|
| Greg's diary |
| November 2023 |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Wednesday, 1 November 2023 | Dereel → Ballarat → Dereel | Images for 1 November 2023 |
| Top of page | ||
| next day | ||
| last day |
|
Into Ballarat again
|
Topic: general, health, opinion | Link here |
Into Sebastopol and Ballarat today, primarily to take Yvonne in to pick up her car, but also had a haircut.
Then on to SpecSavers to pick up the glasses they forgot to tell me about. There's something funny about picking up glasses: they always seem so much better than the old ones. And so it was today.
Bridge Mall is looking worse all the time:
I've been puzzled about the decision to allow vehicle traffic again, but if it means that people can drive up to a shop and park in front of it, it might be of some use. That assumes that there are enough parking places, which seems unlikely. Maybe 5 minute parking places would be a good idea, but I'll wait until I see the results.
|
Australian and Chinese food
|
Topic: food and drink, language, opinion | Link here |
Then to the Fruit Shack, where I found an interesting distinction:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Australian choi sam and (real) choi sam? The “Australian” choi sum looks more like pok choi, and even before I knew the distinction, I didn't want it.
Another surprise was this bag of some kind of snack:
|
|||||||||||||||||||||||
I tend to try to decipher the names of Korean food, and I had come up with Jio Ri Pong, but it was written underneath in Roman script: “Jolly Pong”, clearly a name not thought up by a native English speaker. And the l in “Jolly” can't be represented in Hangul.
|
The daily font pain
|
Topic: technology, opinion | Link here |
I really should be doing something else, but my difficulties with X fonts are still getting on my nerves. Why can't I find a font that provides representations for all characters that I am likely to find? And how do fonts add up? I've installed additional fonts for East Asian character sets, and they seem to get found automatically. I'm missing a lot of basic understanding.
One thing that really irritates me is that Microsoft has none of these problems. Why not? Off searching, and discovered: https://forums.freebsd.org/threads/x-org-ms-truetype-fonts.617/ contains:
just copy the ttf files from a windows install (c:\windows\fonts\ , I think) to /usr/local/share/fonts/ . This should work for the new C fonts (Candara, Calibri, Consolas etc) as well, if you've got them.
Is that legal? Worth checking anyway. This page suggests that it might be. It also contains the interesting information:
They contain all 652 glyphs defined in WGL4 and are available in the following encodings: ...
So there's something to follow up there.
But there must be some way to find out which characters each font represents. Off looking and came up with How to view a TTF font file?, which recommends display, part of ImageMagick. But that's not really much help:
=== grog@hydra (/dev/pts/15) ~ 38 -> display /usr/local/share/fonts/noto/NotoSans-Regular.ttf
|
|||||||||||||||||||||||
That's the Chinese font, but it only shows the first page. That's not much use.
The first link also contained a link to Font Manager, so I tried that:
=== root@hydra (/dev/pts/4) /usr/local/share/fonts/urwfonts 197 -> pkg search font-manager
font-manager-0.8.8_4 Font management application for the GNOME desktop=== root@hydra (/dev/pts/4) /usr/local/share/fonts/urwfonts 198 -> pkg install font-manager
...
==> Running trigger: glib-schemas.ucl
Compiling glib schemas
==> Running trigger: gtk-update-icon-cache.ucl
Generating GTK icon cache for /usr/local/share/icons/hicolor
==> Running trigger: desktop-file-utils.ucl
Building cache database of MIME types=== root@hydra (/dev/pts/4) /usr/local/share/fonts/urwfonts 201 -> man font-manager
...
DESCRIPTION
font-manager is intended to provide a way for average users to easily
manage desktop fonts, without having to resort to command line tools
Those triggers look dangerous. And savour that “without having to resort to command line tools”. Yes, it's nice and graphic and not as archaic as xfontsel, but it didn't help either.
Then there was How to find out which unicode codepoints are defined in a TTF file?. That proves to be part of TeX, and it wasn't installed on hydra. It was on eureka, however, and it actually works:
=== grog@eureka (/dev/pts/4) ~ 197 -> otfinfo -u /usr/local/share/fonts/Lohit/Lohit-Tamil-Classical.ttf | less
...
uni0BAA 75 u0BAA
uni0BAE 76 u0BAE
uni0BAF 77 u0BAF
(etc)
But that's far too hard to interpret. Surely there must be something that just displays the characters.
| Thursday, 2 November 2023 | Dereel | Images for 2 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
New mi udang
|
Topic: food and drink, opinion | Link here |
I've found a new kind of mi udang (prawn noodles) at the Fruit Shack:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Tried them today. Yes, despite the name they're not as pedas as the Tean's paste, but possibly not as interesting in flavour either. They're somewhat irritating in that a sachet produces 1400 ml of broth, meaning either 5 portions of 280 ml (too little) or 4 portions of 350 ml (too much).
|
More lawn mower pain
|
Topic: gardening, opinion | Link here |
Paul Donaghy along today, along with some suggestions from Jono and Johno, the makers of the engine. It seems that the electrical connection on the carburettor that we saw last time was a petrol on/off switch. That would have to mean that it is only on when the engine is running, so a loose connection could be the problem. He took out the spark plug, replaced it and... it ran! Paul thinks it's the spark plug, I think it's the wiring.
But then he refilled the tank and the old behaviour happened again. Why? This is really puzzling.
|
Return of old memories
|
Topic: history, general, opinion | Link here |
One of the highlights of my life was the “Asia trip” in April to June 1967, when my father and I drove from Singapore to London. It was supposed to be one of many trips that we had planned, but in the end we didn't repeat it.
But of course the matter interested me greatly, and recently I discovered a film series “Last Overland”, about a journey from, well, Singapore to London. And the name Tim Slessor cropped up. Hmm, that name rings a bell. It churned around a couple of times, and then I recalled a book: “First Overland”.
Oh, that sounds similar. Yes, I think I had the book. What did it look like? Into the library where my car and travel books were. “Round the world in a 2CV”, in German, but no “First Overland”.
But I recall a detail from the book: the cars were equipped with winches, and if I recall correctly, this was reported as wenches in an Indian newspaper article of the time. So I must have the book. But where is it?
Watched the first episode of “Last Overland”, which wasn't very convincing. The biggest detail was that Tim Slessor, who had supposed to be taking part, was a non-starter due to health issues. At the age of 87 that's not surprising, but the hype was misleading. And both the map of the route and the videos looked unconvincing.
But wait, there's more! A video called “First Overland” was released a few year ago, and it's available on YouTube in high-resolution 640x480, so I looked at that. Not quite what I was expecting: some clips of the (colour) film that they took at the time, along with some which looks like it was taken much later:
|
|||||||||||||||||||||||
|
|
|||||||||
I took the first photo on 6 May 1967. Looking at the trees along the sides, the other one must have been taken some years later. The number of people there tends to confirm that.
But there was also a view of the book. Yes, that's just like the one I remember. Where can it be? Into Yvonne's bedroom, which doubles as her personal library, and looked in the English fiction books. Yes! That's the wrong place, of course, but I have found it.
Did I ever read it? I can't remember much of it. But it looks like it's worth reading (again?). The real thing of interest is how details of the book returned to me gradually over a couple of days: first Tim Slessor, then the name of the book, then the appearance of the book.
|
Bruno the mountain goat
|
Topic: animals | Link here |
Bruno continues to explore heights. He got up in front of the TV again today, looking at the cupboard that he jumped on to on Tuesday. So I removed the flash unit, but in the end he decided to jump onto the TV, not a good idea: clearly he couldn't sit on top of it, and I had to help him down. Hopefully he won't try that again, though I suspect we'll see him on the curtain rails in the not-too-distant future.
| Friday, 3 November 2023 | Dereel | Images for 3 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
hydra: next steps
|
Topic: technology, opinion | Link here |
Why am I not getting more done with hydra? One obvious reason is that I want to get things right rather than getting them done quickly. And this font stuff is still irritating me. But it'll have to go on the back burner for a while. It's time to finally get the X configuration working. At least I should be able to use the current configuration for server 1, and all I really need is another monitor for the transition period. And Chris Bahlo has one that she can lend me for a while, so that's the next step to plan.
|
Information overload
|
Topic: general, opinion | Link here |
Another reason why I'm not getting much done, of course, is sheer information overload. Just following the crises in the world takes so much time, and I can measure my overload by the number of messages in my mail inbox; my criterion is that it be not more than one page, or 67 messages, and preferably not more than 10. Lately I've been running into a second page far too often.
|
Bank of Melbourne bugs: insight?
|
Topic: technology, general, opinion | Link here |
Bill from Devin Lafranchi today. Pay by bank transfer. I have enough in my ANZ account, so went through all the pain of setting up a new payee and paying him.
“We'll need confirmation that it's you. We'll send an email to axxxxxx@lemis.com”. Maybe they did, too, but of course that's not my registered email address; they're just too polite to write the real address. But they have it on record, and I get mail from them from time to time. After 5 minutes, nothing. This happens far too often (every time, it seems). Cancel.
OK, Bank of Melbourne it is. Logon fail: “Please match the requested format” on the “Security number” field. Dammit, this has been happening for over 5 years! And they still haven't fixed it.
But I need to pay. Called up their “help” line on 132266 and spoke to David, explained my situation—he even seemed to understand. Yes, they've been having problems recently, and he does apologise. But that doesn't cut it. Please connect me to the web people so they can tell me what's wrong.
To my surprise, he did, and I was connected more quickly than the 10 minutes I had to bear listening to excruciating “music” while waiting for him. Spoke to Matt, who did seem to understand, and after some discussion came to the conclusion that it could be due to incorrectly saved parameters, specifically saving the “security number” as the •••••• that it displayed. At least he took my suggestion on board to have a button to display the real number. That doesn't explain why the page is dead in the water after such an error message, that it doesn't even try to reauthenticate. But certainly after restarting the page and entering the number correctly, it logged in correctly.
That could be coincidence, of course—I've seen login failures followed by success for no obvious reason before—but maybe this was the missing detail. In any case, it was nice to talk to somebody who seemed to have a brain. I wonder how much harm companies are doing themselves by having “help” lines that don't.
| Saturday, 4 November 2023 | Dereel | Images for 4 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
X on hydra
|
Topic: technology, opinion | Link here |
Enough procrastination! Today I continued my work on getting X to run on hydra. In principle it should have been a no-brainer: I already had a configuration file and most monitors in place. Only one monitor was missing. Well, I had the 7" monitor for my cameras, but that was as good as useless. So today Yvonne went and picked up a loaner from Chris Bahlo.
First, though, I've been migrating my photo processing from eureka to hydra, and I need a number of programs. One is s3cmd, which loads my photos to DigitalOcean. It's in the FreeBSD Ports Collection, but they have chosen to call it net/py-s3cmd. That installed without problems, but it wouldn't run without a configuration file ~/.s3cfg. That's easy: copy it from eureka. I can't recall having set anything specific, though it does contain the destination URL.
No HDMI input, so I had to swap for a DVI output. Reconfigure with nvidia-setup, this time really saving the configuration file. The result was interesting:
=== root@hydra (/dev/pts/4) /etc/X11 228 -> diff -wu xorg-1.conf.backup xorg-1.conf
--- xorg-1.conf.backup 2023-10-22 14:27:24.881413000 +1100
+++ xorg-1.conf 2023-11-04 16:37:37.925309000 +1100
@@ -1,11 +1,12 @@
-# nvidia-xconfig: X configuration file generated by nvidia-xconfig
-# nvidia-xconfig: version 525.116.04
+# nvidia-settings: X configuration file generated by nvidia-settings
+# nvidia-settings: version 470.86
Section "ServerLayout"
Identifier "Layout0"
- Screen 0 "Screen0"
+ Screen 0 "Screen0" 0 0
InputDevice "Keyboard0" "CoreKeyboard"
InputDevice "Mouse0" "CorePointer"
+ Option "Xinerama" "0"
EndSection
The difference in version number between nvidia-xconfig and nvidia-settings is interesting. And the Screen entry has gained additional numbers. Most interesting, though, is the Xinerama option, whose absence I had already noted.
That worked fine until the screen saver cut in. When it came back, it didn't want to know about the changes: I couldn't even load them again: I had to restart X. Why?
Moving on, there was a change for a monitor entry because, well, I changed a monitor. And then
@@ -38,7 +42,7 @@
Identifier "Device0"
Driver "nvidia"
VendorName "NVIDIA Corporation"
- BusID "PCI:1:0:0"
+ BoardName "NVIDIA GeForce RTX 3060"
EndSection
Section "Screen"
@@ -46,6 +50,12 @@
Device "Device0"
Monitor "Monitor0"
DefaultDepth 24
+ Option "Stereo" "0"
+ Option "nvidiaXineramaInfoOrder" "DFP-0"
+ Option "metamodes" "HDMI-0: nvidia-auto-select +0+0, DP-1: nvidia-auto-select +3840+216, HDMI-1: nvidia-auto-select +1920+0, DP-3: nvidia-auto-select +5206+0"
+ Option "SLI" "Off"
+ Option "MultiGPU" "Off"
+ Option "BaseMosaic" "off"
SubSection "Display"
Depth 24
EndSubSection
There are a number of entries here that I should probably research. Clearly the long metamodes line gives a sequence, but how? The outputs appear to be called HDMI-0, HDMI-1, DP-1 and DP-3. Apart from the discrepancy in numbering, they don't match the sequence of monitors, which have HDMI, HDMI, DVI and DVI. Still, it seems to work. My guess is that the real information is the geometry that is also specified.
OK, start with my startx1 script, which sets options:
startx -- :1 -config xorg-1.conf -listen tcp
That worked, and I got a display along with a window manager, but the menus didn't work. That proved to be due to inadequacies in my convoluted .xinitrc file.
And that got me thinking. I have files like ~/.Xdefaults-eureka, ~/.Xdefaults-tiwi and ~/.Xdefaults-dereel, but no ~/.Xdefaults-hydra. Obvious: create it.
But that's tacky. I have X configuration files in my home directory, in ~/.fvwm and in ~/keyboard. I even found remains of older attempts:
=== grog@eureka (/dev/pts/4) ~ 225 -> l ~/fvwm
-rwxr-xr-x 1 grog lemis 86 10 Jun 2000 /home/grog/fvwm=== grog@eureka (/dev/pts/4) ~ 226 -> cat ~/fvwm
fvwm2 -s -display freebie:0.0&
fvwm2 -s -display freebie:0.1 -f .fvwm2rc-freebie:0.1&=== grog@eureka (/dev/pts/4) ~ 234 -> l ~/.fvwm2rc
-r--r--r-- 1 grog lemis 22,704 4 Jun 2010 /home/grog/.fvwm2rc
I really need to put them into a single directory. ~/X sounds like an obvious choice. Should I do it now or think it through?
And then there's this line in ~/.xinitrc:
kbdcontrol </dev/ttyv0 -r 250.46
That's been there at least since 7 December 2002, and at the time I needed to select /dev/ttyv0 to get it to work at all. And it should work if I start X from /dev/ttyv0, as I normally do. Only here it didn't. Is there a more robust way?
And then there's this:
[fvwm][GetWindowSizeHints]: <<WARNING>> The application window (id 0x800003)
"NVIDIA X Server Settings" has broken size hints (inconsistent with current size).
fvwm is ignoring those hints.
Harmless enough, but it's nice to know that I'm not the only person having pain here.
On the positive side, I was able to connect the servers eureka:0 and hydra:1 with x2x, though the mouse moved rather faster than I wanted. Potentially something needs looking at there.
|
More firefox pain
|
Topic: technology, opinion | Link here |
Since my last update of firefox, to version 118.0.2, things have become much less stable, and once again I'm experiencing multiple crashes per day. That's presumably because I'm running across the network, and hopefully it will go away when the hydra migration is done. But I should really try to find out how to run multiple instances of chrome on one server.
| Sunday, 5 November 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
hydra X installation, day 2
|
Topic: technology, opinion | Link here |
Spent most of the day with X on hydra. It was even more work than I expected.
First updating my personal X configuration files on hydra: .xinitrc, .Xdefaults, and the non-existent .fvwm/fvwm2rc-hydra:1.0/ needed to start things properly. Also ~/xe, which along with its helper file ~/xemarkers.
Restart X. “Couldn't find any screens”.
Huh? That was just running. More examination and discovered that the original xorg.conf file worked, but the new one, updated by nvidia-settings, didn't. More playing around with nvidia-settings, which was also not very cooperative:
=== grog@hydra (/dev/pts/2) ~ 1 -> nvidia-settings
nvidia-settings: Fatal IO error 22 (Invalid argument) on X server hydra:1.0.
Huh? What's happened now?
=== grog@hydra (/dev/pts/2) ~ 2 -> echo $DISPLAY
hydra:1.0=== grog@hydra (/dev/pts/2) ~ 3 -> DISPLAY=:1=== grog@hydra (/dev/pts/2) ~ 4 -> nvidia-settings=== grog@hydra (/dev/pts/2) ~ 5 -> DISPLAY=hydra:1=== grog@hydra (/dev/pts/2) ~ 6 -> nvidia-settings
nvidia-settings: Fatal IO error 22 (Invalid argument) on X server hydra:1.=== grog@hydra (/dev/pts/2) ~ 7 ->
In other words, nvidia-settings is too polite to run over an Internet domain socket. And of course, far too polite to say what the real issue is.
Finally I ran it and came up with a new updated file, which differed slightly from the one I made yesterday:
=== grog@hydra (/dev/pts/15) /etc/X11 66 -> diff -wu xorg-1.conf-broken-by-nvidia xorg.conf
--- xorg-1.conf-broken-by-nvidia 2023-11-04 16:37:37.925309000 +1100
+++ xorg.conf 2023-11-05 13:52:14.275346000 +1100
@@ -1,6 +1,9 @@
# nvidia-settings: X configuration file generated by nvidia-settings
# nvidia-settings: version 470.86
+# nvidia-xconfig: X configuration file generated by nvidia-xconfig
+# nvidia-xconfig: version 525.116.04
+
Section "ServerLayout"
Identifier "Layout0"
Screen 0 "Screen0" 0 0
@@ -13,6 +16,7 @@
EndSection
Section "InputDevice"
+
# generated from default
Identifier "Mouse0"
Driver "mouse"
@@ -23,13 +27,13 @@
EndSection
Section "InputDevice"
+
# generated from default
Identifier "Keyboard0"
Driver "keyboard"
EndSection
Section "Monitor"
- # HorizSync source: edid, VertRefresh source: edid
Identifier "Monitor0"
VendorName "Unknown"
ModelName "Ancor Communications Inc VS248"
@@ -43,6 +47,7 @@
Driver "nvidia"
VendorName "NVIDIA Corporation"
BoardName "NVIDIA GeForce RTX 3060"
+ BusID "PCI:1:0:0"
EndSection
Now isn't that puzzling? The same program working on the same configuration file, and it came up with different results! The most important one is clearly the BusID option that I've tripped over before, but why the different comments at the top and the empty lines in the InputDevice sections?
So finally I had an X display across 4 monitors and with a responsive keyboard, mouse and window manager. Time to finally process yesterday's house photos. rdesktop didn't start. Oh. First I need to wake the target machine (distress). For that, I need /etc/ethers. And it still didn't work: permissions errors on /dev/bpf*. I need wake to be setuid:
=== root@hydra (/dev/pts/4) /etc/X11 235 -> wh wake
5296747 -r-xr-xr-t 1 root wheel 7976 18 Oct 14:06 /usr/sbin/wake=== root@hydra (/dev/pts/4) /etc/X11 238 -> chmod 4555 /usr/sbin/wake=== root@hydra (/dev/pts/4) /etc/X11 239 -> wh wake
5296747 -r-sr-xr-x 1 root wheel 7976 18 Oct 14:06 /usr/sbin/wake
Next, connect to distress:
Connecting to distress
ATTENTION! The server uses and invalid security certificate which can not be trusted for
the following identified reasons(s);
1. Certificate issuer is not trusted by this system.
Issuer: CN=distress
...
Do you trust this certificate (yes/no)? yes
And fortunately that only needed to be done once. Sadly, though, my configuration spreads the distress window over all four screens, not what I really wanted, so until I set up the :0 server I'll do that on eureka
Next, set up a housephoto-notes file. Emacs converts c-f into c-s-f again, like it did last month! Grrr!
Ha ha, only joking. That's what happened if I use the keyboard on eureka, connected via x2x. Using an identical keyboard on hydra, things worked normally.
But that was only part of it. Running Hugin, I moved the fast panorama preview window to another screen, and some of the display remained in the old place, something that I'll need to revisit. Also, the arrow keys didn't work. But apart from that and the uncomfortable access to the display, things worked. And I had the satisfaction of having top show me nona using 3213% CPU (and thus max out the machine). I also used up to 40 GB of memory, still only a small fraction of the machine's 192 GB.
Or at least I thought all was well. The results were not encouraging, and I had to process them again on eureka (run the cursor over an image to compare it with its neighbour):
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That looks like a problem with Hugin, specifically enblend, not with X or hydra. The version on eureka is 4.1.4, and on hydra it's 4.2. Each pair of images was created with identically the same .pto file.
Also, the arrow keys on the local keyboard didn't work. Checking with xev showed that they were completely incorrectly mapped. How could that happen? The keyboards on each machine were the same (Sun Type 7). But that can be remapped.
And xv didn't seem to work at all. It displayed its standard startup page, but it didn't react to the mouse keys. Why?
So: where am I now? I need to:
Is that all? It seems almost simple.
| Monday, 6 November 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Zoom pain, a year later
|
Topic: technology, opinion | Link here |
Spent fully 30 minutes this morning trying to set up a one-on-one Zoom conference with a FreeBSD team member in the USA this morning. Zoom means camera and microphone, of course, and that's only on bde.lemis.com, my hand-me-down laptop. But I was sent the meeting code by email on eureka. How do I put it into the browser on bde? x2x to the rescue.
And once again I had problems, insurmountable ones. We tried for half an hour, and it wasn't until after the discussion (on the phone) that I discovered that I had exactly the same problem a year ago today. On that occasion, I had to fall back to Microsoft. Until proof of the contrary, I have to assume that it just doesn't work under FreeBSD. And on this occasion, though not last year, it seems that Zoom and x2x got into a fight about cursor ownership, resulting in the cursor running aimlessly across the screen.
The phone call was no better. We both managed to send each other incorrect phone numbers, but even with the correct ones I only got a busy signal. Call to Aussie Broadband: it seems that my lines weren't configured for international calls. Presumably the costs were the issue: 1¢ per minute for the international call, as opposed to 35¢ for 1300 numbers in Australia.
They fixed that and told me to wait half an hour. Yes, I've heard that before. Called immediately, and was connected. For about 5 minutes, then the line dropped. After 3 such incidents, my partner called me and things were OK. I don't suppose I can even blame Aussie, since it was all within the half hour. But hopefully that's the reason.
In passing, Aussie's web site offered me that alternative of “live chat”. Selected that and was told that I would have an estimated wait time of 9 minutes before somebody could be with me. That's not “live” in my book. The callback from Aussie support was faster.
| Tuesday, 7 November 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Strange change in hydra security output
|
Topic: technology | Link here |
Seen in the nightly output from hydra today:
- 6173502 -r-sr-xr-x 1 root wheel 8928 Oct 14 13:46:53 2023 /usr/local/libexec/Xorg.wrap
+ 6172201 -r-sr-xr-x 1 root wheel 8928 Nov 2 13:46:40 2023 /usr/local/libexec/Xorg.wrap
What's that? The man page doesn't say. The DESCRIPTION only describes the permissions needed to run it, something that really needs fixing. But it seems to be something (new) that starts the X server. What starts it? I thought that maybe /usr/local/bin/startx might, but if so, it's well hidden.
And why has the file changed? Ah, right, I did a pkg upgrade.
|
Groggy, the Jew
|
Topic: politics, general, opinion | Link here |
Survey request about the Gaza war from the Times of Israel today. I get the Times newsletters daily, though they add to the information overload. So I filled it out; they wanted extreme checks, including Two-factor authentication with my phone number. And yes, the fact that I'm not in Israel was not an issue.
But they wanted to know if I'm Jewish. Well, in the distant past, yes, so I answered “yes”. Wrong answer. Lots of questions about religious observances and the stream of Judaism to which I adhered, at least giving me the possibility of an honest answer. And then the questions about the war, few by comparison. “What's your opinion about the current war?”. Disgust. And yes, it was there for me to choose.
It will be interesting to see how they report on the survey. In the meantime, they have given me a survey URL to pass on to fellow Jews, gaining me $10 for everybody who fills out the survey. Is that a way to get a representative opinion?
|
More garden stuff
|
Topic: gardening, opinion | Link here |
Jesse Walsh around today for more weeding in the garden. Gradually it's looking a little better. He also brought along a reel of copper wire to wrap around our dead Alyogyne huegelii. I fear he has chosen a poor example to demonstrate his hypothesis.
|
Chromium and multiple profiles
|
Topic: technology, opinion | Link here |
One of the reasons I still use firefox—a browser that I have always hated—is because I couldn't get Chromium to support multiple profiles. But it seems that it does now.
Does it also now work across a network? Tried it out. Not what I expected:
=== grog@hydra (/dev/pts/15) /etc/X11 70 -> DISPLAY=eureka:0.1 chrome
[29146:-1859547136:1107/102603.781897:ERROR:process_singleton_posix.cc(353)] The profile appears to be in use by another Chromium process (60349) on another computer (dereel). Chromium has locked the profile so that it doesn't get corrupted. If you are sure that no other processes are using this profile, you can unlock the profile and relaunch Chromium.
[29146:-1859547136:1107/102603.781911:ERROR:message_box_dialog.cc(146)] Unable to show a dialog outside the UI thread message loop: Chromium - The profile appears to be in use by another Chromium process (60349) on another computer (dereel). Chromium has locked the profile so that it doesn't get corrupted. If you are sure that no other processes are using this profile, you can unlock the profile and relaunch Chromium.
Where did dereel come into it? Is there really a Chromium running on dereel?
=== grog@hydra (/dev/pts/6) ~/Downloads 10 -> l ~/.config/chromium/
total 1
...
drwx------ 2 grog lemis 512 3 Aug 2021 Safe Browsing
-rw------- 1 grog lemis 20,480 25 Jul 2021 Safe Browsing Cookies
-rw------- 1 grog lemis 0 25 Jul 2021 Safe Browsing Cookies-journal
lrwxr-xr-x 1 grog lemis 20 31 Jul 2021 SingletonCookie -> 13443178945937354382
lrwxr-xr-x 1 grog lemis 12 31 Jul 2021 SingletonLock -> dereel-60349
Aaargh! File names with spaces in them! And a two year old reference to a long-dead process on a different machine! Clearly nothing to fall in love with immediately.
|
Boiling what?
|
Topic: general, technology, opinion | Link here |
Question in the German Quora today: Wie lange sollte man seine Eier kochen? On the face of it, that's “how long should you cook your eggs?”. But „Eier“ can mean eggs or balls (testicles). And I've been there before.
| Wednesday, 8 November 2023 | Dereel | Images for 8 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More work overload
|
Topic: general, technology | Link here |
The FreeBSD Core Team doesn't require much work, but when it does, it comes unevenly. And the last couple of days have been a particularly busy time. Once again my mail inbox has overflowed onto a second page.
|
eBay: More security
|
Topic: technology, opinion | Link here |
Mail from eBay today, offering me a thing called a Passkey. To quote:
Passkeys are an alternative to passwords that allow you to sign into your eBay account using biometric authentication such as a fingerprint, pattern or PIN.
OK, let's think about this:
And then there's the offer of two-factor authentication. If you've got this far, you already have the phone and it's active, so responding to an SMS is trivial. No additional security, only additional effort required.
Yes, security is important. But the issue is clear: mobile phones. I wonder if RFID implants wouldn't be a better way. To steal them you'd need to have the owner, or at least part of him, nearby.
|
All phones down?
|
Topic: technology, opinion | Link here |
I've been having issues with phones lately, notably Monday's call to the USA. I had attributed that to the configuration change to enable international calls But today I had more problems calling a number in Ballarat.
Why? We'll see if they continue, but it seems that Optus had a significant outage that they still haven't completely fixed (nor even identified the cause). “Over 10 million customers and 400,000 businesses are affected by the outage”. 10 million is 40% of the population of Australia, and Wikipedia states “Optus is the second-largest wireless carrier in Australia, with over 10 million subscribers as of 2022”. So that means effectively all customers were affected. Were my problems connected to that outage?
|
Chicken wings revisited
|
Topic: food and drink, opinion | Link here |
I eat deep fried, marinated chicken wings with nasi lemak. The recipe I have been using requires them to be cooked sous vide, which has a couple of disadvantages: it's fiddly, the bags are non-recyclable, and I lose a lot of the juice, which can be used as a broth.
Today was the day to make a new batch. What alternatives do I have? This one, maybe? I do this every few months, but I can't remember how I did it last time. I must have cooked them normally, because I recall doing them twice in the same broth; possibly there was little more than water to start with.
Today, anyway, was time to write things down:
| quantity | ingredient | step | ||
| 1 kg | chicken wings (about 10) | 1 | ||
| 20 g | garlic | 1 | ||
| 10 g | ginger | 1 | ||
| 45 g | light soya sauce | 1 | ||
| 10 g | dark soya sauce | 1 | ||
| water to cover | 1 | |||
| 15 g | dark soya sauce | 2 | ||
| 10 g | garlic | 2 |
Mix the step 1 ingredients, bring to the boil and simmer for 30-40 minutes. If there are more, cook them in the same broth, adding soya sauce and garlic (step 2).
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
| Thursday, 9 November 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
ALDI: Stewed Christmas pastries
|
Topic: photography, food and drink, opinion | Link here |
Yesterday Yvonne went shopping and brought back the ALDI catalogue, as usual. Today's was particularly noteworthy:
|
|||||||||||||||||||||||
That's an el-cheapo casserole made out of aluminium (so that you can't use it on an induction stove), though they call it a sauté pan, something completely different. But look at the content! Baked pastries in the form of a Christmas tree! On the one hand that shows ALDI's problems with the calendar (one of the best reasons I can think of to celebrate something like US Thanksgiving), and on the other hand it shows the photographer's complete lack of understanding of the purpose of the utensil. Bravo, ALDI!
|
More weather station pain
|
Topic: technology, opinion | Link here |
Since my workarounds last month, my weather station has been running relatively reliably. Until today:
./wh1080 starting
Can't read device: Unknown error or Device busy (16)
That's nothing new. It happens about 20 times a day, and I just restart the wh1080 process (I can't recover programmatically). But this time it just kept repeating, and on one occasion I even got:
Can't query database: MySQL server has gone away (2006)
That must be one of the report processes, since wh1080 only writes to the database. But why? Checking showed that the MySQL server was running happily, as it has for decades.
A more interesting case was
./wh1080 starting
Rain prev: 7245, now: 7245, raw 7245.0
Can't read device: Unknown error or Device busy (16)
Here the program got as far as reading some data from the station, but not enough to create a database record. How does this happen?
OK, disconnect the USB cable and reconnect it? No help. Power cycle the station. Yes! So it is a weather station issue, not FreeBSD or my coding.
But then it happened again, and this time nothing would help. In the end, I just gave up; I have enough other things to do. Since the beginning of the week I haven't even found time to look at the X configuration on hydra.
But right, what about hydra? At some point I need to migrate the software to it. That should be relatively straightforward. First install MySQL on hydra, which came up with the interesting message:
MySQL80 has a default /usr/local/etc/mysql/my.cnf,
remember to replace it with your own
or set `mysql_optfile="$YOUR_CNF_FILE` in rc.conf.
Oh. Yes, there was something. But there's no /usr/local/etc/mysql/my.cnf on eureka. A quick search showed:
=== grog@hydra (/dev/pts/7) ~ 5 -> ls -lrt /eureka/home/var/db/mysql/my.cnf /eureka/usr/local/my.cnf /usr/local/etc/mysql/my.cnf
-rw-r--r-- 1 mysql mysql 39 3 May 2006 /eureka/home/var/db/mysql/my.cnf
-rw-r--r-- 1 root wheel 943 13 Apr 2019 /eureka/usr/local/my.cnf
-rw-r--r-- 1 root wheel 2039 4 Nov 13:28 /usr/local/etc/mysql/my.cnf
What's in there? Bloat? Maybe, but also lots of comments that I need to read, and only possibly stuff that I need to change. It's interesting to note how long it has taken to get the file in the right place, though the first one dates back to when I worked for MySQL, so maybe that was my problem.
Still, let's build the weather station software first. Now we have this Clang compiler with the ugly (but monochrome) error message reporting. And sure enough:
cc -g -Wall -Wno-parentheses -I /usr/local/include -c -o local-compare.o local-compare.c
local-compare.c:175:20: warning: format specifies type 'int' but the argument has type 'time_t' (aka 'long') [-Wformat]
local_time,
^~~~~~~~~~
...
cc -lm -L /usr/local/lib -L /usr/local/lib/mysql -lusb -lmysqlclient local-compare.o util.o db.o -o local-compare
ld: error: duplicate symbol: mysql
>>> defined at local-compare.c:29
>>> local-compare.o:(mysql)
>>> defined at db.c:31
>>> db.o:(.bss+0x8)
Oh. Yes, that looks like a day-one bug that had slipped through for nearly 14 years. What is local-compare.c anyway? But my available time was up, so I got no further. For the time being, the weather station is offline.
| Friday, 10 November 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Where does the time go?
|
Topic: general, opinion | Link here |
It's been 4 days since my last configuration work on hydra. What have I been doing? Yes, there has been some work, but I think it's really the effect of many things happening together. About the best I can say for today is that I have managed to reduce the size of my inbox. Can I now finally get back to hydra?
|
Weather station insights
|
Topic: technology, general, opinion | Link here |
It's really irritating that the weather station has died now, just as the weather starts to get interesting. Can it be that the USB interface (or whatever it is) has recovered in the meantime? Tried starting it again on eureka, with the same results.
OK, shoot down the window to kill all the processes that won't respond to a simple ^C and try again. No go. Check with ps. Dozens of dorun processes! dorun is a simple script that ensures that all the other processes keep running, and I had thought that it would go away if I shut the xterm window, but it didn't.
OK, shoot down all the doruns and try again. Success! And to make my point, I actually found 7 records overnight in the database. So this particular issue is at least because I don't know how to write scripts that stop all processes on ^C. Something to do with process groups?
And yes, today's top temperature was 36.9°.
| Saturday, 11 November 2023 | Dereel | Images for 11 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
50 years since COBOL
|
Topic: history, technology, opinion | Link here |
In my calendar this morning:
Nov 11 Greg Lehey writes GOPU, 1973
What's so special about GOPU? It was the only COBOL program I ever wrote, and it wasn't very clever. It wasn't even completely COBOL: it was an extension designed to talk to a CODASYL database. But I, the master assembler programmer, hated the idea of having to write in COBOL, and I took it out on the layout of the program and the identifiers. Things like code reviews didn't happen in those days: it worked, and that was enough. But 50 years!
|
More hydra pain
|
Topic: photography, technology, opinion | Link here |
House photo day today, time to play around with Hugin on hydra again.
Not a success. In fact, an absolute failure. I couldn't even start Hugin! No obvious reason, just a couple of meaningless error messages followed by:
/usr/local/share/hugin/data/plugins/top_five.py
CAT:Control Points
NAM:keep 5 CPs per image pair
fails @api-max
/usr/local/share/hugin/data/plugins/woa.py
CAT:Control Points
NAM:Warped Overlap Analysis
fails @api-max
What does that mean? Did some package upgrade break this horribly fragile python? Still, no time to do anything today, so I processed the photos on eureka instead. And even there things didn't go smoothly: one of the panoramas came out upside-down. I'm wondering if it's not due do the control point detector finding spurious control points on unrelated images. It seems that there's a --linearmatch option to cpfind:
--linearmatch
Enable linear images matching (default : all pairs)
What does that mean? Ah, isn't it obvious? But elsewhere in the cpfind man page there's an explanation:
Linear match
This matching strategy works best for single row panoramas:
cpfind --linearmatch -o output.pto input.pto
This will only detect matches between adjacent images, e.g. for the 5
image example it will matches images pairs 0-1, 1-2, 2-3 and 3-4. The
matching distance can be increased with the switch --linearmatchlen.
E.g. with --linearmatchlen 2 cpfind will match a image with the next
image and the image after next, in our example it would be 0-1, 0-2,
1-2, 1-3, 2-3, 2-4 and 3-4.
That should be worth trying. It also has the advantage that it should be faster. But, once again, not today.
|
More garden stuff
|
Topic: gardening | Link here |
Jesse Walsh along today for yet more garden work. He's an energetic worker, and he gets things done fast, but today he still wasn't finished. I'm really happy that I don't have to do it myself.
He also took a sickly looking Strelitzia reginae and planted it on the other side of the house:
|
||||||||||||||||||||||||
It's not clear why, and he didn't get round to explaining. Clearly the place is inappropriate, since the wildflower to the left would completely envelop it next year. Could it be that the Strelitzia prefers shade?
|
More lawn mower pain
|
Topic: gardening | Link here |
Paul Donaghy along today to pick up the lawn mower, which is still misbehaving: after filling up the tank, and only then, the engine ran unevenly and kept cutting out. Juha Kupiainen thinks that maybe the petrol feed solenoid might be sticking. But why only after filling? Today it occurred to me: when the engine is cold, the solenoid works normally. When it's running, it doesn't need to change position. But to fill the tank, it needs to turn off again, and when it's done, it's still hot, and it sticks.
Today, though, it didn't work like that. It started misbehaving while still cold. So Paul took it with him to the Men's shed, where they're hoping for somebody to fix it.
|
Where are my pig's ears?
|
Topic: rant, general, animals, opinion | Link here |
Every day Yvonne gives the dogs a treat, dried pig's ears. We get them from Bendigo Discount Pet Supplies, and I ordered a new batch last month.
Where are they? Off to check. They've been waiting for us at Napoleons since 25 October!
|
|
|||||||||
What's wrong with this picture?
We weren't informed. Normally we should get a notification in our mailbox, though I've noted that they've been sloppy about that in the past.
By now they should have been returned to the sender (2 weeks, I think). Fortunately, Yvonne called and found that they were still there.
We've picked up other things in that time frame, but nobody told us about them.
AusPost has further worsened their tracking information: “Tracking history” is obviously too understandable, so they've changed it to “Journey”.
|
|
|||||||||
They still spell “Napoleons” as “NAPOLEON”. You'd think that the post office would know the names of all towns, but especially those with a post office.
|
|
|||||||||
The information contradicts itself. It's awaiting collection, but it includes delivery information:
|
|
|||||||||
What a mess!
|
Ducks in the dam
|
Topic: animals, photography | Link here |
Yvonne has been reporting for some days now that there's a big family of ducks round the dam. This morning they were particularly prominent.
So I off down to the dam to take a look. They had moved on, but I saw a few on the other side:
|
|||||||||||||||||||||||
Moved around to get a better view, but they saw me. All I got was a photo worth of the Facebook group Crap bird photography:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
Latest Academia nonsense
|
Topic: technology, general, opinion | Link here |
In the mail today:
27 N 11-11-2023 To groggyhimself@l ( 957) Academia.edu N You have 1 highly engaged reader
What's a “highly engaged reader”? How can I find out? Read the message and I get:
You have 1 highly engaged reader
A total of 7 people have read your papers on Academia.
Click here to see who: https://www.academia.edu/keypass/bm1K...
That goes on to tell me that, for only $5, I can find out who the “highly engaged” reader is. More to the point, though: only 7 people? That doesn't sound encouraging, and I've lost count of the number of this kind of message I have received.
| Sunday, 12 November 2023 | Dereel | Images for 12 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Another Pelargonium?
|
Topic: gardening | Link here |
A few hundred metres north of our driveway there are some wild Pelargoniums:
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
They've been there for as long as we can remember. But today, for some reason, Yvonne overcame her dislike of pink (“it's really violet”) and decided she wanted some, so I broke off a stem for her. They do have the advantage that they're hardy, clearly what we need here.
| Monday, 13 November 2023 | Dereel | Images for 13 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
KL Hokkien Mee mods?
|
Topic: food and drink | Link here |
I've been freezing the broth from cooking Wednesday's chicken wings in 50 g portions, the amount I need to make the sauce for KL Hokkien Mee. But I ended up with a non-integral multiple of 50 g: the last portion was 70 g.
What do I do? Plenty of options, of course, but what would happen if I made the sauce with 40% more ingredients. This:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Is it worth it? I don't think so. But it was worth comparing.
|
More lawn mower fun
|
Topic: gardening, opinion | Link here |
Paul Donaghy back with our lawn mower, which he had taken to the Men's shed on Saturday. Had it been repaired? No. The problem had gone away into
hiding. He had mowed “several cricket pitches” with it, and it worked fine. OK, off he
went to mow the lawn. He made about 30 m and it started with the same old problems.
What is it? The more I look at this engine, the more it confuses me:
|
|||||||||||||||||||||||
That's the carburettor at bottom left, with the bronze-coloured solenoid on the right. That's the thing that Paul had taken apart and “cleaned” (it hadn't needed cleaning) on Saturday. But then there's this black thing at top centre. Paul tells me that it's some pressure thing to ensure something to do with fuel flow.
So: the first thing is that I don't understand the construction. Off to search the web for descriptions, of which I found none. But the “surging” (most popular term) is a relatively common problem, and normally it seems to be related to dirt in the idle jet. Could that be the case here too? It doesn't match the “now you see it, now you don't” nature of the problem. But it can't harm to check. Where is it?
Bloody lawn mowers! They're almost as unreliable as my weather station.
|
More bulbs
|
Topic: gardening | Link here |
Yvonne received a package from Jane Ashcroft today:
|
|||||||||||||||||||||||
If I have this right, they're Haemanthus coccineus. One (the small one?) might be Haemanthus albiflos. Now to find a place to put them.
|
Hugin on hydra, again
|
Topic: technology, photography, opinion | Link here |
So what's the issue with Hugin on hydra? So tar I have:
So somehow it must have something to do with X on hydra. By chance, the latest version of Hugin has just been released, and the FreeBSD port is available, so built that—for once with no problems—and tried it. It stopped in the same way, as I had expected.
OK, start writing a message to the Hugin developers list. Test again to show exactly what appeared on the screen and—it worked!
Dammit, I hate these bugs that go into hiding. And that's the second today. OK, create an xorg.conf file for hydra:0 (four screens), calm down a bit, and hope that I'll get it to work tomorrow.
|
Bad grammar makes me sick
|
Topic: language, rant, opinion | Link here |
It's no secret that I hate abuse of grammar. But now I discover that I'm not alone. This article vindicates me: bad grammar is bad for your health.
| Tuesday, 14 November 2023 | Dereel | Images for 14 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Another monitor fade
|
Topic: technology | Link here |
Over the last couple of months, my old “Matrix” monitor has given cause for concern, briefly turning off or distorting the display, only once per day. But it has behaving for some time now. The last time I noted any problem was on 26 October.. Today it gave me a distorted display, only for a few seconds, but 4 times in a row. And then nothing again...
|
Lawn mowing, again
|
Topic: gardening | Link here |
When I had my fun with the lawn mower yesterday, our neighbour Garry Marriott was looking on. And today he brought a short-term solution: he brought his mower over and mowed the lawn for us.
His mower looks very much like ours, also Husqvarna, so I took a look. An original motor that looks nothing like ours. And also, nothing like my experience, he has never had an issue with it. But he did come up with an interesting observation: he had noticed it surging months ago, so the issues we've had in the last couple of weeks are possibly only the end of a long development phase.
|
Understanding hydra problems
|
Topic: technology, opinion | Link here |
One of the reasons I'm taking such a long time setting up hydra—apart from the fact that it's possible—is that I want to understand the right way to do things. And currently I'm completely baffled by a number of things, notably the different characters generated by the same keyboard on eureka and hydra.
Both keyboards are Sun Type 7 keyboards with a USB interface. In each case I load a key map file that, in particular, remaps the Control key to Alt, the CapsLock key to Control, the keys on the left to F1 to F12, and the keys at the top to F21 to F32.
But what I get is completely different. The Control and Alt keys get mapped correctly. So do most of the top function keys, almost to F21 to F32. But for some reason F3 gets remapped to F21, when it should be F23. And the arrow keys are completely crazy.
With the help of xev, I've established the following keycode mappings:
| key | hydra | generates | eureka | generates | ||||
| down | 116 | 104 | ||||||
| up | 111 | BS | 98 | |||||
| left | 113 | Alt-R | 100 | |||||
| right | 114 | 102 | ||||||
| PgUp | 112 | KP_Divide | 99 | Prior | ||||
| PgDown | 117 | Multi_key | 105 | Next | ||||
| Stop | 136 | 145 | F1 | |||||
But why? I thought that keycodes were directly related to keyboard scan codes, and they're demonstrably the same on both keyboards: I can swap the keyboards and the problem stays with hydra.
Of course, I can just go and build a new key map for hydra, but that way I don't have an answer to the question why?. What I need to know is why the keycodes are different. Is there another layer of translation between the scan codes and key codes? Spent a lot of time chasing that one, ultimately giving up.
|
X server 0 on hydra
|
Topic: technology | Link here |
Another issue with the X setup on hydra is that there seems to be no way to revert to a default key map. xmodmap is the program, and it'll happily modify the current key map any way you want. But to get back to the default, it seems that I need to restart X. And given the issues with Hugin, I don't want to do that yet: first I want to set up server 0 (one screen per monitor).
In preparation, did a couple of other things. First, I'm using an old two-keyed mouse until I get the configuration finished. I've been here before, but the current way to set middle button emulation appears to be:
xinput set-prop 10 287 1 # Logitech wired
I've already found instructions for setting up one screen per monitor, and I created a config file that I thought would do the job:
WARNING: this configuration doesn't work. It's here to compare what I need to change to get it to work correctly.
Screen 0 "Screen0" 0 0
Screen 1 "Screen1" RightOf "Screen0"
Screen 2 "Screen2" RightOf "Screen1"
Screen 3 "Screen3" RightOf "Screen2"
Section "Device"
Identifier "Device2"
Driver "nvidia"
VendorName "NVIDIA Corporation"
BoardName "NVIDIA GeForce RTX 3060"
BusID "PCI:1:0:0"
Screen 2
EndSection
And that looks very much like the example in the manual, modulo the fact that the manual only has two screens, and that it includes stuff that I suspect is not necessary, like the Display subsection in the Screen sections.
OK, try it out. It came up so fast that I thought it was something that had already been there. But all 4 displays showed the same content, and the display resolution was the resolution of the smallest monitor. What went wrong there? Under those circumstances it wasn't surprising that the window manager went mad trying to find displays :0.0 to :0.3.
So that, too, will have to wait for another day.
Getting back to the keyboard issues, it seems that there's a new layer of stuff which recognizes keyboards and tries its own mapping. This could be the missing link between scan codes and keycodes. This page refers to a program xkbcomp, a keymap compiler, unfortunately without giving enough information to understand what's going on:
The source may specify an X display, or an .xkb or .xkm file; unless explicitly specified, the format of destination depends on the format of the source. Compiling a .xkb (keymap source) file generates a .xkm (compiled keymap file) by default. If the source is a .xkm file or an X display, xkbcomp generates a keymap source file by default.
But I can't find any file with a name ending in .xkb or .xkm. I was able to find a reference in some file to Sun Type 7, so maybe this is part of the puzzle.
Then there's this reference, in German, which suggests that you need specific InputDevice options to use the keyboards. In that case, I might have been barking up the wrong tree.
But then, I've been here before, over 10 years ago. At the time it didn't worry me as much:
The keyboard map ended up for my Sun Type 7 keyboard being different from what it was on dereel. In particular, I still have to run xmodmap twice, but the resultant behaviour is different. And on defake I could switch to vtys with Ctrl-Alt-F1 to Ctrl-Alt-F8, but only the ones marked like that above the main keyboard, though they're bound to F21 to F28. Now the keyboard map seems to have noticed that. This is part of a general issue with xmodmap.
I got past those issues at the time, but it just goes to show how incomprehensible this is.
|
Jenny's Lammtopf again
|
Topic: food and drink, opinion | Link here |
Jenny's Lammtopf (I must really find out what the correct name is) for dinner tonight. In the past I have been frustrated by the quantities, and it occurred to me that I'm simply putting in too much water. Up to now the quantities have been one part lamb mix, one part risoni and one part water. What about 2 parts mix, one part risoni and no water?
|
|||||||||||||||||||||||
No, too dry. Add water:
|
|||||||||||||||||||||||
And that was surprising because for my portion I had 240 g mix, 70 g risoni and 44 g water, and for Yvonne's half portion 120 g mix, 35 g risoni, but 40 g water instead of the 22 g I had expected.
The results?
|
|||||||||||||||||||||||
That looked good enough, but my portion was particularly al dente. Yvonne's was OK. I also was not able to finish mine, though strangely Yvonne did finish hers. So it looks like still more playing around to get things right.
|
Dying duck
|
Topic: animals | Link here |
Yvonne tells me that there's a dead duck in the “dam paddock”. It has to go, but she hates touching them. So off looking where she told me where it was. No duck, the search hampered by duck-sized piles of horse dung. Finally Garry Marriott (who has been surprisingly in evidence lately) found it, in a completely different place.
|
|||||||||||||||||||||||
Off to pick it up, but... it resisted. Not yet dead. What are the options? I could kill it, of course, and maybe put it out of its misery. But what if it could recover? I've frequently seen birds that I had thought to be dead recover and fly off again. So I left it there. We'll see tomorrow.
| Wednesday, 15 November 2023 | Dereel | Images for 15 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
X on hydra: progress
|
Topic: technology, opinion | Link here |
Why am I having so much difficulty setting up X on hydra? Surely there must be an easier way. So far, based on Nvidia's instructions for Configuring Multiple X Screens on One Card I have created a configuration file that should do the trick. But it doesn't.
OK, copy the configuration details exactly from that page and try again. That doesn't work either! Off searching for web pages that go beyond the documentation I had already found, and came up with How to set up dual or multiple monitors, also from Nvidia. That looks promising. It starts with:
Open the NVIDIA Control Panel. You can do this from the popup menu when you rightclick on the desktop.
So much bad language at once! What's an NVIDIA Control Panal? Which popup menu? Which desktop? Clearly it's making lots of assumptions, presumably Microsoft. Would it even apply to Apple? “Is this answer helpful?”. No. “Thanks for your feedback”.
Still, there's this “Control Panel” thing. But what I have is nvidia-settings and nvidia-xconfig, the latter particularly useless. Can I find a “Control Panel” somewhere? If I can, I still don't know how. Can it be that “Control Panel” is a Microsoftism for one of those two programs? Certainly nvidia-settings doesn't offer a tab "Set up Multiple Displays" under anything that could match “the Display category”. What about nvidia-xconfig? RTFM?
To my surprise yes, I found a man page. And it includes
--separate-x-screens, --no-separate-x-screens
A GPU that supports multiple simultaneous display devices can
either drive these display devices in a single X screen, or as
separate X screens. When the '--separate-x-screens' option is
specified, each GPU on which an X screen is currently configured
will be updated to have two or more (depending on the capabili-
ties of that GPU) X screens configured. The '--no-sepa-
rate-x-screens' option will remove any extra configured X
screens on each GPU. Please see the NVIDIA README description
of "Separate X Screens on One GPU" for further details.
--x-screens-per-gpu=X-SCREENS-PER-GPU
A GPU that supports multiple simultaneous display devices can
either drive these display devices in a single X screen, or as
separate X screens. When the '--x-screens-per-gpu=<quantity>'
option is specified, each GPU on which an X screen is currently
configured will be updated to have <quantity> X screens. <quan-
tity> has to be greater than 0. Setting <quantity> to 1 is
equivalent to specifying the '--no-separate-x-screens' option.
Please see the NVIDIA README description of "Separate X Screens
on One GPU" for further details.
Amazing! OK, try that. And yes, it produced a configuration file that looked surprisingly like the example from the instructions. Will it work?
Yes! Well, the mouse didn't, for some reason. It didn't start working until I prodded it with eureka's mouse via x2x. But finally I have my four screens on hydra. Well, sort of. The display resolutions were still all over the place. Further investigation showed that the resolutions in the config file bore no resemblance to the resolutions of the monitors. Two of them are the standard 1920x1080, one is 1920x1200, and the last is 1366x768. But all the mode sections showed:
SubSection "Display"
Depth 24
Modes "1600x1200" "1024x768" "800x600" "640x480"
EndSubSection
That's so far from being correct that it took me a while to catch my breath. All resolutions are for monitors with an aspect ratio of 4:3, while most modern monitors are 16:9, and have been for at least 15 years. The results were interesting: it set the DVI outputs to 1600x1200, and the HDMI outputs to 1024x768.
In addition, the window manager showed strange differences. :0.0 and :0.1 were correct, but :0.2 and :0.3 had something that bore no resemblance to my configuration. That proved to be my fault: I had only created fvwm2 configuration files for the first two screens, and fvwm2 winged it (presumably a default configuration) for the other two.
Still, that's progress. What I need to do now:
And then I checked the man page for nvidia-xconfig again:
=== root@hydra (/dev/pts/8) /etc/X11 46 -> man nvidia-xconfig
No manual entry for nvidia-xconfig
Huh? Oh, I found the man page on eureka, and it's over 7 years old:
=== grog@eureka (/dev/pts/27) ~/keyboard 21 -> l /usr/local/man/man1/nvidia-xconfig.1.gz
-r--r--r-- 1 root wheel 7,909 28 Sep 2016 /usr/local/man/man1/nvidia-xconfig.1.gz
Further investigation shows that there's a --help option to nvidia-xconfig, which with some prompting produces a badly formatted version of what at least looks like most of the information. Who needs man pages, anyway?
If one thing is clear from the pain that I've been going through, it's not (all) my fault. I had started to wonder if I had missed something obvious. But if Nvidia's own documentation contradicts itself, and the programs generate invalid configuration files, clearly much of the blame must go elsewhere.
|
X keyboard insights
|
Topic: technology, opinion | Link here |
The other thorn in my side with hydra X setup is the keyboard, of course. This time I started X without loading a key map. I saved the defaults to use for comparisons later on. After loading the keymap and saving it, I was able to confirm that the map was the same as on eureka. And the F3 mapping? My fault. I had mis-mapped it from day 1, over 10 years ago, and never noticed. But there's lots of fun ahead comparing the maps.
|
Dead duck
|
Topic: animals | Link here |
Out again today to look for yesterday's duck. Yes, it's dead, Jim:
|
|||||||||||||||||||||||
Clearly something else found it before I did:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
| Thursday, 16 November 2023 | Dereel | Images for 16 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More keyboard insights
|
Topic: technology, opinion | Link here |
One of the things that I still need to understand is why the identical keyboards on hydra and eureka generate different keycodes under X. Today Callum Gibson suggested a few things not related to X on IRC.
First, you can test your keyboard type from a vty with kbdcontrol:
=== root@hydra (/dev/pts/8) /etc/X11 56 -> kbdcontrol -i </dev/ttyv0
kbd0:
kbdmux0, type:AT 101/102 (2)
Oh. Not exactly the answer I was expecting, nor anything of use. But
=== root@hydra (/dev/pts/8) /etc/X11 56 -> kbdcontrol -d < /dev/ttyv0 | less
# alt
# scan cntrl alt alt cntrl lock
# code base shift cntrl shift alt shift cntrl shift state
# ------------------------------------------------------------------
...
034 'g' 'G' bel bel 'g' 'G' bel bel C
...
096 fkey51 fkey51 fkey51 fkey51 fkey51 fkey51 fkey51 fkey51 O
There are a surprising number of high-numbered function keys. What are they? But that was on hydra, and very few corresponded directly with eureka:
=== grog@eureka (/dev/pts/27) ~ 50 -> kbdcontrol -d < /dev/ttyv0 | less
# alt
# scan cntrl alt alt cntrl lock
# code base shift cntrl shift alt shift cntrl shift state
# ------------------------------------------------------------------
...
034 'g' 'G' bel bel 231 199 135 135 C
...
096 fkey51 '9' '9' '9' 185 185 185 185 N
But the differences are just what happens with the modifier keys. I can't see anything that would explain the different keycodes that I saw on Tuesday. But it does give a little further insight. Time to RTFM. Well, no, no time now. That will come later.
|
Updating window manager config
|
Topic: technology, opinion | Link here |
One of yesterday's action items was to update the window manager (fvwm2) configuration. And that relates to my gradual conversion from ISO 8859-1 to UTF-8. Currently I have two ways to start an xterm, one with each encoding. I seem to have fixed most of the problems with UTF-8, so it's time to remove the ISO-8859-1 versions. And in the process, I need to set my shell environment accordingly.
In more details:
-export LC_ALL=en_AU.ISO8859-1
+# Bite the bullet (20231116)
+export LC_ALL=en_AU.UTF-8
Next, restart the window managers on eureka:0. But they didn't want to go:
=== grog@eureka (/dev/pts/27) ~/keyboard 45 -> killall fvwm2
No matching processes belonging to you were found=== grog@eureka (/dev/pts/27) ~/keyboard 46 -> ps aux|grep fvwm2
grog 25950 0.0 0.0 74388 4284 v0 I 7Nov23 0:04.03 fvwm2 -s -display :0.0 -f /home/grog/.fvwm/fvwm2rc-eureka:0.0 (fvwm)
grog 25951 0.0 0.0 74388 4604 v0 S 7Nov23 0:21.76 fvwm2 -s -display :0.1 -f /home/grog/.fvwm/fvwm2rc-eureka:0.1 (fvwm)
grog 25952 0.0 0.0 74388 4956 v0 I 7Nov23 0:19.16 fvwm2 -s -display :0.2 -f /home/grog/.fvwm/fvwm2rc-eureka:0.2 (fvwm)
grog 25953 0.0 0.0 74388 4412 v0 I 7Nov23 0:04.55 fvwm2 -s -display :0.3 -f /home/grog/.fvwm/fvwm2rc-eureka:0.3 (fvwm)
grog 89379 0.0 0.0 74388 2844 v4 I 4Oct23 0:04.96 fvwm2 -s -display :1.0 -f /home/grog/.fvwm/fvwm2rc-eureka:1.0 (fvwm)
grog 14491 0.0 0.0 18848 2332 27 S+ 3:05pm 0:00.00 grep fvwm2
Huh? Why that? Killing them individually worked, but I found another bug in my .xinitrc: they didn't automatically get restarted.
Finally it was done, though, just in time for the next surprise. The menus look different. Here the old, then the new:
|
|
|||||||||
|
|
|||||||||
Why has that changed? I haven't even mentioned encodings in .fvwm2/mkrcfiles.
On the whole, though, things went relatively smoothly. But I've discovered other issues:
xclock*hands: green
xlogo*Background: magenta
xclock*Geometry: 200x200
e*display: unix:0.1
emacs-Right*Geometry: 110x50
emacs*Geometry: 110x75
emacs-Left*Geometry: 110x50
e*Geometry: 110x50-0+0
eg*Geometry: 110x50
emacs*font: 9x15
e*font: 9x15
xmailbox*mailSndFile: /usr/local/bin/sound /home/grog/Sounds/newmail.au
I must have put that all in myself, but I can't recall any of it. The files in ~/Sounds date to March 1997. And now I'm replicating this stuff, one file per system, so there's an .Xdefaults-hydra that contains the same reference to a non-existent directory. I need to find a way to generate the file, like I do with the window manager files.
=== grog@hydra (/dev/pts/7) ~ 9 -> ssh freefall
agent key RSA SHA256:S7sZHLcY4dgw53/rF70vrScdPuGef3enHdJzuYA1WDo returned incorrect signature type
grog@freefall.freebsd.org: Permission denied (publickey).
Disconnected at Fri 17 Nov 2023 11:55:54 AEDT
so that's another one I'm going to have to chase up.
So how does my to-do list look now?
|
Sweet and sour fish revisited
|
Topic: food and drink, opinion | Link here |
Sweet and sour fish for dinner again today. The recipe contained some interesting historical ingredients: bamboo shoots and water chestnuts. It's been years since I've seen them in Chinese recipes.
Set off half-cocked and didn't read the web page, which states:
Deep fry the fish after all, use different vegetables, and reduce the quantity of sauce.
Today I was more concerned with the vegetables, and in fact came to a conclusion close to what I did last time, only 3 months ago. And I came to the same conclusions as then, too much sauce:
|
|||||||||||||||||||||||
But I forgot the pineapple that I put in last time, and I didn't note the detail about frying the fish. I really need to read what I've written.
| Friday, 17 November 2023 | Dereel → Ballarat → Dereel | Images for 17 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Next health check
|
Topic: health | Link here |
Into Ballarat today to talk with Dr Paul Smith about my blood test results. Nothing unexpected, except the MCV results. We had had a bet on the result: Paul 102 fl, I 104 fl. And what was the result? 103 fl! We hadn't considered that possibility, the only one that could result in a draw. So this time Paul has bet 105 fl, and I'll stick with 104 fl.
The other thing of interest is that I'm now over 75, and so I'm due for a health assessment. Somehow this reminds me of the diabetes assessment 10 years ago, which didn't seem very useful. Still, it does no harm, and this time Paul will be involved too, so I'm booked in for 11 December.
|
Buying tomatoes
|
Topic: food and drink, language, opinion | Link here |
While in town, Yvonne had asked me to buy some “mini Roma” tomatoes. OK, can do. Or can I?
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Sweet delights! Petite [sic] tomatoes! Intensely Sweet Cherry Vine Tomatoes! Even some with no name at all:
|
|
|||||||||
But no Romas.
Fool! You're looking in the wrong place:
|
|
|||||||||
The new name for “Mini Roma” is “Petite tomatoes”. Now isn't that obvious?
| Saturday, 18 November 2023 | Dereel | Images for 18 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More garden stuff
|
Topic: gardening | Link here |
Jesse Walsh along today for more garden work, including weeding the bed south of the house, planting the Haemanthus bulbs and the Strelitzia reginae, which proved to consist of no fewer than 5 plants:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
House photo proceessing, another week
|
Topic: technology, photography, opinion | Link here |
Saturday is house photo day, and for the past couple of weeks I've been trying to migrate the processing to hydra, which is much faster than eureka.
Much of the photo processing is either on distress, my Microsoft machine, or done by scripts. Many of the scripts are single-threaded, so that seemed to be a good place to start. mkpto builds a project file (name ending in .pto), and it takes a certain time to process. Do all the .pto files (one per photo) to speed things up? That's almost as simple as changing a ; to a &.
Then process on eureka:1 to select the view and maybe set some masks (not necessary today). The first two were OK, but then something went wrong, and all the others showed the view from the house entrance:
|
|||||||||||||||||||||||
Yes, this is a different image. I forget why, but it turned out not to be important.
What's wrong with Hugin this time? At least it crashed. OK, try it on hydra after all. And yes, I had had a Hugin instance running on hydra:1 since Monday, but the configuration file was wrong. OK, put in the correct configuration file and restart.
It crashed! OK, at least that's a clue about what's wrong. Back to the previous, functional config file. It crashed! About the only insight was that the Hugin splash screen appeared before it crashed. But whatever it is, it's strange.
Start Hugin on hydra:0 (individual screens, not what I want for Hugin). It worked as expected, but I still got the wrong images. Off to look at the .pto files. Yes, with one exception they all contained the same files.
Ah! Race condition in setting up the conversion commands? Off to look at the code, but so far I haven't found anything obvious. But it took enough time that I didn't get finished.
|
Gigot d'agneau again
|
Topic: food and drink, opinion | Link here |
Leg of lamb for dinner tonight, not something that we eat very often. It's only a coincidence that the last time we ate one was almost exactly a year ago: we're coming up to the 60th anniversary of the “Shepherd's pee” incident, so it's time to have the ingredients for a shepherd's pie.
How long do I cook it? It was 2 kg and boned. According to my cooking times page, I need to cook to 55° meat temperature, which should take somewhat less than 55 minutes per kilogram. That would make about 100 minutes for this one.
But it's long and thin. Does that make a difference? Better be safe than sorry. But yes, it only took 65 minutes. How do you account for that in a guidelines page?
And how did it taste? Boring! Are we maybe getting past the idea of roast meat without any herbs or spices?
| Sunday, 19 November 2023 | Dereel | Images for 19 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Another catchup day
|
Topic: general, technology, photography, opinion | Link here |
Another day spent doing lots of things with nothing to show for it. Found the race condition in mkpto1 (temporary files with fixed file names), finished yesterday's panoramas, and somehow that seems to have been all.
| Monday, 20 November 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Understanding hydra ssh issues
|
Topic: technology, opinion | Link here |
So why can I connect from dereel to freefall with no issues, but I can't from hydra with the same credentials? Spent an inordinate amount of time today chasing things up. Firstly, to my surprise, hydra (last update 18 October) is running OpenSSH_9.2p1, OpenSSL 1.1.1t-freebsd 7 Feb 2023, while dereel (last update 20 September)) is running OpenSSH_9.4p1, OpenSSL 3.0.10 1 Aug 2023. But then dereel is running FreeBSD 15-CURRENT, while hydra is running 13.2 STABLE. It shouldn't make a difference, but is it worth trying running dereel's version on hydra?
Well, it might be worthwhile, but nothing useful happens:
=== grog@hydra (/dev/pts/7) ~ 67 -> /dereel/usr/bin/ssh freefall
ld-elf.so.1: Shared object "libcrypto.so.30" not found, required by "ssh"
And the other way doesn't do anything useful either:
=== grog@dereel (/dev/pts/1) ~ 34 -> /hydra/usr/bin/ssh freefall
OpenSSL version mismatch. Built against 300000a0, you have 1010114f
OK, give up on that. Do I have a configuration mismatch? What files does ssh look at, anyway? Ran ktrace against each and came up with /etc/ssl/openssl.cnf, /etc/ssh/ssh_config and ~grog/.ssh/config. And all were the same, with the exception of slightly different comments in /etc/ssh/ssh_config.
OK, what do they say to each other? Ran ssh -v and later ssh -v -v, which proved interesting. Running diff against the hydra version (prefixed -) and the dereel version (prefixed +) gave me
-debug1: kex: host key algorithm: ssh-ed25519
+debug1: kex: host key algorithm: ecdsa-sha2-nistp256
Now why did it choose a different algorithm? The configurations are all the same. Could it be a different default? This alone might potentially be enough to explain the issue. But continuing:
debug1: SSH2_MSG_KEX_ECDH_REPLY received
-debug1: Server host key: ssh-ed25519 SHA256:oJ7FKX5UTBWP4CncsrsaIb1JbfbtqzKOMYni3oVLAo0
-debug2: ldns: got 8 answers from DNS
-debug2: ldns: trying to validate RRset
-debug2: ldns: got 1 signature(s) (RRTYPE 46) from DNS
-debug2: ldns: RRset validation failed: General LDNS error
-debug1: found 8 insecure fingerprints in DNS
-debug1: Fssh_verify_host_key_dns: matched SSHFP type 4 fptype 2
-debug1: Fssh_verify_host_key_dns: matched SSHFP type 4 fptype 1
-debug1: matching host key fingerprint found in DNS
+debug1: Server host key: ecdsa-sha2-nistp256 SHA256:qbhR/gKDUzkxEvdNtsTlR7uM6mbj8UQ2gMQhobXrQg8
What's all that? ldns? Insecure fingerprints? Spent a lot of time trying to find information, without success.
debug1: rekey in after 134217728 blocks
-debug2: get_agent_identities: ssh_agent_bind_hostkey: agent refused operation
+debug1: get_agent_identities: bound agent to hostkey
debug1: get_agent_identities: agent returned 1 keys
This one is particularly interesting because only the debug2 output clarified any kind of refusal. The first time round (-v, debug1) just showed dereel getting “bound agent to hostkey”, without clarifying whether it was important or not. So the following was to be expected:
debug1: Server accepts key: /home/grog/.ssh/id_rsa RSA SHA256:S7sZHLcY4dgw53/rF70vrScdPuGef3enHdJzuYA1WDo agent
-agent key RSA SHA256:S7sZHLcY4dgw53/rF70vrScdPuGef3enHdJzuYA1WDo returned incorrect signature type
...
-grog@freefall.freebsd.org: Permission denied (publickey).
+Authenticated to freefall.freebsd.org ([96.47.72.132]:22) using "publickey".
So: I've done a lot of investigation, spent a lot of time, and I still don't have any idea why this problem still exists. Why is this so complicated?
Later, in the lounge room, I tried again. Yes, tiwi can't connect either. Fire up a hydra xterm, try to connect. It works!
Huh? Now it works, now it doesn't? What's going on? Does it have something to do with the X server on eureka? To be investigated. Mañana.
| Tuesday, 21 November 2023 | Dereel | Images for 21 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More monitor dropouts
|
Topic: technology, opinion | Link here |
More dropouts from my “Matrix” (brand) monitor this morning, and they lasted longer, up to about a minute each. Clearly it's trying to tell me something, and it's time that I listened. Time for a new monitor. Daniel O'Connor pointed out that I can get a new monitor with similar specs for under $200. So now the question: do I want a monitor with similar specs? 27" seems to be about the biggest I can fit on my desktop. Should I stick with the current 2560×1440 resolution or go to 3840×2160? My eyes aren't what they used to be, and 3840×2160 might cause problems. But then I can always run it at a lower resolution—can't I?
|
hydra installation: progress
|
Topic: technology, opinion | Link here |
So are my authentication problems with freefall related to eureka? Yesterday's investigations suggested “yes”. This morning, first thing, tried to access freefall directly from hydra. Success!
Well, I've worked around the issue. I still have not the slightest idea why, or what the problem was, but somehow it seems that something in X has changed since my last build on eureka in 2015. But in the 25 years before that I have never seen this kind of incompatibility. And then there's the issue with Emacs, where Emacs windows on eureka get control codes changed (c-f becomes c-s-f, making it useless).
OK, those are open questions, but they don't stop me from continuing with the migration. What next? Finally X server :0? How hard can it be? Last week I created an X configuration with one screen per monitor. Yes, the sequence was wrong, and this horrible broken nvidia-xconfig generated resolutions that were just plain impossible. But I can fix that—can't I?
The layout issues were easy to address: move the monitors. This proves to be convenient, so I can use the right-hand two monitors for hydra, and the small monitor moves further to the left.
OK, how do I fix the resolutions? Can nvidia-settings help? Yes, if reluctantly. Once again it didn't look at the EDID, but it did allow me to select the correct values. Then save the configuration:
|
|
|||||||||
OK, save that, merging with what I already had. The result: 7 screens! nvidia-xconfig and nvidia-settings can't agree on the screen names, so I ended up with both. OK, save again, restart X. Success! Now I really have something that looks like what I want, at least for server 0.
So now my to-do list is:
| Wednesday, 22 November 2023 | Dereel | Images for 22 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More hydra progress
|
Topic: technology, opinion | Link here |
More work on hydra today. First, how do I use my new-found screen real estate? Can I put a browser on hydra:0.3 (using the monitor for eureka:0.0, which thus became inaccessible)? Yes! Can I run mail on hydra and send mail too? That had been an issue with dereel that I had put into the “too hard” basket. But here, too, it worked with no problems. I just needed to start an Emacs on hydra:0.3 to avoid the control character mutilation that I had experienced (c-f changed to c-s-f, for example). And that also worked! The only issue was missing spell checker support. Somehow I had ended up with only a partial hunspell, which seems to be for Hungarian. What do I install? ispell? aspell? Both? Tried aspell, and that worked.
Moving on, can I display web pages from mail (on a window on eureka) on the web browser on hydra? That, too, works. Can I get proper sound output from hydra? In the past I have had issues there. But yes, that also works. For once most things seem to work.
OK, back to yesterday's to-de items. Two items stood out:
The keyboard mapping isn't that important, since I'm connecting with x2x anyway. And Hugin?
=== grog@hydra (/dev/pts/2) ~/Photos/20231122 48 -> hugin
bash: hugin: command not found
Huh? More investigation showed that yes, Hugin was no longer installed. So were a number of others. What's up?
=== grog@hydra (/dev/pts/26) ~ 18 -> ls -dl /usr/local/bin/
drwxr-xr-x 2 root wheel 27,136 22 Nov 10:45 /usr/local/bin/=== grog@hydra (/dev/pts/26) ~ 20 -> ls -cl /usr/local/bin/aspell
-rwxr-xr-x 1 root wheel 146,352 22 Nov 10:45 /usr/local/bin/aspell
A smoking gun! Installing aspell had removed Hugin! And for once I hadn't saved the output of pkg install.
OK, I had built Hugin locally. Install and start:
=== root@hydra (/dev/pts/8) /usr/ports/graphics/hugin 68 -> hugin
ld-elf.so.1: Shared object "libwx_baseu-3.0.so.0" not found, required by "hugin"
Dammit, what else had it removed? Started building the port, and about 20 minutes into a maze of twisty little dependencies, all enormous, decided to go from the package. And that at least started. But clearly it's a bad idea to set ASSUME_ALWAYS_YES = true in /usr/local/etc/pkg.conf.
But it wasn't all plain sailing. Watching a video with firefox, I switched to the full-screen display. And... the window manager crashed! That's so unusual that I didn't understand what had happened.
And I couldn't restart it!
=== grog@hydra (/dev/pts/26) ~ 10 -> fvwm2 -s -display :0.3 -f ~/.fvwm/fvwm2rc-hydra\:0.3
[fvwm.3][CMD_EdgeResistance]: <<DEPRECATED>> The command EdgeResistance with three arguments is obsolete. Please use the following commands instead:
EdgeResistance 250
Style * EdgeMoveDelay 250
Style * EdgeMoveResistance 10
[fvwm.3][pixmapPath_function]: <<ERROR>> PixmapPath is deprecated since 2.3.0; use ImagePath instead.
[fvwm.3][iconPath_function]: <<ERROR>> IconPath is deprecated since 2.3.0; use ImagePath instead.
[fvwm.3][menustyle_parse_old_style]: <<DEPRECATED>> The old MenuStyle snytax has been deprecated. Use 'MenuStyle * "mwm", Foreground "black", Background "gold", Greyed "brown", Font "-*-*-medium-r-*-sans-12-*-*-*-p-*-*-*", "AnimationOff"' instead of 'MenuStyle black gold brown -*-*-medium-r-*-sans-12-*-*-*-p-*-*-* mwm'
[fvwm][FlocaleGetFontSet]: (-*-*-medium-r-*-sans-12-*-*-*-p-*-*-*) Missing font charsets:
ISO8859-5, KOI8-R, ISO8859-7, JISX0208.1983-0, KSC5601.1987-0, GB2312.1980-0, JISX0201.1976-0
Assertion failed: (ret != inval_id), function _XAllocID, file xcb_io.c, line 626.
Abort trap (core dumped)
How much of that is the normal grumbling that this kind of program tends to do behind the scenes, and how much is a real error message? But I couldn't find a way round the problem. I had to restart X.
OK, if firefox is even more unreliable, how about Google Chrome? Tried to start from the window manager. Nothing. Try from an xterm:
=== grog@hydra (/dev/pts/21) ~ 1 -> chrome
[40931:-1238556672:1122/123008.451596:ERROR:process_singleton_posix.cc(353)] The profile appears to be in use by another Chromium process (60349) on another computer (dereel). Chromium has locked the profile so that it doesn't get corrupted. If you are sure that no other processes are using this profile, you can unlock the profile and relaunch Chromium.
[40931:-1238556672:1122/123008.451624:ERROR:message_box_dialog.cc(146)] Unable to show a dialog outside the UI thread message loop: Chromium - The profile appears to be in use by another Chromium process (60349) on another computer (dereel). Chromium has locked the profile so that it doesn't get corrupted. If you are sure that no other processes are using this profile, you can unlock the profile and relaunch Chromium.
I seem to have seen this before. Ah, right, only 2 weeks ago. And in the config directory I found, once again:
lrwxr-xr-x 1 grog lemis 20 31 Jul 2021 SingletonCookie -> 13443178945937354382
lrwxr-xr-x 1 grog lemis 12 31 Jul 2021 SingletonLock -> dereel-60349
lrwxr-xr-x 1 grog lemis 50 31 Jul 2021 SingletonSocket -> /tmp/.org.chromium.Chromium.dueGS8/SingletonSocket
So another instance of chrome on a different system, over 2 years ago, is still
raising its ugly head. And the current version of chrome (Version 119.0.6045.159
(Official Build) (64-bit)) is too polite to even mention it. How do people with desktop
managers work around a problem like this? For me, of course, it's simple: remove those 3
symlinks, and all was well worked.
For a while. Then I discovered that I couldn't search. xev told me that I had the dreaded stuck modifier key (mod4, 0x40, which proved to be Super or Hyper). I have already addressed that in my keyboard mapping, but that's not in place yet. But xmodmap confirms:
=== grog@hydra (/dev/pts/26) ~ 12 -> DISPLAY=:0 xmodmap -pm
xmodmap: up to 4 keys per modifier, (keycodes in parentheses):
shift Shift_L (0x32), Shift_R (0x3e)
lock Caps_Lock (0x42)
control Control_L (0x25), Control_R (0x6d)
mod1 Alt_L (0x40), Alt_R (0x71), Meta_L (0x9c)
mod2 Num_Lock (0x4d)
mod3
mod4 Super_L (0x73), Super_R (0x74), Super_L (0x7f), Hyper_L (0x80)
mod5 Mode_switch (0x8), ISO_Level3_Shift (0x7c)
But there's a way to disable them, the way I do in the keymap. But it doesn't work if the modifier is stuck:
=== grog@hydra (/dev/pts/26) ~ 14 -> DISPLAY=:0 xmodmap -e 'clear mod4'
xmodmap: please release the following keys within 2 seconds:
Super_L (keysym 0xffeb, keycode 115)
...
xmodmap: please release the following keys within 32 seconds:
Super_L (keysym 0xffeb, keycode 115)
xmodmap: unable to set modifier mapping, keyboard problem
It's interesting that it told me which of the 4 keys was stuck. But it didn't help much: once again I had to restart X.
Still, I'm making progress. Isn't it easy to tell?
|
A new scratching post for Bruno
|
Topic: animals, general, opinion | Link here |
Bruno is gradually accepting life only indoors, but for us it's something new. One thing we're discovering is that his claws are getting too long, and they're catching in things.
That's what a scratching post is for, of course, and of course we have one. But he doesn't use it. While showing it to him, I discovered why: it's too light, and when he scratches on it, he just pulls it with him.
OK, we need a bigger post. Today Yvonne brought one:
|
|||||||||||||||||||||||
It's certainly big enough. We took some video of him taking his first looks, but he wasn't very impressed. Later, though, he took to it:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The middle photo shows clearly how his claws get caught in things.
|
A new monitor?
|
Topic: technology, opinion | Link here |
So it's clear that I should be looking for a new monitor to replace they dying “Matrix” monitor. Yes, 27" is really the widest I can do (about 64 cm). But shouldn't I really be looking for a 3840×2160 resolution?
As it happens, Friday is Thanksgiving in the USA, and the following day is called Black Friday because of the time skew. And the idea seems to have taken hold with retailers round the world. So I got mail from Umart offering big discounts on monitors. Even one that looks useful, the LG 27” 27UP850-W 4K UHD UltraFine™ IPS Monitor.
In fact, they were too polite to put the “27UP850-W” in the heading, but it's necessary.
Only $449! A quick check on eBay shows that the cheapest price there is round $556, so it's a real discount. But still more than twice the price of the one that Daniel O'Connor found yesterday. What do I get for it? The higher resolution, of course, along with confirmation that it will run at 2560×1440 if I want. But also calibration for photos, something that I could clearly use.
And there starts the fun. What alternatives are there? What results do I get from a Google search for “Best 27" photo monitor”? Nothing useful. The results are so one-sided and vague that I can't be bothered looking through them. On the other hand, StaticIce gives me prices ranging from $129 to $7,999. Once again I find the choice more work than it seems to be worth. About the only thing I did find is that the LG monitor is probably not the best choice. But why is this all so difficult?
| Thursday, 23 November 2023 | Dereel | Images for 23 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Garden flowers in late spring
|
Topic: gardening, opinion | Link here |
It's a month to the solstice, time for the monthly garden photos. And once again I look at these photos and wonder why I bother. They look worse every year. Today I discovered that our Helichrysum, the one that flowered so profusely in previous years, is no longer there. It used to look like the first photo (August 2021), but all I have left is the second photo:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Presumably it fell foul of an overzealous weeder. And I didn't even notice!
And once again there are lots of dying plants. Here the Hibiscus rosa-sinensis “Uncle Max”, of which I was so proud, then the indestructible Alyogyne huegelii, and of course the Camellia japonica that still hasn't forgiven us for moving here nearly 9 years ago:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And our remaining Clematis managed to flower last month, but it seems that it was more like a swan song:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And on one occasion I forgot to water the oak tree sapling in the pot on the verandah. It was not amused, but on closer examination it seems that it's recovering:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Similarly, the Trachelospermum jasminoides (I think that's what it is), the one that should have covered the garden arch years ago, is still barely alive, though a recently arrived Pelargonium is showing what can be done:
|
|||||||||||||||||||||||
Is there anything good to report? Well, the Curry tree (Murraya koenigii), now over 13 years old, has recovered from once again being put out on the verandah too early, and it looks like it's about to flower:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And the roses look surprisingly good:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And to the east of the house the Corymbia ficifolia is looking better than it has typically done at this time of year:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And the Hebes are flowering more profusely, presumably because of the overhauled irrigation:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Now that we're getting a little more attention to the garden, maybe things will pick up.
|
Waiting for a storm
|
Topic: gardening | Link here |
We're running out of eucalyptus chips for mulching the garden. Called up DELA on 5336 4442 and spoke to Sharon, who told me the startling news that they didn't have any, and they didn't know when they would get any. Currently the tree trimming people were somewhere south of Rokewood, maybe not so far from us, but they weren't prepared to take the stuff to Ballarat. Now if there would be a storm, blowing down a lot of trees, things could change...
|
hydra: what next?
|
Topic: technology, opinion | Link here |
On with hydra today. For the most part it's almost usable. Only minor things get on my nerves, in particular the realization that a 180° view angle is not good for my neck. hydra:0.0 and hydra:0.1 are almost useless as a result. I think that 4 monitors is probably as much as I can comfortably use.
And then there's another strangeness with x2x. It works surprisingly well, but now I have one joining eureka:0.0 west to hydra:0.3. In fact, they're both on the same monitor, so I can't see the cursor move from eureka:0.1 to hydra:0.3; there's just no cursor while traversing where eureka:0.0 should be.
But the cursor doesn't always enter at hydra:0.3. Sometimes it seems to bypass it and go to hydra:0.2, and I have to move it back. And when I do, it frequently stops at the left of the screen. To get it to move on the screen, I need to move the mouse to the left.
Why? “It's a bug”, I'm sure. Clearly it has nothing to do with the missing monitor. And then another raised its ugly head: once again I ended up not being able to do anything because the mod4 key had stuck again. But hadn't I disabled it? Yes, but see if the keyboard routines care. Once again I needed to shoot down and restart X.
So it seems that I have had at least three strangenesses or incompatibilities between X on eureka and hydra, along with the c-f bug with Emacs. Hopefully all will go away when I finally migrate.
So: next step is the keyboard. How do I do that? Started using xev, but that's quite painful. Is there maybe a remap utility available? The real issue is accessing the keysym directly. And only today did I notice that the keymap format produced by xmodmap is different, at least for some of the more unusual keys. Here a couple of keys from the keymap I've been using on eureka for over ten years:
keycode 38 = a A a A
keycode 67 = F21 XF86Switch_VT_1 F21 XF86Switch_VT_1
And here's what xmodmap produces now:
keycode 38 = a A a A
keycode 67 = F1 F1 F1 F1 F1 F1 XF86Switch_VT_1
The difference between F1 and F21 is part of my remapping, and the XF86Switch_VT_1 is something that I hadn't noticed before, but which explains why I can switch to /dev/vty0 with S-F21. But clearly more modifiers are involved, and so far I don't know what they are. The alternation a and A suggests that the second or fourth entry relates to the Shift key, but that might mean that I would need some other combination on the new map.
But there's something more pressing: a new monitor. Tomorrow is Black Friday, and gradually more special offers are appearing (1920×1080 monitor for $49, 27" 2560×1440 for $149). That's not what I'm looking for, but there's a good chance I'll get something at a good price. Now I just need to do hours of research to find what I want. The one page that I did followed had recommended a monitor for something like $1,500, which just seems too much to me.
| Friday, 24 November 2023 | Dereel | Images for 24 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Much domestic activity
|
Topic: general, animals | Link here |
We don't often have much in the way of physical contact with the outside world, but somehow today was different. Apart from Petra Gietz' weekly visit, today we had two visitors, one early, one late, so they both arrived just as we were having breakfast: Dave Pinder, the farrier, and Carola Sahler, a friend who wanted to buy some riding boots from Yvonne, and who was gone again in a matter of minutes.
|
Matrix monitor insights
|
Topic: technology | Link here |
My old “Matrix” monitor faded again today, first briefly, then long enough to be annoying. OK, where do I go from here? Check the power supply? How about turning it off at the switch first? As soon as I touched the switch, it came back again. Is this a flaky switch?
It's also interesting to note the strange chirping noises that it makes when it comes back. I'm not sure whether this is new, or whether I just hadn't noticed before. It could be trying to sync the input, for example.
|
New monitor: done
|
Topic: technology, opinion | Link here |
Spent several hours today trying to decide on a new monitor. How do I make a choice? Pages that I looked at included:
So what do I do? It seems that the LG 27” 27UP850-W 4K UHD is not such a bad choice after all. Somehow, though, it's really frustrating not to be able to find the information that I need to make a choice.
So: the LG it is. What does it cost? The cheapest price is $449, from 8 different suppliers. And shipping? Ah, another thing that you can't find out until you almost commit to buy. But one of the suppliers, Scorptec, lets the cat out of the bag: “Free delivery”.
OK, Scorptec, thanks for making that clear. It also makes it clear that it's the cheapest, so I decided to buy it. The usual pain signing up, and then to checkout:
|
|
|||||||||
Huh? $4.49 surcharge for PayPal? I've never seen that before. OK, how about credit card? Same again! Can I do a bank transfer? No. The other options looked even more dubious. Is Scorptec on the level?
About the only other payment option that I could use was Afterpay. I've been there before, and I hate the idea. But how much surcharge do they need for that? None!
That's really puzzling. I can pay now and pay $4.49 extra for the privilege, or I can spread the payment over 6 weeks and not pay any surcharge. What kind of sense does that make?
On IRC, asked about the credentials of Scorptec. It seems that Alastair Boyanich does a lot of business with them, and he's quite satisfied. That's good to hear, but what a way to spend an afternoon!
|
Calibrating the new monitor
|
Topic: technology, opinion | Link here |
I was right that the new LG 27” 27UP850-W 4K UHD has a calibration function, right? Checked the manual (with the descriptive file name ENG.pdf) and found nothing. More searching on the web brought me to this page, describing the calibration software. So yes, all seems well
| Saturday, 25 November 2023 | Dereel | Images for 25 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More garden work
|
Topic: gardening | Link here |
Jesse Walsh along today to continue weeding. It's a slow business, and I'm glad I don't have to do it myself any more.
|
Honesty and Black Friday
|
Topic: general, opinion | Link here |
I've seen a couple of newspaper articles in the last couple of days suggesting that the claimed discounts on Black Friday (today in the USA) are misleading or downright lies. But then the Washington Post came up with incontrovertible evidence, two of a number of emails sent to by backup email address:
16 N 21-11-2023 To washingpost ( 697) The Washington Post N Unlimited access is A99c for a limited time
25 N 24-11-2023 To washingpost ( 694) The Washington Post N Nothing says “Black Friday” like our A99c sale
Now isn't that a saving? Of course, the price is good; that 99¢ is for a month's subscription. But do they really think that their readers are that gullible?
|
More strange spam
|
Topic: general, opinion | Link here |
Another message that I saw today was interesting because it wasn't clear who should believe it:
54 N + 24-11-2023 To groggy@freebsd ( 469) Sainte Apparel N + Last Chance before you lose access to Telstra services !
“Saints Apparel”. No idea what that is, but I'm sure it doesn't have anything to do with Telstra. And that sent to my FreeBSD address.
|
More Nvidia setup
|
Topic: technology, opinion | Link here |
In preparation for today's house photos I tuned the X setup for hydra:1, which I hadn't looked at for nearly two weeks. Based on my experience since then, fired up the server and ran nvidia-settings against it. Yes, I was able to move the displays around, set the resolutions and save the configuration file. But what a difference! The configuration file expanded from 69 to 178 lines, including detailed definitions for each monitor. It look as if nvidia-xconfig is old and worn-out, just barely able to create an X configuration at all, and that to get anything useful, you need to pass it through nvidia-settings.
|
Hugin on hydra
|
Topic: photography, technology, opinion | Link here |
House photo day again today, and for once the weather was appropriate. So, process the panoramas on hydra?
Yes, it almost worked. Once again I had white blanked-out areas on hydra:1, but I also got the expected windows, here on the two monitors to the right. But the other two monitors were blanked out (completely “white”). And when I selected “Crop”, the first image of the sequence appeared outside any window on the left-most monitor. Here a photo after that point, and just as the background on the small monitor was changing back to the xearth display:
|
|||||||||||||||||||||||
What causes that? It could be an issue with Hugin, X or the Nvidia driver. I'll observe when the final monitors are in place, but that's hardly likely to change anything.
What does look like a Hugin bug is various issues with stitching. Here
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
And here a closeup of the area (run the cursor over an image to compare it with its neighbour):
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That was done with an identical project file, first with the old (2018) version of Hugin (and almost certainly enblend), then with the latest (2023) version. And there's no obvious difference between the hugin configuration files on eureka (old) and hydra (new). I wonder what to do there.
|
More on monitor calibration
|
Topic: photography, opinion | Link here |
Now that my calibratable LG 27” 27UP850-W 4K UHD monitor is on its way, it's probably a good idea to learn more about monitor calibration. And just by chance I found not one, but two articles in Digital Photography Review: an introduction, which looks interesting. There's also an introduction that probably isn't necessary. But one thing looks likely: to do it right, I need a colorimeter.
| Sunday, 26 November 2023 | Dereel | Images for 26 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
The effort in a cooked breakfast
|
Topic: food and drink, opinion | Link here |
Decided to eat nasi lemak for breakfast today. A series of coincidences made it take me nearly 2 hours and just about fill a dish washer:
The result:
|
|||||||||||||||||||||||
That's extreme, of course, but it's worth bearing in mind.
|
hydra: Next step
|
Topic: technology, opinion | Link here |
What do I do now with hydra? The good news is that I can use it, and I'm doing so. In principle I could completely move away from eureka, but I still need the correct monitor setup and keyboard mappings. It doesn't seem worth messing around with the monitors until the new monitor arrives, but I could do something about the keyboard.
Or could I? After multiple RTFM I still don't understand the expressions for describing key bindings with xmodmap, like this one:
keycode 67 = F1 F1 F1 F1 F1 F1 XF86Switch_VT_1
The man page does divulge that there are a maximum of 7 bindings for a keycode. By inspection and comparison with other keycodes, the first is the normal binding and the second is the binding with Shift:
keycode 38 = a A a A
But what are the others? There appears to be no documentation.
|
ssh authentication issue, ancient insights
|
Topic: technology, history | Link here |
A year ago today I discovered that I couldn't ssh from tiwi.lemis.com to freefall.freebsd.org—but only when the X server was started from eureka.lemis.com. That confirms the suspicion that I had last week.
|
Yet another Matrix monitor issue
|
Topic: technology | Link here |
Another short dropout with my “Matrix” monitor today. Tried to touch the switch, only half managed. But it came back round that time.
|
Artificial stupidity
|
Topic: technology, opinion | Link here |
For reasons I don't completely understand, but which I suspect is related to Indian school syllabus, there are lots of questions on Quora like If the time is 12 noon at 0° longitude, what will be the time at 75°E longitude?, and also lots of answers from people who have never heard of time zones. But Quora can do better: a ChatGPT bot that not only ignores time zones, but also can't count:
At 12:00 PM at 0° longitude, the time at 75°E longitude would be 4:30 PM. This is because every 15 degrees of longitude represents a time difference of one hour, and 75 degrees is equivalent to five hours.
My guess is that this is a combination of various other incorrect answers, a recognition that the time zone in India (which straddles 75° E) is UTC+5:30, and an answer written in a time or location where the time zone at 0° was UTC+1.
O Brave New World, that has such bots in't.
| Monday, 27 November 2023 | Dereel | Images for 27 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Ancient history in configuration files
|
Topic: technology, history, opinion | Link here |
I have a list of TV programmes to search for, and once a month I send myself a list to remind me, using cron, of course. But where do I store the list? Off looking through crontabs to find it. I failed: it must be in a script with a non-obvious name.
But while looking, I found this entry:
0 0 * * * root cp -p /src/FreeBSD/5.0-CURRENT/src/usr.bin/calendar/calendars/* /usr/share/calendar >/dev/null 2>/dev/null
That made perfect sense—once. It copied the latest version of the calendar data files from the FreeBSD-CURRENT tree to my local system. But look at the path name: FreeBSD-5.0-CURRENT must have stopped well over 20 years ago. And there's no point in updating the crontab entry: the project has decided that our version of calendar, which was in fact the definitive version also used in Linux and MacOS, doesn't belong in the tree any more.
|
New wildflower?
|
Topic: gardening, opinion | Link here |
What's this?
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
After some comparison, I've come to the conclusion that it's Cytisus scoparius, the same bush that normally blooms profusely a little earlier in the year:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
It's interesting how the appearance of the flowers changes.
|
A Grevillea for Yvonne
|
Topic: gardening, opinion | Link here |
While walking the dogs today, Yvonne found a Grevillea rosmarinifolia seedling:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
There are plenty of them round there, but so far there has been no successful attempt to plant them on our property.
|
More hydra fun
|
Topic: technology, opinion | Link here |
So it's clear that my next step with hydra is to remap the keyboard. That would mean that I could use it normally, so it also makes sense to move the small monitor to the extreme left and use the other three normally.
I've done this before. nvidia-settings to the rescue. But today it was obviously in a bad mood. Yes, I was able to move it, but nothing I could do would get it to renumber the screens, so I now have, from left to right, the sequence Screen 1, Screen 0, Screen 2 and Screen 3.
What do I do? Nothing. The new monitor is on its way, and then things will change again.
So, back to the keyboard. It's relatively simple: xmodmap will do what I want. Yet another RTFM, and this time I came across:
keycode NUMBER = KEYSYMNAME ...
The list of keysyms is assigned to the indicated keycode (which
may be specified in decimal, hex or octal and can be determined
by running the xev program). Up to eight keysyms may be
attached to a key, however the last four are not used in any
major X server implementation. The first keysym is used when
no modifier key is pressed in conjunction with this key, the
second with Shift, the third when the Mode_switch key is used
with this key and the fourth when both the Mode_switch and
Shift keys are used.
OK, what's Mode_switch? Tried remapping an unused key (KP_Insert, the 0 on the keypad) to "a b c d e f g h" on eureka. And yes, I could produce a normally, b with Shift—and that was all. More investigations needed. I have a Mode_switch on keycode 93, but how do I find what that is?
Putting that onto the tuit queue, back to hydra. Get rid of that horrible CapsLock key and replace it with Ctrl. xmodmap happily complied. The result? Nothing. It was still CapsLock. OK, update the keyboard map and read it in again. Yes, now it works. A bit of random messing around with the keyboard, and somehow I ended up with CapsLock set—and no way to reset it. I had cut off the branch I was sitting on. Nothing to do but to restart X.
And that's all that time allowed. Why is this all so complicated?
|
Irrigation fun
|
Topic: gardening, opinion | Link here |
I was woken up at 4:00 this morning by the sound of the irrigation sprinklers outside my bedroom window. That's normal enough, but today things sounded louder. Out in the afternoon to take a look:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Somebody must have trodden on the irrigation line and disconnected it. And that after I paid the best part of $1000 to have it overhauled! Fortunately it looks pretty moist in the coming days, so Jesse can fix it when he comes on Saturday.
| Tuesday, 28 November 2023 | Dereel | Images for 28 November 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Ten year anniversary
|
Topic: history, Stones Road house, opinion | Link here |
Ten years ago today, after much searching, we finally found a building site for a new house, in Stones Road. We bought it, and that's where we live now. How time flies! And how the place has changed. Today I braved sun, flies and mechanical problems to repeat a panorama that I took at the time—as it seems, even the time of day was almost the same. But it's almost impossible to see any similarities (run the cursor over an image to compare it with its neighbour):
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That's at least partially because of the choice of location, of course, but it's also interesting to see how much of the surrounding vegetation has changed.
|
41 year anniversary
|
Topic: history, general, opinion | Link here |
Why only one anniversary? Today's the 41st anniversary of my wife swap, the day I moved my old wife (Doris) out of the house and the new one (Yvonne) in.
|
The window manager killer identified
|
Topic: technology, opinion | Link here |
The problems I had last week with dying window managers haven't gone away. It happened again today, again while doing something with firefox. And restarting the window manager failed again with the same message:
Assertion failed: (ret != inval_id), function _XAllocID, file xcb_io.c, line 626.
OK, if it's firefox's fault, what happens if I shoot down firefox? I can restart the window manager. I was also able to restart it. So something in the previous process was responsible for the problem, and that it wouldn't go away.
|
More Black Friday pricing
|
Topic: general, opinion | Link here |
So now Black Friday is over, even Cyber Monday. Have the Washington Post's prices gone up? Of course not:
16 N 21-11-2023 To washingpost ( 697) The Washington Post N Unlimited access is A99c for a limited time
25 N 24-11-2023 To washingpost ( 694) The Washington Post N Nothing says “Black Friday” like our A99c sale
28 N 27-11-2023 To washingpost ( 699) The Washington Post N Get The Post for A99c
|
Dinner with the Marriotts
|
Topic: general, food and drink, history, opinion | Link here |
In honour of today's anniversary, we invited our neighbours Garry and Diane for dinner. We've been living next door for 8½ years, but somehow we don't see much of each other. We had dinner at their place a while ago—hardly more than 5 years. And we still don't know very much about each other, for example that Garry bakes his own bread too. They're in the process of moving into a house closer to Ballarat. When we bought the property, Garry said that when he moved out of the house, it would be feet first, so I'm trying to persuade him to let me take a photo of him in that position. But I didn't even get any photos this evening.
| Wednesday, 29 November 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Yet more new hardware needed
|
Topic: technology, opinion | Link here |
Seen in my nightly log files:
+ahcich4: Timeout on slot 11 port 0
+ahcich4: is 00000000 cs 00000000 ss 00000800 rs 00000800 tfd 40 serr 00000000 cmd 0000cb17
+(ada2:ahcich4:0:0:0): READ_FPDMA_QUEUED. ACB: 60 08 68 5a 6f 40 0d 00 00 00 00 00
+(ada2:ahcich4:0:0:0): CAM status: Command timeout
+(ada2:ahcich4:0:0:0): Retrying command
+ahcich4: Timeout on slot 24 port 0
+ahcich4: is 00000000 cs 00000000 ss 01000000 rs 01000000 tfd 40 serr 00000000 cmd 0000d817
+(ada2:ahcich4:0:0:0): READ_FPDMA_QUEUED. ACB: 60 08 a0 5a 6f 40 0d 00 00 00 00 00
+(ada2:ahcich4:0:0:0): CAM status: Command timeout
That's my 8 TB photo disk. It looks as if it's on its way out. OK, simple: replace it.
Oh. Not so simple. What do I replace it with? An 8 TB disk costs round $370. What does an 8 TB SSD cost? Are there even any 8 TB SSDs? Yes, though probably not as fast as the 4 TB one that I have in hydra. And they cost about twice as much. Clearly the tendency is for them to drop, but in the meantime it's just one more thing to wonder about. Are 8 TB even enough? Currently /Photos is nearly full:
=== grog@eureka (/dev/pts/43) ~ 1 -> df /Photos
Filesystem 1048576-blocks Used Avail Capacity Mounted on
/dev/ada1p1 7,629,565 6,684,090 869,179 88% /Photos
That's misleading, though. A lot of the stuff in there are TIFF intermediate files that could (should) be removed. But 12 TB might be a better size—if I can find one.
|
Keyboard remapping: done!
|
Topic: technology, opinion | Link here |
Why am I having so much trouble remapping my Sun Type 7 keyboard? It worked fine last time. But this time I've been trying to do things incrementally, and that seems to be a problem.
OK, start again from scratch. What do I want to do?
Save the default configuration as revision 1.1 of ~/keyboards/sun-map-hydra.
Remap the keys above the main keyboard, marked F1 to F12. Set them to F21 to F32. The key codes are 67 to 76 (F1 to F10), then 95 and 96 (F11 and F12).
The 2 keys to the left of the F keys (marked Help, Esc): Map to F11 and F12. Key codes are 146 and 9.
2 columns of keys to the left of the main keyboard, marked Stop to Cut. Map them to F1 to F10. Keycodes are 136 137 ... 145. On eureka they're 145 to 154.
Grave/Tilde key at top left of the main keyboard, to the right of Again. Map to Esc.
CapsLock key: map to Control_L. Key code is 66.
Bottom row of main keyboard (Control, •, Alt, space, Alt Graph, •, Compose: Map to Meta_L, Grave/Tilde, Meta_L, (space), Mode_Switch, ISO_Level3_Shift, Multi_key. Keycodes are 37, 133, 64, 65, 108, 134, 135. And yes, it has two Meta_Ls. I don't use the second one, so there's an opportunity there. On eureka the codes are 37, 115, 64, 65, 113, 116, 117.
Some of this is confusing. According to the man page, the second keysym is chosen when the key is pressed along with the shift key. But on eureka I have:
keycode 37 = Meta_L NoSymbol Meta_L
keycode 113 = Alt_R Meta_R Alt_R Meta_R
The first one is the key I use for Alt, but it doesn't have that name, and Alt-Shift combinations work fine. And there doesn't seem to be any distinction between Alt and Meta. xev also doesn't show any difference.
And then there's the question of the modifiers. The man page for xmodmap only describes the first four keysyms, but this page, courtesy of Daniel Nebdal (again), states:
Each keysym column in the table corresponds to a particular combination of modifier keys:
But the default keymap that I dumped had 8 keysyms, and the eighth is frequently different, like for the F keys above the main keyboard:
keycode 67 = F1 F1 F1 F1 F1 F1 XF86Switch_VT_1
How is that supposed to work?
And then there's this:
keycode 92 = ISO_Level3_Shift NoSymbol ISO_Level3_Shift
How does that tie in with the sixth keycode in the list above?
Still, I changed things as above, loaded the map and... it worked! That should have been obvious, but even now I have unanswered questions.
|
More X woes
|
Topic: technology, opinion | Link here |
It's high time I complete my migration to X on hydra. There's something seriously wrong with x2x between eureka and hydra. The cursor seems to land at random places, not made any easier by the fact that I can't see it disappear left from eureka:0.0 (there's no monitor connected). And maybe another problem is related, the random killing of the window manager on hydra:0.3. I had blamed it on firefox so far, but today it happened while running chrome. High time to migrate.
What's stopping me? For reasons I don't understand, I can't get the Alt bindings to work. I started without the all-important entries in ~/.Xdefaults:
! Make alt key work again (20221124)
*VT100*utf8: 1
*VT100*eightBitInput: false
*VT100*eightBitControl: false
*VT100*eightBitOutput: true
OK, that's easy to fix. I've done that already on eureka, and it worked with the running X server. But this time it didn't make any difference. Still more stuff to investigate.
|
More bad dictionaries
|
Topic: rant, language, general, opinion | Link here |
It's been five years since I ranted about poor definitions in some dictionaries. I was particularly irritated by the definition of “local time” in the online version of the Merriam-Webster dictionary:
: time based on the meridian through a particular place as contrasted with that of a time zone
That's nonsense, of course, and the examples they produced at the time clearly referred to time zones. But they have changed; they seem to choose recent quotations at random and without checking. Still, it's worth a comment, which will probably land in /dev/null:
Your definition claims that "local time" is not related to time zones, but the examples all relate to time zones. I believe that your definition is incorrect, not the examples. All other reputable dictionaries that I have found equate local time to the prevailing time zone. See OED for a clear definition, and http://www.lemis.com/grog/diary-nov2018.php?subtitle=Local%20time%20and%20standard%20time&article=D-20181130-012938#D-20181130-012938 for further background.
It'll be interesting to see if I get any reaction.
| Thursday, 30 November 2023 | Dereel | |
| Top of page | ||
| previous day |
|
More pancreas problems?
|
Topic: health | Link here |
Yvonne had a call from Kontoku Shimokawa today about the results of her latest pancreas MRI examination. Something could be happening. So far the worst news is that she will need another ordeal at the Epworth Hospital, an EUS and maybe FNA, like she had 5 years ago
|
hydra: cutover?
|
Topic: technology, opinion | Link here |
So: what do I need to do to cut over the X display from eureka to hydra? In principle the only outstanding issue is that I still can't use the Alt key with bash. But that can be fixed later.
OK, move to the left. It's amazingly uncomfortable. Yes, I knew that it was only a temporary arrangement, but I was surprised how uncomfortable it is. Different screen heights, drawers in front of the left end of the displays, a mouse with a cord! That, too, I had forgotten.
OK, back to check the Alt key issue. When I first fixed it on eureka, I noted two things: set the values in ~/.Xdefaults and add some entries to inputrc. But I only did the first: the inputrc I found had the fully qualified name /compat/linux/etc/inputrc, clearly a Linuxism, and things proved to work without it.
OK, research further. Came up with this page, which was in fact interesting: it's part of readline, and it's used by bash, not xterm. . Does my bash use it? ktrace shows:
1063 bash CALL openat(AT_FDCWD,0x826511b40,0<O_RDONLY>)
1063 bash NAMI "/home/grog/.inputrc"
1063 bash RET openat -1 errno 2 No such file or directory
1063 bash CALL openat(AT_FDCWD,0x826511b40,0<O_RDONLY>)
1063 bash NAMI "/usr/local/etc/inputrc"
1063 bash RET openat -1 errno 2 No such file or directory
OK, we can fix that. Copy the file to /usr/local/etc/inputrc and try again. Yes, it finds it, reads it in, finds the correct entries (and a whole lot more). And nothing changes.
What characters is the keyboard generating? Ran hexdump against the keyboard, thus bypassing bash. On eureka, I got (for Alt-a):
=== grog@eureka (/dev/pts/48) ~ 1 -> hexdump -C
^[a^[a^[a^[a^[a^[a^[a^[a
00000000 1b 61 1b 61 1b 61 1b 61 1b 61 1b 61 1b 61 1b 61 |.a.a.a.a.a.a.a.a|
But on hydra I got:
=== grog@hydra (/dev/pts/54) ~/keyboard 1 -> hexdump -C
áááááááááááááááá
00000000 e1 e1 e1 e1 e1 e1 e1 e1 e1 e1 e1 e1 e1 e1 e1 e1 |................|
So clearly nothing that I change in readline is relevant to the problem. The rest of inputrc contains the standard Emacs bindings, so I assume that the values are the defaults.
OK, found another thread on the X web site, with an interesting reply:
Edit ${HOME}/XTerm and add the following lines:
*vt100.altIsNotMeta: true
*vt100.altSendsEscape: true
That seems to have fixed it for the original poster, and it seems to make sense. It also brings home to me things that I had forgotten decades ago. First, the application default files, in this case /usr/local/lib/X11/app-defaults/XTerm. And arguably that's where the settings all belong, not ~/.Xdefaults. But then, the xterm man page contains, in the FILES section,
/usr/local/lib/X11/app-defaults/XTerm-color
the xterm color application resources. If your display supports
color, use this
*customization: -color
in your .Xdefaults file to automatically use this resource file
rather than /usr/local/lib/X11/app-defaults/XTerm. If you do not
do this, xterm uses its compiled-in default resource settings for
colors.
And also, in passing, the reference to VT100, a terminal that was introduced 45 years ago. Much of this looks like ancient history.
OK, so has my default XTerm changed? No, the new file is identical to the one on eureka, starting with
! $XTermId: XTerm.ad,v 1.101 2016/01/28 02:15:29 tom Exp $
Are different values compiled in, maybe?
Still, I should try out the suggestion. Only: I want to understand what's going on. Is there a way to display the settings that the xterm instance is using? More research brought me to this page, which suggests using editres to get the Xt widgets tree. Do I have editres? It wasn't there, but I can install it. But by this time it was evening, I was in the lounge room, and the editres display on my TV was too small to read. Mañana.
Before completely leaving the subject, discovered that there's also a -query option to xrdb, and that seems to be useful.
|
Which 8 TB SSDs?
|
Topic: technology, opinion | Link here |
Spent a little more time looking round for 8 TB SSDs. There's precious little choice: only the Samsung 870 QVO. Scorptec, Umart, Mwave offer it for $899. Megabuy has it for only $886.88, though it's not clear whether that's the cheapest, since all are too polite to mention shipping charges. That makes it almost worthwhile, except that I'd really prefer a 12 TB drive. But I think the days of spinning rust are over. Should I wait until the current drive shows more problems, or should I bite the bullet?
Surprisingly, only 9 months later the product “has been discontinued”, and there seems to be no replacement.
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 |
