogs were outside, of course.
|
More X configuration fun
|
Topic: technology | Link here |
What's next with my X configuration? Where should I start? A good one would be to interface with the clipboard. I've tried this before, but it didn't work as I expected. Time to look again.
Another answer in the page that described the xterm feature recommends adding this to .Xdefaults:
xterm*VT100.Translations: #override \
Ctrl Shift <Key>V: insert-selection(CLIPBOARD) \n\
Ctrl Shift <Key>C: copy-selection(CLIPBOARD)
But that's strange. I don't see anything like that elsewhere, and I'd prefer to do it with the mouse. Other pages suggest that you can set mouse bindings there too, but they don't make it very clear what to do. How about dumping the current xterm bindings? A quick search pointed me at editres. Tried that, didn't get a response. A web search showed that I'm not alone, though there weren't many hits, and one of them was this one, which I wrote 7 months ago.
There I found a way to dump the current resource database entries with xrdb -query, but that only shows, well, the resource database, not the compiled-in defaults that I was looking for. Once again a dead end. About all I found was:
*VT100*utf8: 1
*VT100.translations: #override <Key>Next: scroll-forw(1,page) \n\
<Key>Prior: scroll-back(1,page)
I also found this page, which I haven't digested yet. It points to documentation of the resource entries in X(7), and mentions that the name at the beginning of a resource specification can be either an instance or a class, though it's not (yet) clear how they differ. In the example, instances start with a lower case letter and classes with an upper case letter, but it's not clear that that's significant.
While searching for answers, came across this page, which purports to be a complete list of xterm control sequences. Can it help with this horrible page resizing during ports builds that I complained about in <mumble>? Well, I thought I did. When building ports, something changes the number of lines on the page:
===== Fri 19 Jul 2024 15:05:41 AEST on hydra.lemis.com: Make config
[1;24r^M
===> Options unchanged
The result is that text at the bottom of the page stays there, and it seems that the build has hung, when in fact the real output is only ⅓ of the way from the top. Why do they do that? And how do I find it? The page proved to be singularly useless, since I can't find that control sequence there. But it brought back to me that there's One More Thing to look for.
| Monday, 22 July 2024 | Dereel | Images for 22 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Understanding the CrowdStrike failure
|
Topic: technology, language, general, opinion | Link here |
A couple of days ago I engaged in guesswork about how the CrowdStrike bug caused the global system crash. Today Ginger Wolnik posted a reference to this video on the Tandem Alumnni mailing list:
It pretty much supports my guesswork: a kernel module had an unrecoverable error. But the details are interesting. Microsoft has a bad reputation for security, but in this case it seems that it's not to blame. As I guessed, the CrowdStrike software includes kernel modules. But Microsoft tests all third-party kernel modules before giving them its seal of approval.
So: did they slip up? Well, sort of. They left a loophole: modules can load data files that aren't certified. And these data files can contain pseudocode, and that's where the problem occurred.
Why did CrowdStrike resort to this method? They need to react quickly to threats, and to be responsive they can't put every change through the certification process. So Microsoft is blameless.
Well, almost. They have allowed this loophole to occur. But potentially the same problem could happen on just about any other platform. And clearly CrowdStrike is to blame by rolling out the change to everybody all at once.
The video is interesting for another reason: it has automatically generated subtitles, and they come up with a number of amusing misinterpretations. The problem occurs in curdle mode because of inadequate air checking. And anti-stress (NT stress) didn't find them
Don't like curdle? How about colel?
And then, at 9:37 into the clip, I find entries for blackboxbsd, blackboxntfs and blackboxpnp. OK, NTFS I know. Black Box I know to: it was our ill-fated police interview project 18 years ago. NTFS I also know. And of course I know BSD. But blackboxbsd has a completely different meaning: “Boot Status Data”.
|
A hairball
|
Topic: animals | Link here |
Mona came into my office today and promptly vomited:
|
|||||||||||||||||||||||
A hairball. I've heard of them, but in all my time with cats this is the first that I have seen. At 2.5 cm it looked quite big, but it seems that they can get much bigger.
|
Garden flowers in mid-winter
|
Topic: gardening, opinion | Link here |
It's a month after the solstice, time for the monthly garden photos.
It's the middle of winter, and a month ago we had a relatively heavy frost. That was particularly evident on my Hibiscus rosa-sinensis “Uncle Max”, which had fortunately been protected by some dead cannas.
|
|||||||||||||||||||||||
But then somebody removed them, and now it looks like this:
|
|||||||||||||||||||||||
Will it survive? It has done so before.
The frost also means that there are even fewer flowers than normal for this time of year. In particular, there are almost no roses:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The spring flowers are on their way:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The Strelitzia nicolai is already showing a bud:
|
|||||||||||||||||||||||
But the Strelitziae reginae that were replanted are not looking any happier:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The tree fern is also looking unhappier than for a while, perhaps because it's being throttled by some creeper:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And despite all my exhortations, the weeds are winning against the gardener:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Not for the first time, I wonder why I bother.
|
Resetting photo numbers
|
Topic: photography | Link here |
I've been keeping records of my camera battery life for the best part of 15 years. Every time a battery needs recharging, I note the file number of the last image it took, and compare it with the one before.
But that relies on the file numbers incrementing uniformly. What happens if the file number changes for some reason? That happened recently with my “new” Olympus OM-D E-M1 Mark I, which was set to reset the file number to 1 every time all files were erased. There's another option, which I should have set, to just keep counting. That way I can also tell how many photos the camera has taken.
So how do I recover? First, find what the number should be, the last 4 digits of the shutter count. And I can find that in the secret menus: 9916 total. Then, using mtools:
=== root@hydra (/dev/pts/32) ~ 101 -> mcd c:dcim/100olymp=== root@hydra (/dev/pts/32) ~ 101 -> mcopy c:97220057.ORF c:97229916.ORF
The 9 at the beginning of the file name is a setting I applied: this is our 9th Olympus camera that supports renaming files, so it helps me keep the images apart.
In passing, the Olympus OM-D E-M1 Mark II now has taken 63310 photos with the shutter.
| Tuesday, 23 July 2024 | Dereel | Images for 23 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Network failure out of sympathy?
|
Topic: technology, language, opinion | Link here |
What does “riven” mean? It's clearly the past participle of a strong verb, but which? Does it conjugate like “strive”? Off to check the found of all wisdom, the OED.
Are you human? Oh, no. They've added a CAPTCHA to insult their users. But after correctly answering the CAPTCHA, I got another one. And another one. The thing was stuck in a loop. Called up the State Library of Victoria, where Tom confirmed that OED had changed something, and that they (SLV) had been having trouble, as had all other libraries, and that it wouldn't be solved today.
In the meantime I had tried logging in to the National Library of Australia. Timeout. I could access the site, but not log in. No hope of even getting as far as the OED.
Back to the SLV and tried to access the Macquarie Dictionary. No, login timeout, reported as “error code: 502”, clearly an end-user report. All this goes far beyond whatever OED has done.
And then the Bureau of Meteorology got in on the act:
|
|||||||||||||||||||||||
No “Internet”? Has my link gone down? No, clearly something at the BOM. It seems to have lasted all day, but it was gone by the evening.
Clearly none of this can have anything to do with the CrowdStrike incident. Sympathy, maybe?
| Wednesday, 24 July 2024 | Dereel | Images for 24 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
New display card
|
Topic: technology, opinion | Link here |
The display card that I ordered last week is there. For once it's not the fault of Australia Post: it arrived in Napoleons (sorry, NAPOLEON) a week ago, just after Yvonne did her shopping, and since it wasn't urgent, I waited until today.
It's enormous! The package measured 31×43x6.5 cm, or 8.66 l!
|
|||||||||||||||||||||||
But most of it was air: the real package measured 29.5×21×5.3 cm, only 3.28 l:
|
|||||||||||||||||||||||
That's still ridiculous. The card itself measures 16×7×1.7 cm, or 0.19 l, only 2.2% of the volume of the package:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Why do people do this? Gigabyte really doesn't need to make such an enormous, wasteful package.
|
A new monitor
|
Topic: technology, opinion | Link here |
Finally the monitor that I ordered last week—a day before the display card—has arrived. I paid for it on Sunday, 14 July, but it wasn't sent until the afternoon of Monday, 22 July. And to make up for that, it arrived today, less than 2 days after being sent, and it was delivered to my door. Not bad for free postage.
Now the real work starts: tidying up the monitor arrangement that once displayed eureka, which has been there for over 9 years:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
Can't save Emacs buffer
|
Topic: technology | Link here |
Writing an email today, I tried to save the buffer. No go:
select-safe-coding-system: Loading charset map: No such file or directory, MULE-uviscii
Huh? What's that? It seems to be related to the fact that the message contained non-ASCII characters. A bit of a web search brought this page to light, which helped me work through to discover the directory /usr/local/share/emacs/30.0.50/etc/charsets/. Previously it contained MULE-uviscii.map, but now it doesn't. Clearly it relates to my recent rebuild of Emacs, but how? I found an older version, and it doesn't look like something that changes frequently, so putting it in place worked around the problem. But what other issues are lurking in the shadows?
Ports Collection: well over a quarter of a century of pain.
| Thursday, 25 July 2024 | Dereel → Sebastopol → Dereel | Images for 25 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Dog escape
|
Topic: animals, opinion | Link here |
While I was messing around in the office, Yvonne came in and said “the dogs are out the front”.
OK, good to know. But then she said “I'll get in the car and go and look for them”. Huh? Ah, “outside” meant “out on the road”. How did that happen? Off with her in the car, where the Swifts and a neighbour from down Progress Road were out discussing the matter. They said that the dogs had gone down Progress Road, and sure enough, we soon found Elena. And Larissa? Yvonne had seen her further down the road, so off to the end without finding her, though we found a sheep that looked surprisingly similar.
Then a phone call from Graeme Swift: she had been caught in a fence, Graham had freed her, and all was well—no injuries.
But how did it happen? The pedestrian gate was open. Graeme, always the cautious one, suggested a padlock on the gate, like he has. I'm more inclined to look for alternative, less extreme solutions. But for the moment the dogs aren't allowed out in front of the house when we're not there.
|
Office restructure, next step
|
Topic: technology | Link here |
Spent much of the day moving monitors around. It shouldn't have been difficult, but there were constraints:
In the process, I found a lot of old documentation, including appointment cards dating back as far as June 2017. And I managed not to disconnect any cables. Sadly, the power connector to eureka was loose, so I did have to reboot it—exactly at a time where it had no monitor connected. But I got things done faster than expected, by midday. Here before and after:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Now to get used to being in my normal position again, after 9 months of being displaced.
|
The end of the SPARCstations
|
Topic: technology, history, opinion | Link here |
What's constant about these photos?
|
|
|||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Clearly not the image quality. Clearly not the aging person between keyboard and chair. Also not the location, nor the displays. Not even the keyboards, which I had to change before the last photo. But there is one constant: the monitors are mounted on SPARCstation pizzaboxes. I had long since stopped using them as computers, but they did very well at propping up the monitors, and they have been doing it for round 30 years.
No more:
|
|||||||||||||||||||||||
The new monitors have their own stands, no SPARCstation needed. And while changing over, it became clear how much space they had taken up. I now have more space for appointment cards and things under the monitors. And things look tider from behind too. Here before and after:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Yes,still a lot of cables, but far fewer things for the cats to throw onto the floor.
|
A haircut
|
Topic: general | Link here |
Into Sebastopol today for a haircut. Originally I had planned a whole lot of things, but somehow they either sorted themselves out, or became unimportant. So a long drive just for that.
|
Carlotta returns
|
Topic: animals, general | Link here |
It has only been a little over a year since Yvonne swapped her horse Carlotta for Samba. But the conditions for keeping her were not ideal, and now Jane and Yvonne have come up with the idea of bringing Carlotta back here, and Jane will come here every few months to ride her. Today Carlotta arrived. Clearly she feels happy to be back home:
|
||||||||||||||||||||||
|
||||||||||||||||||||||
|
X configuration, part 4723
|
Topic: technology, opinion | Link here |
The good news about connecting up my new monitor was that it worked out of the box. Of course, it was connected to a running output (hydra:0.2) configured for 1920x1080, but it picked that up and ran with it.
That was the good news. I wasn't expecting the reconfiguration to be easy, and my expectations were met. Although I had an output on the rightmost monitor (to become hydra:0.3, but currently hydra:0.0), the server and Nvidia software didn't want to know about it, thus the blue screen on that monitor:
|
|||||||||||||||||||||||
OK, run nvidia-settings. I've been there before, and it wasn't quite as painful. But once again I couldn't get it to position the screens correctly, and it also wouldn't save the configuration file where I wanted it to. Finally I saved it, but it was broken. Despite “finding” the new monitor (“LG Electronics LG ULTRAFINE”), it included the old name in the configuration file. It knew about the refresh rates, though. Here an excerpt from the Monitor sections:
Section "Monitor"
Identifier "Monitor0"
ModelName "Ancor Communications Inc VS248"
HorizSync 30.0 - 83.0
VertRefresh 50.0 - 76.0
EndSection
Section "Monitor"
Identifier "Monitor1"
ModelName "LG Electronics LG ULTRAFINE"
HorizSync 30.0 - 135.0
VertRefresh 40.0 - 60.0
EndSection
Section "Monitor"
Identifier "Monitor2"
ModelName "Ancor Communications Inc VS248"
HorizSync 30.0 - 135.0
VertRefresh 40.0 - 60.0
In addition, it wouldn't let me position the monitors correctly. More playing around, and after a while found the answer (maybe) in my December diary:
|
|
|||||||||
Does that work? No, but for a different reason:
=== root@hydra (/dev/pts/32) /usr/ports/editors/emacs-devel 120 -> nvidia-settings=== root@hydra (/dev/pts/32) /usr/ports/editors/emacs-devel 121 -> nvidia-settings
nvidia-settings: Fatal IO error 22 (Invalid argument) on X server hydra.lemis.com:0.0.
The first invocation worked, but when I did it the second time, some time later, it had apparently changed something in the server, and I could no longer start the program. It also apparently reset the mouse bindings and the keyboard map. It wasn't the server itself, which has been running for well over a month:
USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
root 2307 0.0 0.1 25948804 226184 v0 S 15Jun24 4710:04.67 /usr/local/libexec/Xorg :0 -config xorg-0.conf -logverbose 6 -listen tcp -auth /home/grog
So: frob the configuration file manually and check as server 4.
|
An offer you can't refuse
|
Topic: technology, general, opinion | Link here |
Seen on the Ballarat Courier today:
Who can refuse that?
| Friday, 26 July 2024 | Dereel | Images for 26 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Yet another noodle dish?
|
Topic: food and drink, opinion | Link here |
One of the things that has surprised me over the years is how many different kinds of Chinese noodles there are. I've been playing around with a number of them, but it has recently occurred to me that I knew almost none of them in my childhood. They're Chinese, and not Malaysian.
One typical kind of noodle that I do know are these:
|
|||||||||||||||||||||||
And it seems that the noodle dishes made with them are different, too: “mah mee”, now known as mee goreng. I have a recipe somewhere, but today I just tried it out. The result:
|
|||||||||||||||||||||||
Somehow it bears too much resemblance to KL Hokkien Mee.
|
Completing the restructure
|
Topic: technology, general | Link here |
More rearranging in the office today. The monitors for hydra have been in the correct positions since yesterday, though I still need to complete the X configuration. But for the moment things work.
Next, eureka. Three monitors which barely fit in the space left over. But that problem “solved” itself: the old Matrix monitor didn't power up properly. My latest guess is that there are issues with the power connector, and disconnecting and reconnecting it was the last straw. So: goodbye Matrix, you have served me well for well over 10 years. I paid a surprisingly low price for it at the time, only round $300, and I wasn't expecting it to last long. In fact, it has lasted longer than any other monitor I have had, and the display is still clear and unfaded. A pity with the power.
| Saturday, 27 July 2024 | Dereel | Images for 27 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
hydra update: the end in sight?
|
Topic: technology, opinion | Link here |
More work updating X on hydra.lemis.com today. To avoid shooting myself in the foot, I created Yet Another configuration file and a server hydra:4, then played around with nvidia-settings. That was more work than I expected, especially because nvidia-settings kept trying to change things without warning. It seems to love changing display positions from relative (“RightOf Screen0”) to absolute (“+1920+0”) and then complaining about it not being a good idea. But finally I had a configuraton that looked good.
Only screen 2 (the second LG 27UP850) didn't display anything. More investigation showed that it would only work at resolutions of 2560×1440 or lower, but not at its native resolution of 3840×2160.
Why? It's hardly likely that it's defective. Could it be the HDMI connection? Could it be that the total resolution exceeds some limit? I can't imagine that the brand new monitor could be defective.
Still, things work like that; after all, until I decided to buy the second monitor, I was planning to run the 2560×1440 Matrix monitor in that position. Hopefully I'll solve this problem too.
There were others, of course. xv will no longer display PNG files, and neither will ImageMagick. Time up upgrade my ports, made more attractive by the fact that for once it didn't want to delete anything. But 6 GB of download! First I should bring my system up to date.
Make -j 25 buildworld kernel
...
>>> Installing kernel GENERIC completed on Sat Jul 27 14:43:51 AEST 2024
1050.52 real 21569.72 user 1856.37 sys
That wasn't just the kernel. The time includes buildworld as well. Aaaah!
|
Catch that mouse!
|
Topic: technology, opinion | Link here |
I now have a second J.BURROWS mouse, Officeworks' special. The side buttons are so convenient, so now I have one for eureka and another for tiwi.
But they're so skittish! Once upon a time mice were so sluggish that you needed to speed them up with xset m 3 or some such. But now even a setting of 0 is too fast.
More searching, and found this page: “mouse speed too fast”. And it pointed me to an xinput variable, libinput Accel Speed. By default it's 0, but I find that something round -0.5 is a reasonable speed.
|
Can't stop Perfectly Clear
|
Topic: technology, photography, opinion | Link here |
Messing around on distress, one of my Microsoft boxes today, discovered that “Perfectly Clear“ was full screen. No, not almost full screen like most programs, which leave the bar at the bottom free. It completely covered the screen, and with Microsoft's pitiful excuse for window management, that means that I couldn't resize or iconify it.
OK, c-q should stop it, right? Yes, probably, it should stop it. But the maker hasn't made any provision for stopping it! Presumably people do it by the window decorations, which were missing here.
Key bindings? Off looking for the documentation (“Perfectly Clear is no longer maintained. Go Radiant”), but finally found something that I needed to read three times to recognize. No, no normal function key, no normal key: `. Now why did they choose that?
| Sunday, 28 July 2024 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More upgrade fun
|
Topic: technology, opinion | Link here |
Yesterday's world build on hydra finished in record time. But that was yesterday, and since then my cron jobs had updated the tree. Nothing for it, another world build, this time with -j 45:
>>> World build completed on Sun Jul 28 08:41:17 AEST 2024
>>> World built in 922 seconds, ncpu: 32, make -j45
...
>>> Kernel build for GENERIC completed on Sun Jul 28 08:42:12 AEST 2024
>>> Kernel(s) GENERIC built in 55 seconds, ncpu: 32, make -j45
...
>>> Installing kernel GENERIC completed on Sun Jul 28 08:42:13 AEST 2024
978.87 real 24411.08 user 2256.35 sys
Reboot, then brave upgrading all my ports, 6 GB worth:
The following 413 package(s) will be affected (of 0 checked):
...
Number of packages to be installed: 26
Number of packages to be upgraded: 373
Number of packages to be reinstalled: 14
The process will require 874 MiB more space.
6 GiB to be downloaded.
It worked!
Well. xv still doesn't want to know about PNG files, though ImageMagick now does. Put it on the back burner: it still works on eureka.
OK, next: why does monitor 2 (the second LG 27UP850) not come up in native 3840×2160 mode? Looking at the configuration file was interesting. Here the “metamodes” for each monitor:
Option "metamodes" "DP-0: nvidia-auto-select +0+0"
Option "metamodes" "HDMI-1: 3840x2160 +0+0; HDMI-1: 1024x768 +0+0; HDMI-1: 800x600 +0+0; HDMI-1: 640x480 +0+0"
Yes, both of those should work, but there's a difference. Why? Is is part of the problem? Tried to start
=== grog@hydra (/dev/pts/4) /etc/X11 12 -> Startx 4
startx -- :4 -config xorg-4.conf -logverbose 6 -listen tcp
...
(==) Log file: "/var/log/Xorg.4.log", Time: Mon Jul 29 12:00:17 2024
(++) Using config file: "/etc/X11/xorg-4.conf"
(==) Using system config directory "/usr/local/share/X11/xorg.conf.d"
(EE)
Fatal server error:
(EE) no screens found(EE)
Oh. What's that? What does the log file say?
[ 97972.221] (EE) NVIDIA: Failed to initialize the NVIDIA kernel module. Please see the
[ 97972.221] (EE) NVIDIA: system's kernel log for additional error messages and
[ 97972.221] (EE) NVIDIA: consult the NVIDIA README for details.
And /var/log/messages?
Jul 28 13:47:02 hydra kernel: NVRM: API mismatch: the client has the version 550.54.14, but
Jul 28 13:47:02 hydra kernel: NVRM: this kernel module has the version 535.104.05. Please
Jul 28 13:47:02 hydra kernel: NVRM: make sure that this kernel module and all NVIDIA driver
Jul 28 13:47:02 hydra kernel: NVRM: components have the same version.
In passing, why do I have to look in three different places to find out what went wrong? But what's NVRM? “Never mind”, it seems. Not the solution. A number of searches drew a blank. What's the client? But looking at the kernel modules, it's clear that the kernel modules didn't participate in the ports upgrade:
=== grog@hydra (/dev/pts/7) ~ 22 -> l /boot/modules/
total 63
-r--r--r-- 1 root wheel 1,717,824 16 Jul 17:12 nvidia-modeset.ko
-r--r--r-- 1 root wheel 61,469,296 16 Jul 17:12 nvidia.ko
So I'm going to have to upgrade it manually. Hopefully things will still work afterwards.
|
teevee progress
|
Topic: technology, opinion | Link here |
And teevee? I've moved the box to the other side of my desktop, where I can easily access the back of the box. And there I made a surprising discovery: two DisplayPort connectors! I had considered the ThinkCentres too primitive to even have DVI outputs, one of the reasons I use a display card. The display card is almost certainly necessary anyway, but the DisplayPort output really surprised me.
It came up and seemed to work normally. But once again I had issues with the Alt key functions under bash not working correctly. For example, Alt-F should move one word forward, but instead it capitalizes the word.
I've seen this before and found a solution. Put this in .Xdefaults:
*VT100*utf8: 1
*VT100*eightBitInput: false
*VT100*eightBitControl: false
*VT100*eightBitOutput: true
But it seems that I forgot that, and a year later I found another solution:
*vt100.altIsNotMeta: true
*vt100.altSendsEscape: true
It seems that I used the former method:
=== grog@teevee (/dev/pts/5) ~ 5 -> diff -wu .Xdefaults-teevee /tiwi/home/grog/.Xdefaults-tiwi
--- .Xdefaults-teevee 2017-11-16 19:37:42.000000000 +1100
+++ /tiwi/home/grog/.Xdefaults-tiwi 2023-12-10 17:04:23.162348000 +1100
@@ -22,6 +25,11 @@
! *VT100*color5: #8000f0 ! magenta
! *VT100*color6: #0080ff ! cyan
! *VT100*color7: BlanchedAlmond ! white
+! Make alt key work again (20221124)
+*VT100*utf8: 1
+*VT100*eightBitInput: false
+*VT100*eightBitControl: false
+*VT100*eightBitOutput: true
But that opens another can of worms. I'm doing more and more with .Xdefaults, and I need a better way to keep things in sync. That's even the case on hydra and eureka, where I have two different environments for each machine.
In passing, saw a reference to another xterm resource in last year's entry:
*customization: -color
Something to play with later, maybe. It could help me work around this obscenity:
=== root@hydra (/dev/pts/11) /usr/ports/editors/emacs-devel 11 -> git status
On branch main
Your branch is up to date with 'freebsd/main'.
Changes to be committed:
(use "git restore --staged <file>..." to unstage)
modified: Makefile
Yes, there's a file name on the last line, but git has chosen to set the foreground to white, leaving the background as it was (white). The only way to see anything is to highlight it with the mouse.
|
Where's my stash?
|
Topic: technology, opinion | Link here |
After fixing the build of editors/emacs-devel last week, I ran into a problem: git point blank refused to update the entire ports tree. So I had to “stash” the Makefile.
Today, after the ports upgrade, I needed to rebuild the Emacs port. Where's my Makefile? Read various things that didn't make any sense to me, but I couldn't find it. In the end, just reapplied the patch: it's only two lines, and I even had the diffs in this diary. But why is everything to do with git so difficult?
| Monday, 29 July 2024 | Dereel | Images for 29 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Fake Korean side dish
|
Topic: food and drink, language, opinion | Link here |
We ate asparagus for dinner on Saturday, and I had a few ends left over, too tough for steaming. How about boiling them and turning into a Korean side dish for bibimbap?
Boiled them for 15 minutes and added some of this paste:
|
|||||||||||||||||||||||
I hadn't opened it, and inside I found this:
|
|||||||||||||||||||||||
What's that? A spice packet? Time for Google Translate.
|
|||||||||||||||||||||||
|
|
|||||||||
Oh. Something like silica gel in dry packages, though clearly this is something else. But you'd think that at least the warning would be written in English too.
While I was at it, translated the outside of the package:
|
|||||||||||||||||||||||
It turned out to be much saltier than I expected. Put together the following combination:
| quantity | ingredient | step | ||
| 35 g | asparagus stem ends, well cooked | |||
| 3 g | sesame oil | |||
| 5 g | bean paste | |||
| 10 g | light soya sauce |
|
|||||||||||||||||||||||
I'll eat it tomorrow.
|
CrowdStrike hits FreeBSD
|
Topic: technology, opinion | Link here |
When I had time this morning, continued with the hydra X configuration. As expected, I couldn't unload the nvidia.ko kernel module. I had to reboot.
But what happened then was anything but what I expected:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Panic, even in single user mode. I have almost exactly repeated the CrowdStrike incident on FreeBSD! A kernel module from a third party caused the system to get stuck in a reboot loop. And what's more, since it happens even in single user mode, I needed external help to recover: boot from a USB Stick and change the system configuration to not load the modules on boot.
Catch my breath. Then: what do I do now? I need that module, and no other. Remove it, reinstall it, cross fingers and load it.
Jul 29 15:27:05 hydra kernel: nvidia0: <NVIDIA GeForce RTX 3060> on vgapci0
Jul 29 15:27:05 hydra kernel: vgapci0: child nvidia0 requested pci_enable_io
Jul 29 15:27:05 hydra kernel: nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms 550.54.14 Thu Feb 22 01:05:40 UTC 2024
No crash! And it worked! Tried my modified configuration file on server 4, but I still couldn't get the second LG 27UP850 to work in native 3840×2160 mode.
So what's the difference? Before I had:
=== root@hydra (/dev/pts/8) ~ 58 -> l -c /boot/modules/
total 63
-r--r--r-- 1 root wheel 1,717,824 28 Jul 10:56 nvidia-modeset.ko
-r--r--r-- 1 root wheel 61,469,296 28 Jul 10:56 nvidia.ko
And now I have:
=== grog@hydra (/dev/pts/3) ~ 63 -> l -c /boot/modules/
-r--r--r-- 1 root wheel 1,717,824 29 Jul 15:25 nvidia-modeset.ko
-r--r--r-- 1 root wheel 61,469,296 29 Jul 15:25 nvidia.ko
Further checking showed that the files are indeed identical. I have no idea why it crashed, but clearly it's a good idea, at least for the moment, to load the modules after the system is in multi-user mode. I wonder if that it an option for the CrowdStrike module too.
Back to the issues with display 2. Can it be the monitor? I was sure that it wasn't, but there's a simple way to check: connect to the other monitor. And as I expected, the other monitor wouldn't work in native mode either.
So, what is it? It's an HDMI connection, while the other monitor is connected by DisplayPort. In principle I wanted both connected by DisplayPort, but my cables aren't long enough. There's a simple answer to that, of course: buy a longer cable. They're available for under $10.
But are they the right cable? As H.L.Mencken almost said, “For every problem there is a solution that is simple, obvious and wrong”. Off reading about DisplayPort releases, and gradually my eyes went funny. And when I understand the real issues, I still have the variables of the Nvidia driver and the display card. A more obvious thing to do is to reconfigure yet again and put the monitor where I can connect it with a DisplayPort cable. Too much work for today.
Only much later did I read:
Jul 29 15:27:45 hydra kernel: nvidia-modeset: WARNING: GPU:0: Unable to read EDID for display device LG Electronics LG ULTRAFINE (HDMI-1)
Jul 29 15:27:45 hydra kernel: nvidia-modeset: WARNING: GPU:0: Unable to read EDID for display device HDMI-1
That does rather suggest that there's an issue with the HDMI cable.
Other things: yes, now Hugin works, and so does ImageMagick. But xv still doesn't. And then I discovered that it doesn't work on teevee either. A new version of a program that I have been using unchanged for over 25 years, and it doesn't handle PNG any more! Why is there so much breakage lately?
| Tuesday, 30 July 2024 | Dereel | Images for 30 July 2024 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Asparagus dish
|
Topic: food and drink, opinion | Link here |
As planned, ate bibimbap for breakfast today, with the asparagus stems that I made yesterday:
|
|||||||||||||||||||||||
How did it taste? I could eat it. But I won't make it again.
|
hydra: tying up loose ends
|
Topic: technology, opinion | Link here |
Why does screen 2 on hydra not work at native resolution? Yesterday's console messages suggested the HDMI cable. I have two cables to two monitors. What happens if I swap them?
It works! What a relief! So it was something to do with the HDMI cable.
Well, it worked for a while. Then I got random lines on the screen, and at one point it briefly went blank. Playing around with the HDMI connectors worked around that, but clearly they're not up to the job. This isn't the first time that I have had trouble with HDMI. I should change the cable for a DisplayPort cable.
And now? I have the largest display I have ever had, 20,736,000 pixels, the equivalent of 10 1920×1080 monitors. And I need to come to terms with that. Next I need to work on the configuration for server 1, which I last looked at last November.
And then there's rdesktop. When I trie to access distress, the Microsoft box, nothing happened. It didn't get woken. Somehow wake(8) had lost its setuid bit, so it didn't get woken. That's the first time this has happened.
|
Fixing xv
|
Topic: technology, opinion | Link here |
The version of xv on tiwi works with PNG files, while the version on hydra and teevee doesn't. OK, copy across. Needs two libraries that weren't installed on hydra, libtiff.so.5 and libjasper.so.6. Copy them across. Runs. Doesn't display PNG! So there must be something in some other library that breaks it. What a pain.
|
Goodbye chilorio
|
Topic: food and drink, opinion | Link here |
Tuesday's dinners are almost always “leftovers” from the weekend, but today we really only had enough for Yvonne. Never mind, there's this chilorio that I made 3 years ago. At the time I wrote that Yvonne liked it, but I'm not sure that was correct. So today I thawed some out and ate it, while Yvonne ate the (better) leftovers.
No, that's really not worth repeating. Like the asparagus at breakfast, I was able to eat it, but I'd prefer not to.
| Wednesday, 31 July 2024 | Dereel | Images for 31 July 2024 |
| Top of page | ||
| previous day |
|
Air conditioner repair
|
Topic: Stones Road house, general, opinion | Link here |
Up at 6:15 this morning to receive Nathan Barnes of South West Electrical, who came to replace the inside coil of the old air conditioner. And sure enough, he was there at 7:30.
The weather was appropriate: when he arrived, the outside temperature was round 0°, having just recovered from the minimum -1.9°, and the little Mitsubishi unit, which had been running all night, was struggling to maintain 20°. They replaced things with little difficulty, and I had a chance to take a look at the coil:
|
|||||||||||||||||||||||
The bare metal is damage caused during removal of the coil. But there are two very obvious things to observe: first, once again the thing is clogged with some kind of fibrous mess, and secondly the bottom left hand corner is wet, due to an oil leak. There's not much oil in there, but some gets transported with the refrigerant.
Where's the leak? It doesn't seem to be easy to determine, but clearly it's at this corner:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And what's the story with the fuzz? Where does it come from? It's clearly quite thick, and that's what Tony Nesci removed less than two years ago:
|
||||||||||||||||||||||||
So where does it come from, and what is it? More photos don't show very much:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Both Tony and Nathan thought that it was dog hair (I hear H.L.Mencken laughing in the background), but there's no way it could have got into the return air in any quantity. But on the other hand, it also doesn't seem that it could be spider's webs: for that it's too firm. Animals?
By chance, when Nathan was measuring temperatures coming from the ducts, this lump came out:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
It wasn't there before, and since it was the closest duct to the coil, it seems that it was related to the repair. But what are those black lumps in the fuzz? Could it be mouse droppings? It would make sense that the warm coil would attract animals, so maybe everything happened up against the coil.
The circumstances make today's temperature graphs particularly interesting:
They're easier to see on the big image (click on the small one). The orange line shows inside dew point, but by chance it also shows the effect of the de-icing cycles between 2:00 and 8:00 better than the blue inside temperature line. And at almost exactly midday they started running the old air conditioner in cooling mode, showing a noticeable drop, again more obvious in the dew point curve. The subsequent increase in room temperature was due to the outside temperature.
An interesting observation from Justin, Nathan's apprentice: the inverter in the garage makes what appears to be a continuous high-pitched whine. My old ears don't hear that. I wonder if it's normal.
|
Ten year itch
|
Topic: Stones Road house, general | Link here |
It has been nearly 10 years since we started construction of the house, and things are continuing to fail. The front gate was installed almost exactly 10 years ago, some time between 9 and 16 August 2014, as the house photos show. It hasn't weathered well: it's almost completely rusty. And today it wouldn't shut:
|
|||||||||||||||||||||||
Oh. That doesn't look right. It has fallen off its mounting pin:
|
|||||||||||||||||||||||
OK, lift it up and put it back. No, that won't work:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The bottom of the tube has been torn out. We need a new gate. Same again, or something better?
|
hydra progress
|
Topic: technology, opinion | Link here |
More work on hydra today. What do I need to do with the configuration for server 1? Fired it up, and it came up, just with the monitors in the wrong sequence. And nvidia-settings didn't want to know about the fourth monitor until I insisted.
OK, moving the monitors is almost straightforward, except now there are no X display 1 and friends. I have to tell nvidia-settings the (sometimes ambiguous) name of the monitor. Save the config file, and we're done. Start the server with the new config file.
No devices detected.
Fatal server error: no screens found(EE)
Huh? What went wrong there? I just rearranged the monitors:
- Option "metamodes" "HDMI-0: nvidia-auto-select +0+0, HDMI-1: nvidia-auto-select +5760+0, DP-3: nvidia-auto-select +7680+0, DP-0: nvidia-auto-select +1920+0"
+ Option "metamodes" "HDMI-0: nvidia-auto-select +9600+0, HDMI-1: nvidia-auto-select +5760+0, DP-3: nvidia-auto-select +0+0, DP-0: nvidia-auto-select +1920+0"
Ah, but that wasn't enough for nvidia-settings. It removed vital configuration information:
@@ -45,18 +43,16 @@
Driver "nvidia"
VendorName "NVIDIA Corporation"
BoardName "NVIDIA GeForce RTX 3060"
- BusID "PCI:1:0:0"
EndSection
Replace that, and all was well. But that must affect a lot of people. Why haven't they fixed it?
In passing, there's something very strange about this configuration file. It only defines one monitor! And that monitor doesn't even exist any more. I idly wonder whether I couldn't remove all monitor definitions.
|
Fixing firefox
|
Topic: technology | Link here |
One of the problems with a 3840×2160 resolution monitor is that web browsers like firefox display everything too small. Spent some time today changing the default font sizes, with less than stellar results.
Surely there's a better solution? A web search pointed me at the configuration parameter layout.css.devPixelsPerPx in about:config. By default it's set to -1.0, but it can be set to between 1.0 and 2.0 to scale all output. I've found that 1.4 works for me, at least for the moment.
Do you have a comment about something I have written? This is a diary, not a “blog”, and there is deliberately no provision for directly adding comments. It's also not a vehicle for third-party content. But I welcome feedback and try to reply to all messages I receive. See the diary overview for more details. If you do send me a message relating to something I have written, please indicate whether you'd prefer me not to mention your name. Otherwise I'll assume that it's OK to do so.
| Top of page | Previous month | Greg's home page | Today's diary entry | Next month | Greg's photos | Copyright information |
