|
| Greg's diary |
| October 2023 |
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Sunday, 1 October 2023 | Dereel | Images for 1 October 2023 |
| Top of page | ||
| next day | ||
| last day |
|
DST again
|
Topic: general, photography, opinion | Link here |
Daylight Saving Time started today, and with it the usual clock resetting.
But my Olympus OM-D E-M1 Mark II and friends have OI.Share, a mobile phone app that connects to the camera, and in the process sets the date. OK, fire up. “Can't connect”. By the time I found out what was wrong, I had established that the other phone (enzian.lemis.com) no longer had the app installed.
OK, why do I need to do it this way anyway? The camera has a relatively straightforward method to set the date, and like almost anything to do with OI.Share, it's so much easier. This app is really completely useless.
|
Non-stick omelette revisited
|
Topic: food and drink, opinion | Link here |
A couple of years ago I did some experiments to fry a thin “omelette” without oil in my then-new non-stick ceramic frying pans. They were never completely successful, mainly because the frying surface was slightly domed, but from time to time I tried again.
Today was the next time. And, I think, the last time. The pans were very non-stick when I got them, but like so many ceramic things, they wear out quickly. Here a comparison between January 2022 and now:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Yes, the colour shows that it's a different pan, but in fact the green one is newer. It still works as well as most non-stick pans, but the biggest issue now is that the thing has a domed surface. So until I find better pans, that experiment has done its time.
|
Android networking: why so castrated?
|
Topic: technology, general, opinion | Link here |
Discussing my issues with FTP on hirse today. Android is based on Linux, which has a full-featured TCP/IP stack. Why did the author of WiFi FTP Server have to reimplement the stack?
Once again people on IRC amaze me. They're not stupid—in fact, quite the contrary—but they seem to accept the bad design decisions as something normal, even necessary. “A mobile phone is not a computer”. "Why use TCP/IP when you can upload to Google and back down again?" That kind of attitude is bad for my blood pressure.
Of course, there's an assumption on my part here: Why, did the author of WiFi FTP Server have to reimplement the stack? Does Android really not supply that as part of the system? How about asking the experts, the artificial Intelligence bots?
|
Android networking: the experts speak
|
Topic: technology, opinion | Link here |
I have assumed that my problems with WiFi FTP Server were due to a poor reimplementation of the TCP/IP stack. But is that assumption correct? Asked my tame artificial Intelligence bots: “Does Android have a full TCP/IP stack?". ChatGPT tells me
Yes, Android does have a full TCP/IP stack. It is based on the Linux kernel and includes all the necessary protocols and components for communication over the internet, such as TCP, IP, UDP, and ICMP. The stack allows Android devices to establish and maintain network connections, send and receive data packets, and handle various network-related tasks.
Bard is even clearer:
Yes, Android has a full TCP/IP stack. This means that it can implement all of the protocols in the TCP/IP suite, from the link layer to the application layer. This allows Android devices to communicate with other devices on the network using a wide range of protocols, including HTTP, FTP, and SMTP.
That makes perfect sense. But why does it work so badly on hirse? Whose fault is it?
|
Next bushfire!
|
Topic: general, technology, opinion | Link here |
The bushfire season is really coming early this year. Nothing more round here, but in Gippsland there are a number. This one concerned me:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Where's that? North of Briagolong, where my cousin Gill and husband Kline live. How close is it? Compare with Google Maps:
|
|
|||||||||
Now doesn't that say a lot? I have the exact coordinates (on nearly every photo I took at the memorial for my aunt Freda last January: the house is at 37°48'40.1"S 147°06'28.0"E). But how do I locate it on the VicEmergency map? That's not directly VicEmergency's fault (this time), but it should be possible to locate places of interest on the map.
My best bet is that they're at the very bottom of the outer area. The good news is that the fire is moving in a north-easterly direction, so they're unlikely to be seriously impacted. And so far we haven't seen anything on Facebook.
| Monday, 2 October 2023 | Dereel | Images for 2 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Panic!
|
Topic: technology, opinion | Link here |
Into the office first thing this morning. Number 2 monitor on eureka displayed
Boot:
Another crash! Why? Recovered as normal, and then saw in the logs:
panic: cannot reassign paging buffer
Well, that's a different one, anyway. Fought my way through the reboot, but I couldn't start the weather station software: device not found. No obvious reason for that, so I shut down, powered off and restarted. Then everything worked, sort of. I heard a regular “click” from somewhere behind the monitors, which I traced to a 2½" USB disk drive that I no longer use, but which I haven't removed because it helps maintain the device numbers. When I picked it up, still running, it stopped clicking. Why?
Picking up the pieces showed that all the normal backups had completed normally. There are only 3 hierarchies that I back up with rsync, including my multiple FreeBSD sources, several million tiny files. I didn't bother to check how far it got there.
Later, looking at the console messages, I saw:
Oct 2 03:47:01 eureka kernel: (da2:umass-sim2:2:0:0): WRITE(16). CDB: 8a 00 00 00 00 04 18 4a c3 68 00 00 00 80 00 00
Oct 2 03:47:01 eureka kernel: (da2:umass-sim2:2:0:0): CAM status: CCB request completed with an error
Oct 2 03:47:01 eureka kernel: (da2:umass-sim2:2:0:0): Retrying command
...
Oct 2 09:47:10 eureka kernel: (da2:umass-sim2:2:0:0): WRITE(16). CDB: 8a 00 00 00 00 04 18 4a c3 68 00 00 00 80 00 00
Oct 2 09:47:10 eureka kernel: (da2:umass-sim2:2:0:0): CAM status: CCB request completed with an error
Oct 2 09:47:10 eureka kernel: (da2:umass-sim2:2:0:0): Error 5, Retries exhausted
Oct 2 09:47:10 eureka kernel: g_vfs_done():da2p1[WRITE(offset=9004760137728, length=131072)]error = 5
Oct 2 09:47:10 eureka kernel: g_vfs_done():da2p1[WRITE(offset=9004760268800, length=131072)]error = 5
(many more)
Oct 2 09:47:10 eureka kernel: g_vfs_done():da2p1[WRITE(offset=23464448, length=10240)]error = 5
Oct 2 09:47:10 eureka kernel: panic: cannot reassign paging buffer
Oct 2 09:47:10 eureka kernel: cpuid = 4
Oct 2 09:47:10 eureka kernel: KDB: stack backtrace:
Oct 2 09:47:10 eureka kernel: #0 0xffffffff80982e10 at kdb_backtrace+0x60
...
Oct 2 09:47:10 eureka kernel: #9 0xffffffff808a433b at g_disk_done+0xfb
Oct 2 09:47:10 eureka kernel: #10 0xffffffff80980579 at bioq_flush+0x39
Oct 2 09:47:10 eureka kernel: #11 0xffffffff8030e91f at dadone+0xa9f
That's a smoking gun: the USB backup disk /dev/da2 was misbehaving. But was that the fault of the disk? Looking at the issue with the weather station, it looks more like the USB bus. And something very similar happened this time last month. Roll on the new machine.
|
Artificial stupidity again
|
Topic: technology, opinion | Link here |
Silly question on Quora today:
Well clearly that's silly. There are a large number of solutions to those constraints. But to my surprise, none of the answers were complete and correct. OK, once again artificial intelligence to the rescue. Bard tells me:
There are only three possible three-digit numbers that satisfy the given conditions:
But that's just plain wrong! It doesn't fit the constraints! Quora's “ChatGPT” says:
Sure, let's think about this. If the tens digit is more than the ones digit, that means that the ones digit must be a number between 0 and 8. If the hundreds digit is seven less than the tens digit, that means that the hundreds digit must be a number between 0 and 8 as well. Putting all of this information together, we can see that the three-digit number must have a ones digit between 0 and 8, a tens digit between 1 and 9, and a hundreds digit that is seven less than the tens digit.One possible number that fits this description is 748. The ones digit is 8, the tens digit is 4, and the hundreds digit is 7. Is this the number you were thinking of?
And the real ChatGPT says:
Hi! Let's solve this together. If the tens digit is more than the ones digit, that means the tens digit is at least 1 greater than the ones digit. And if the hundreds digit is seven less than the tens digit, that means the hundreds digit is at least 7 less than the tens digit.
Let's start by assuming the ones digit is 0. In that case, the tens digit would be at least 1, and the hundreds digit would be at least -7. But since we're dealing with positive digits, let's try assuming the ones digit is 1. In that case, the tens digit would be at least 2, and the hundreds digit would be at least -5.
Let's keep going. If the ones digit is 2, the tens digit would be at least 3, and the hundreds digit would be at least -2. If the ones digit is 3, the tens digit would be at least 4, and the hundreds digit would be at least -1.
Now, if the ones digit is 4, the tens digit would be at least 5, and the hundreds digit would be at least 0. And if the ones digit is 5, the tens digit would be at least 6, and the hundreds digit would be at least 1.
Finally, if the ones digit is 6, the tens digit would be at least 7, and the hundreds digit would be at least 2. And if the ones digit is 7, the tens digit would be at least 8, and the hundreds digit would be at least 3.
So, based on these possibilities, the three-digit number you're thinking of could be 102, 213, 324, 435, 546, 657, or 768.
These, too, are completely wrong! But there are so many wrong answers that I'm left wondering if I'm not the problem. If so, it must be serious, because I don't see it. If that's artificial “intelligence”, I'd hate to see natural ”intelligence”.
|
Human intelligence declining?
|
Topic: general, technology, opinion | Link here |
I'm still trying to understand people who tell me that mobile phones aren't computers, and that it makes sense to copy files to local destinations via Google at the other end of the world. But the nonsense continues. Today I read on Facebook that a class of science students wasn't able to find files on their computers:
Gradually, Garland came to the same realization that many of her fellow educators have reached... the concept of file folders and directories, essential to previous generations' understanding of computers, is gibberish to many modern students.
Arguably there's nothing wrong with this, if there's a good replacement. But I don't see one. The current situation just dumbs people down.
Seen from the other side, Alan Clegg posts:
Microsoft continues to mystify me by putting files into directories that I can't find.
There is a desktop, and then there is a OneDrive desktop that I somehow inherited because I now use Office in the cloud (my words, not theirs).
There is a "Documents" folder that the OS swears is on "My Computer" but I can't find the bits when I look for them in "C:" .. maybe the "me" in "my" is someone in Redmond (or, more likely, Banglore).
The demarcation between what is physically where has become difficult to determine and I am not happy.
But then, Alan's a computer expert. What about non-technical people like Diane Saunders? She has only now, in the last few days, found somebody to show her how to connect to Facebook. But that should belong in the Brave New World of inscrutable “devices”. Why did she have trouble, then?
|
Weather graph strangeness
|
Topic: technology, opinion | Link here |
It's been nearly 14 years since I started using gnuplot to plot the graphs for my weather station, and from the start I had difficulties. One of them was to convert the time to gnuplot internal time (the number of seconds since 1 January 2000, 0:0 local time). That's clear enough: subtract the time_t value of the date. But for reasons I never understood, things only worked when Daylight Saving Time was active, and when it finished I discovered I had to adjust by 2 hours!
Clearly a workaround, but it has worked for well over 10 years—until now. Now the time is correct, but the date is off by a day. Why? “I didn't change anything”.
| Tuesday, 3 October 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Following the Briagalong bushfire
|
Topic: general, opinion | Link here |
When we went to bed yesterday, everything seemed relatively under control in Briagolong. But then the weather changed, and there was a lot of wind. And the wind that was from the south-west yesterday, had changed to north-west. This morning the recommendations were no longer to evacuate: it's too late! And the fire's still not under control.
The only hope: rain. And it was on its way. The weather forecast predicted up to 75 mm of rain! At least that meant that the threat was no longer fire, but flooding.
Somehow the area seems particularly dangerous. 17 years ago we considered moving into the area, but this isn't the first bushfire to cause concern—I had noted one only a little earlier—and it also seems to be prone to flooding. But somehow we can't rely on anything any more.
|
The very last night of the Proms
|
Topic: music, opinion | Link here |
There are two annual concerts that we like to watch: the New Year's concert of the Vienna philharmonic and the Last Night of the Proms. The New Year's concert is relatively easy to get from German TV, and I first watched it 60 years ago. Very Austrian, not completely serious:
|
|||||||||||||||||||||||
The Proms are a different matter altogether, very British of course, very patriotic, and much less serious than the New Year's concert. They're also more difficult to get: the licensing conditions seem to have changed, and last year the event was canceled. But this year we got the recording from a German TV station, and we've been watching it gradually.
I think it's the last time. Somehow it was too commercial and not musical enough. And the performance itself was suboptimal. Much of the music was boring, the performers new to the stage. For the first time, I think, they played songs from other parts of the United Kingdom: Londonderry Air, the Skye Boat Song, and something Welsh that I didn't recognize. Sign of equality? But they left out two verses of the Skye Boat Song, the ones referring to the devastation of Culloden and the return of Bonnie Prince Charlie. Not politically correct, maybe?
Then: Rule Britannia is a powerful song, but this time it was sung by a coloratura soprano who could barely make herself heard (clearly also a technical issue), and here, too, more colonial verses had been omitted, though maybe that's always the case. Even Land of Hope and Glory somehow fell flat. So yes, I think this was our last night of the Proms.
In passing, it's interesting to compare the Musikverein to the Royal Albert Hall. The latter is much bigger and more spacious, and seems much newer. But in fact they are both almost exactly the same age.
| Wednesday, 4 October 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Finally fix the irrigation system?
|
Topic: gardening | Link here |
I've continually had trouble with the irrigation system in the garden, and this year is no exception. For some reason, almost nobody else understands the issues, which seem simple to me. And I'm getting too old to do the work myself. UPI? Called them up, but they're booked out. They gave me the number of Devin Lafranchi, whom I called today. He thinks he can get to it later in the month.
|
Briagolong emergency: now the rain
|
Topic: general, opinion | Link here |
The hot, dry weather has broken. Now we have rain, about 20 mm over the last couple of days. And in Gippsland? Since yesterday midday until 9:00 this morning they had exactly 100 mm of rain! That should put out the bushfires.
But it didn't. The bushfire near Briagolong is still uncontained. At least it's no longer a danger to anybody, and at least where Gill and Kline live they're in no danger of flash flooding.
How will the summer progress?
|
Configuration strangenesses
|
Topic: technology, opinion | Link here |
It's high time to continue looking at the makefs bug. But as a result of Monday's panic, kimchi, my NetBSD VM, had gone away.
Not a problem. Reboot. Oh. I had lost some of the VirtualBox modules necessary to run. Never mind, that's why I documented them.
But it didn't work!
Failed to open a session for the virtual machine kimchi.
Failed to open/create the internal network 'HostInterfaceNetworking-em0' (VERR_SUPDRV_COMPONENT_NOT_FOUND).
Failed to attach the network LUN (VERR_SUPDRV_COMPONENT_NOT_FOUND).
Result Code: NS_ERROR_FAILURE (0x80004005)
Component: ConsoleWrap
Interface: IConsole {872da645-4a9b-1727-bee2-5585105b9eed}
What's that? Much messing around, including (finally!) putting the load commands in /boot/loader.conf, but the problem persisted. Reboot? Not I! Check with other systems. I didn't get that error message when restoring a running configuration, but the network still didn't work.
But then later, suddenly, it worked! Why? I had done enough in between that I don't know. But somehow it seems to be one of those “phase of the moon” things.
Another was with X. For reasons I no longer recall, I have done some modifications on the configuration. One was the xterm configuration change to use UTF-8, which I noted a year ago. My guess is that once again I set the configuration but didn't change to config files, with the result that the Alt key bindings with bash no longer worked.
OK, that one's simple, as I noted at the time. Add to ~/.Xdefaults:
! Make alt key work again (20221124)
*VT100*utf8: 1
*VT100*eightBitInput: false
*VT100*eightBitControl: false
*VT100*eightBitOutput: true
And that worked.
Well, for one xterm, but not for others. Presumably I needed to do it for each of them. But that, too, proved to be incorrect. It almost seems that they gradually learnt of the configuration change, but I can't understand why.
Bottom line: things work, but I'm frustrated because I don't know why it happened this way.
|
British Indian food
|
Topic: food and drink, opinion | Link here |
Yvonne is getting harder to please with south and east Asian food. In particular, she doesn't like more than the tiniest hint of chili any more, to the point where I have almost stopped cooking Indian food.
But this week ALDI had “Indian” food on special. It was mainly from Patak, a British company with Indian-like food about whom I have ranted in the past. But some of the pastes are “mild”, so I left it to Yvonne to decide what to buy.
She returned with a “butter chicken” paste, which wanted one onion, some garlic, some ginger, tomatoes, an unspecified quantity of chicken and a forgotten amount of cream to go with a half jar of paste. How I hate these vague recipes! In the end went my own way with 80 g onion, 15 g each of garlic and ginger, no tomatoes or cream, 500 g of chicken and the half jar of paste. The result? I had forgotten how bad these things taste, all fenugreek. Even Yvonne found it boring and prefers my own chicken curry recipe, the one I no longer like.
Goodbye, Patak. Hopefully I have learnt my lesson.
| Thursday, 5 October 2023 | Dereel | Images for 5 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
System down!
|
Topic: technology, opinion | Link here |
Part of my regular morning ritual is to check the state of my machines. Most are local, but one is in Los Angeles, and another is in Frankfurt am Main. And today I read:
=== grog@ffm (/dev/pts/0) ~ 4 -> date; uptime
Thu Oct 5 00:34:27 UTC 2023
12:34AM up 13:37, 1 user, load averages: 0.19, 0.32, 0.36
Aaarrrgh! Yesterday it was:
=== grog@ffm (/dev/pts/0) ~ 5 -> date; uptime
Wed Oct 4 00:19:25 UTC 2023
12:19AM up 2067 days, 9:22, 2 users, load averages: 0.50, 0.57, 0.48
System down after 5 years, 8 months! My longest uptime ever, cut down in its prime! What happened? /var/log/messages tells me:
Oct 4 06:18:08 ffm named[87410]: creating IPv4 interface lo0 failed; interface ignored
Oct 4 06:18:08 ffm named[87410]: not listening on any interfaces
Oct 4 10:57:34 ffm syslogd: kernel boot file is /boot/kernel/kernel
Oct 4 10:57:34 ffm kernel: Copyright (c) 1992-2017 The FreeBSD Project.
That spurious error message from named repeated every hour, and there would have been another kind of error at 6:27. so the system must have gone down between those times. But it didn't reboot until 10:57, 4½ hours later. So: my guess is that DigitalOcean either had a catastrophic failure, or, more likely, they rebuilt their infrastructure. Once upon a time they told me about this sort of thing, but they seem to have stopped their information emails.
|
Bruno was here
|
Topic: animals | Link here |
|
|||||||||||||||||||||||
|
More Indian spice mixtures
|
Topic: food and drink, opinion | Link here |
Yvonne was off in town all day at a riding clinic, More Indian food for dinner this evening. I have a number of spice mixtures that really need using up, including one for Rogan josh. It's a powder, but strangely the additional ingredients look almost identical to those of the chicken paste yesterday. It's straightforward enough, but once again the result was boring. Where can I find good Indian food?
| Friday, 6 October 2023 | Dereel | Images for 6 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Rumours of my demise, or, Microsoft hates me
|
Topic: technology, general, opinion | Link here |
Mail from Michael James today, with startling information:
Oh. They say that if you're on the Internet, nobody knows if you're dead, and I can confirm that from my list of Facebook friends, which includes at least 5 dead people. But myself? I feel like Mark Twain.
This particular claim came from Bing, a Microsoft product, which I don't use. It's in competition, if that's the word, with ChatGPT and Bard. Did a bit of playing around with it, and it seems that its output is very raw. It no longer told me that I was dead, but it produced output from three sources, mainly my own mini-bio for my O'Reilly books.
The whole thing seemed a little uneven, definitely not the relatively good summary that ChatGPT produced. About the only interesting thing was the background for the claim that I took many pictures of the West End of London in the 1960s. That proved to be a link to Flashbak, a site that I had never heard of. And yes, there are some large, badly sorted images of London that I took between 1965 and 1968. Creative Commons copyright? I can't recall that. I wonder what else there is out there.
|
Strange error message sensitivity
|
Topic: technology, opinion | Link here |
Interesting mail from Michael Hughes today. Summarizing:
Today I was working on a script and did the following command:
echo "" > file 2>/dev/nullWhen I tried this on a file my user didn't have write permissions on, I recieved the error:
cannot create file: Permission denied
The error message didn't go to /dev/null.
But he didn't get an error message when when he tried:
echo "" 2>/dev/null > file
What's going on here?
So my guess is that the error message comes before the command setup is complete, when it runs into the inability to open file. In the first case, stderr is still as it was in the parent process, but in the second case it has already been redirected. An interesting detail, and it's confirmed by:
=== grog@eureka (/dev/pts/4) /var/tmp 30 -> echo "" 2> bah > file=== grog@eureka (/dev/pts/4) /var/tmp 31 -> cat bah
bash: file: Permission denied
Here the error message is the one stored in bah.
|
NetBSD again
|
Topic: technology, opinion | Link here |
It's getting high time to look at the makefs bug, as I've been saying for some time. How? I can work my way through the code, but there's a simpler way: I can compare it to a version that works—on NetBSD. OK, get an up-to-date source tree. Don't I have a script somewhere? Yes!
#!/bin/sh
# $Id: update-NetBSD-cvs,v 1.1 2003/07/20 03:42:35 grog Exp $
PATH=$PATH:/usr/local/bin
echo ====== `date`: Getting NetBSD CVSUP updates
cd /src/NetBSD/1.5-CURRENT/src
cvs -d :pserver:anoncvs@anoncvs.netbsd.org:/cvsroot update -P -d
echo ====== `date`: Getting NetBSD CVSUP updates
But look at that date! Over 20 years ago! Does the method still work? It's certainly not worth trying to update version 1.5; the current is 9.3. OK, RTFM. Not easy, somehow, mainly because the Google search results weren't what I was looking for. But the information is there, and it hasn't really changed. Basically, it's as simple as:
export CVS_RSH="ssh"
export CVSROOT="anoncvs@anoncvs.NetBSD.org:/cvsroot"
cvs co src
And it worked. What a difference from git!.
|
Rani's ghost lives on
|
Topic: animals | Link here |
It's been over 6 years since we reluctantly parted with Rani, our Bengal cat: she had too many bad habits. One of the lesser ones was that she kept trying to get through the ceiling:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
But today we found the real reason for yesterday's dislodged cards on the fridge:
|
|||||||||||||||||||||||
That was Bruno. I had thought that he had climbed on the scratching post, but no, there was another way:
|
|||||||||||||||||||||||
He had jumped from the kitchen work area to the fridge. Later he went even higher:
|
|||||||||||||||||||||||
How do we cure him of that?
| Saturday, 7 October 2023 | Dereel | Images for 7 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Trap for Bruno
|
Topic: animals, opinion | Link here |
How do we stop Bruno from jumping on the fridge? Unpleasant surprise when he gets there?
|
|||||||||||||||||||||||
That might do it, but after some consideration I decided that the collateral damage could be too much.
|
A compost heap!
|
Topic: gardening | Link here |
It's been well over 8 years since we moved in to Stones Road, and almost since the beginning we've had a compost heap behind the dog run. Put stuff in, never take it out. Surprisingly, it's overflowing.
But now that Yvonne has moved her dung heap, the one next to the compost heap is unused. More compost! Today Jesse Walsh came and restructured it, widening the left-hand bay into the ex-dung heap:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Now doesn't that oldest layer look good?
|
Chicken and prosciutto again
|
Topic: food and drink, opinion | Link here |
It's been almost exactly a year since we last made prosciutto-wrapped chicken with lemons, spring onions and potatoes. Time to do it again. In the process, a couple of minor changes, but effectively it's the same.
|
|||||||||||||||||||||||
This time it didn't taste as exciting as previously. Why?
| Sunday, 8 October 2023 | Dereel | Images for 8 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Oh, the springtime it brings on the laying
|
Topic: opinion, food and drink | Link here |
It's spring, and many of our friends have hens. At this time of year the hens lay more than they can eat. So they give them away, including to us:
|
|||||||||||||||||||||||
That photo was taken after Yvonne gave one pack away. What do we do with them all? The bigger ones on the right are duck eggs, which Yvonne eats for breakfast. And I don't eat many; what we have there is enough for me for 6 weeks. Nothing for it: I'll eat an omelette for breakfast.
But what kind? Most of my breakfasts are from south-east Asia or nearby. Recipes? Very few. The standard onion, garlic and ginger, and something else? Decided on prawns today. 4 smaller eggs, together 160 g. 60 g prawns. 10 g ginger, 15 g garlic, and a couple of spring onions:
|
|||||||||||||||||||||||
What a mess! But that's the fault of the once-non-stick frying pan:
|
|||||||||||||||||||||||
The pan frustrates me almost more than the omelette, which was boring. What else can I do with eggs?
| Monday, 9 October 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More monitor dropouts
|
Topic: technology, opinion | Link here |
It's been over a week since a series of daily short dropouts on monitor 2. And then nothing—until today, when it happened in the same manner as before: a short dropout round 10:00. This time I didn't attempt any recovery, and it still returned after a few seconds. Then, for the first time, it was immediately followed by a second dropout, which again recovered by itself.
What is this? Overvoltage? No, Powercor has lowered the mains voltage to “only” round 240 V, and at the time that was the voltage; it was higher later in the day when I was feeding in power to the grid. Nothing much that I can do except observe and maybe replace the monitor or its power supply.
|
More bug hunting
|
Topic: technology, opinion | Link here |
I'm still working at a snail's pace on the makefs bug. One thing that it has brought home to me is how much the time depends on being active in the scene. And I really seem to have little time to spare, so I only work for the odd half hour or so.
The other issue is: how do I proceed? I have two comparisons that show the difference between the bug and correct behaviour: FreeBSD without size specification, and NetBSD with or without size specification. The obvious thing is to step through the program along with one of the others and compare what happens. But which? Today I tried the FreeBSD version and got as far as the function ffs_mkfs(), which proves to be a modified copy of the mkfs() function in /usr/src/sbin/newfs/mkfs.c. And then some optimization threw out my superblock! Do I need to recompile the whole bloody thing?
| Tuesday, 10 October 2023 | Dereel | Images for 10 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Bruno was here
|
Topic: animals, food and drink, opinion | Link here |
Preparing breakfast today, I left some noodles to drain:
|
|||||||||||||||||||||||
Is Bruno particularly bad, or have we just forgotten what it's like to have a cat growing up in the house? Certainly Rani was worse, but the others?
|
Preparing for fire season
|
Topic: general, photography, history, opinion | Link here |
An Eddy, whom I don't know, along today to exhort me to prepare for the coming fire season, which could be bad. He brought a thick envelope and this flyer:
|
|||||||||||||||||||||||
Nice photo. I recognize it: I took it nearly 15 years ago:
|
|||||||||||||||||||||||
They presumably got it from the Wikipedia page Dereel. Were they allowed to take it? It seems that I released it under the Creative Commons Attribution-ShareAlike 4.0 International license. But that word “Attribution”: “You must give appropriate credit , provide a link to the license, and indicate if changes were made”. And they didn't do that.
Should I complain? No, I don't think so, though it might be worth informing the council of their obligations; others might not be as lenient.
|
Software archaeology
|
Topic: technology, opinion | Link here |
Now I'm gradually in a position to debug my makefs bug. After some fighting changes in gdb, I finally got into the meat of the ffs_mkfs() function, which looks like being the culprit. It builds up a superblock, which is referred to as sblock. But gdb doesn't want to know:
Breakpoint 4, ffs_mkfs (fsys=fsys@entry=0x7fffffffe8e1 "./image2",
fsopts=fsopts@entry=0x7fffffffe2f0, tstamp=tstamp@entry=1696902845)
at /eureka/home/src/FreeBSD/git/main/usr.sbin/makefs/ffs/mkfs.c:193
193 sblock.fs_avgfpdir = avgfpdir;=== gdb -> p sblock
No symbol "sblock" in current context.
Huh? Optimized out? Time to build without optimization. How do I do that? Tried simply with
=== grog@dereel (/dev/pts/6) /usr/src/usr.sbin/makefs 12 -> CFLAGS= make
And how about that, it worked! I could find all the other variables, but not sblock. Finally it dawned on me. At the top of ffs_mkfs.c I found:
static union {
struct fs fs;
char pad[SBLOCKSIZE];
} fsun;
#define sblock fsun.fs
And sure enough,
=== gdb -> p fsun.fs
$5 = {
fs_firstfield = 0,
fs_unused_1 = 0,
...
There are many members in the structure, so many that it's difficult to find them. What do things look like in other versions? A bit of investigation showed:
/src/NetBSD/CURRENT/src/usr.sbin/makefs/ffs/mkfs.c
union {
struct fs fs;
char data[SBLOCKSIZE];
} fsun;
#define sblock fsun.fs
This really came from the old mkfs code, now wrapped in newfs. And there I see:
/src/NetBSD/CURRENT/src/sbin/newfs/mkfs.c
union {
struct fs fs;
char data[SBLOCKSIZE];
} *fsun;
#define sblock fsun->fs
/src/FreeBSD/git/main/sbin/newfs/mkfs.c
static struct csum *fscs;
#define sblock disk.d_fs
Only the newest versions of FreeBSD differ. How old is this code? The oldest that I can find was 4.0BSD, dated 1 October 1980:
union {
struct filsys fs;
char pad2[BSIZE];
} fsun;
#define filsys fsun.fs
There is older code in 3BSD, but it doesn't contain any references to sblock. Even the 4.0BSD version of the file differs significantly from current versions; it seems that the code evolved considerably between 4.0 and 4.2BSD (1 August 1983), the first version to include the name sblock:
union {
struct fs fs;
char pad[MAXBSIZE];
} fsun;
#define sblock fsun.fs
So much for history. How have the FreeBSD and NetBSD code bases evolved since then?
=== grog@eureka (/dev/pts/4) /src/FreeBSD/git/main/usr.sbin/makefs/ffs 91 -> diff -wu mkfs.c /src/NetBSD/CURRENT/src/usr.sbin/makefs/ffs/mkfs.c
--- mkfs.c 2023-09-24 09:56:00.612723000 +1000
+++ /src/NetBSD/CURRENT/src/usr.sbin/makefs/ffs/mkfs.c 2023-01-08 06:41:30.000000000 +1100
@@ -1,8 +1,6 @@
-/* $NetBSD: mkfs.c,v 1.22 2011/10/09 22:30:13 christos Exp $ */
+/* $NetBSD: mkfs.c,v 1.42 2023/01/07 19:41:30 chs Exp $ */
12 years (today!) and 20 updates! Interestingly, the latest commit starts:
evision 1.42
date: 2023-01-07 19:41:30 +0000; author: chs; state: Exp; lines: +11 -9; commitid: YhUB07cixFcG6E8E;
ufs: fixed signed/unsigned bugs affecting large file systems
Apply these commits from FreeBSD:
commit e870d1e6f97cc73308c11c40684b775bcfa906a2
Author: Kirk McKusick <mckusick@FreeBSD.org>
Date: Wed Feb 10 20:10:35 2010 +0000
This fix corrects a problem in the file system that treats large
inode numbers as negative rather than unsigned. For a default
(16K block) file system, this bug began to show up at a file system
size above about 16Tb.
That's doubly interesting: first, it was only a few months ago, and it was from FreeBSD—13 years ago! It would be nice to investigate further, but this horrible git makes it so difficult.
Still, there's an issue here. Clearly the NetBSD version of makefs has been maintained much more than the FreeBSD version. And I've already established that there are other issues (for example, in my test case the FreeBSD version only creates one cylinder group). Why not just port the latest NetBSD version?
It's not that simple. First, it would require much testing, and secondly there are things that we don't want to copy. The diff above goes on:
- sblock.fs_nindir = sblock.fs_bsize / sizeof(ufs1_daddr_t);
+ sblock.fs_nindir = sblock.fs_bsize / sizeof(int32_t);
...
- acg.cg_initediblk = MIN(sblock.fs_ipg, 2 * INOPB(&sblock));
+ acg.cg_initediblk = sblock.fs_ipg < 2 * FFS_INOPB(&sblock) ?
+ sblock.fs_ipg : 2 * FFS_INOPB(&sblock);
...
- if (acg.cg_nextfreeoff > (uint32_t)sblock.fs_cgsize) {
+ if (acg.cg_nextfreeoff > (unsigned)sblock.fs_cgsize) {
The lines starting with - are from FreeBSD, and the lines starting with + are from NetBSD. In each case, the FreeBSD version seems cleaner. There are also some local name changes which could be confusing. So it looks as if it'll be a line-for-line comparison. This thing is growing to be bigger than I expected.
|
Backup times
|
Topic: technology, general, health | Link here |
Chris Bahlo is back from Europe, and she is feeling somewhat under the weather. COVID-19? It's worth a check, and since we have a lot of RATs, she came by to pick one up.
A good time for our weekly backup change. Quick backup of my photos, which normally takes 30 minutes.
Tue 10 Oct 2023 11:42:02 AEDT Photo backup started
Tue 10 Oct 2023 11:52:26 AEDT Photo backup ended
Huh? That's not normal. A quick comparison showed:
sent 25,284,870,629 bytes received 34,427 bytes 13,981,147.39 bytes/sec
Sun 9 Jul 2023 05:05:00 AEST Photo backup started
Sun 9 Jul 2023 05:38:28 AEST Photo backup ended
sent 14,958,992,459 bytes received 18,774 bytes 9,942,845.62 bytes/sec
Tue 22 Aug 2023 05:05:00 AEST Photo backup started
Tue 22 Aug 2023 05:33:27 AEST Photo backup ended
sent 4,151,516,529 bytes received 5,909 bytes 3,728,354.23 bytes/sec
Tue 10 Oct 2023 05:05:00 AEDT Photo backup started
Tue 10 Oct 2023 05:24:19 AEDT Photo backup ended
sent 494,704,942 bytes received 1,990 bytes 853,678.92 bytes/sec
Tue 10 Oct 2023 11:42:02 AEDT Photo backup started
Tue 10 Oct 2023 11:52:26 AEDT Photo backup ended
So it seems to be very dependent on the amount written.
Chris later reported the result of the RAT: negative.
| Wednesday, 11 October 2023 | Dereel | Images for 11 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
No TV today
|
Topic: technology, multimedia | Link here |
One of the first things I do in the morning is update my German TV programme list from mediathekview.de, part of which is a cron job that downloads the latest information from verteiler1.mediathekview.de. But today it didn't get loaded: verteiler1.mediathekview.de wasn't found.
But why verteiler1 in the first place? It used to be verteiler.mediathekview.de, but that no longer exists. When the name finally resolved, I checked:
=== grog@eureka (/dev/pts/4) ~ 101 -> nslookup verteiler1.mediathekview.de
verteiler1.mediathekview.de canonical name = liste.mediathekview.de.
Name: liste.mediathekview.de
Address: 88.99.10.178
Name: liste.mediathekview.de
Address: 88.99.10.179
OK, that's good to know. And it makes sense up to a point.
While looking for alternatives, came up with this page. It doesn't seem to offer much.
|
More lawn mowing fun
|
Topic: gardening, general | Link here |
Paul Donaghy along today to mow the lawn. He got about 100 m before the lawn mower gave up on him. We half expected this, since something similar happened last time. But this time he investigated all normal possibilities and came up with a blank. There's a fuel filter between the tank and carburettor:
|
|||||||||||||||||||||||
It passes petrol, so it's unlikely to be a blockage. But somehow this carburettor doesn't look normal. What's this electrical connection at the left?
|
|||||||||||||||||||||||
Could it be that there's some issue that goes beyond my 50-year-old knowledge of internal combustion engines?
|
Bushfire!
|
Topic: general, technology, opinion | Link here |
Yet another bushfire warning from the Ballarat Courier today: bushfire south of Ballarat.
Where was it? That's what the VicEmergency app on my phone should tell me. But it didn't.
We've seen this before. OK, off to the web to see what it had to say. Berringa! That's in my watch zone, and that's what the app told me when I asked.
So why isn't this HORRIBLE VicEmergency app informing me? No notification (though I get them for things like bad weather off the US coast. Spent some time looking for ways to configure the app, but came up with nothing useful.
| Thursday, 12 October 2023 | Dereel | Images for 12 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More disk errors?
|
Topic: technology | Link here |
Checking my overnight logs today, saw this:
Oct 11 21:11:19 eureka kernel: (da2:umass-sim2:2:0:0): WRITE(16). CDB: 8a 00 00 00 00 01 2f 4a 5e 68 00 00 00 80 00 00
Oct 11 21:11:19 eureka kernel: (da2:umass-sim2:2:0:0): CAM status: CCB request completed with an error
Oct 11 21:11:19 eureka kernel: (da2:umass-sim2:2:0:0): Retrying command
This looks like a recovered write failure. But it's still a failure, something you don't want on a backup disk.
But is it a disk failure? This is USB, and in the past it has been horribly unreliable. In recent times it has been good by comparison. And the disk is connected to one of these el-cheapo USB extenders, so el-cheapo that I got a refund for them, but even the seller didn't want them back. So I've changed it so that the disk is directly connected to the motherboard. I have no idea what that will mean for the device numbering on the next reboot.
|
The Palestinians strike back
|
Topic: politics, history, opinion | Link here |
I've been watching with horror the atrocities that have been taking place in Israel and Palestine. Not just since Hamas attacked Israel: this has been going on since before I was born.
What's almost as horrifying is the one-sided reaction from most of the Western world. Yes, the indiscriminate massacres are horrendous. But the people who get up on their hind legs and condemn them are surprisingly quiet about previous history. What has happened in the 75 years since the foundation of Israel? Israel has become more and more fascist in its treatment of a conquered people.
They're certainly not weakening now. A complete blockade of Gaza? I thought that that had been in place for years, but now there's no power, no water, no gas (or is that petrol?).
But this is just the culmination of over 75 years of crisis in the region. Clearly it's not getting better, and looking at how the Israeli government is handling its military superiority, disregarding international law and turning whole countries into prison camps, it's not surprising that they're revolting.
Spent some time reading up on the prehistory, from the Balfour Declaration to the Peel Commission. Multiple videos, documents, and all seem to be different. At the start I had some ideas about how to address the problem, but the more I read, the less hope I have. About the only thing that becomes clear is that I was wrong in my assessment that Israel has become more violent in the course of the years: they were like that even before independence. Hamas may be violent, but are they worse than the Israelis?
|
Terrorists kill 107 civilians near Jerusalem
|
Topic: politics, history, opinion | Link here |
One of the atrocities that wasn't reported over the last few days was far closer to central Israel than the others: only 5 km from Jerusalem, terrorists killed 107 civilians.
Why wasn't it mentioned? Yes, it happened 75 years ago, but massacres are massacres, and terrorists are terrorists.
Or are they? The terrorists in question were Jewish, from the Irgun and Lehi groups, assisted by the Haganah and now part of the Israel Defense Forces. And in their case it was only close to the beginning of an ethnic cleansing that saw over 700,000 people driven from their homes. And that's the situation we still have today, 75 years later, where the majority of Palestinians are refugees.
The Israel problem is doubtless complex, but there was no excuse for these actions, and Israel is still living with them today. Will they ever learn?
|
VicEmergency again
|
Topic: technology, general, opinion | Link here |
Why can't I get alerts from VicEmergency app on my phone? Spent a lot of time looking through their excuse for documentation, coming up with at least the fact that they have a “test notification”. So I tried that. Check notifications, nothing there. But they also have a “send email” function. Set that up. Test. Yes! That's so much better than a mobile phone anyway.
More searching. There are so many notifications that I can hardly see anything. I get a notification every time I wake distress.lemis.com, which connects to the phone. Newspapers, notably the Washington Post, tell me of all sorts of things of only marginal interest. Google does the same. I found a way to tone down the WaPo, but I still don't know how to do it with Google.
And then, on the fifth page of notifications, I found the VicEmergency notifications, five
of them, all together under a notification issued 5 weeks ago. Isn't that clever? Clearly
I need to smash steal swipe them away as soon as I see them.
Will I ever come to terms with Android?
| Friday, 13 October 2023 | Dereel | Images for 13 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Goodbye George
|
Topic: animals | Link here |
It has been over a year since Peggy Naumov and her Borzoi George visited us. Today Yvonne received notification: some while back George woke Peggy up in the middle of the night, whining. She went to him, he put his head on her knee and died. Peggy was so devastated that she couldn't bring herself to talk about it until now.
What was the cause? My best bet is heart failure. Somehow it's particularly sad.
|
The Palestinian Untermensch
|
Topic: politics, opinion | Link here |
The situation in Gaza has provoked even worse reactions from Israel than I had feared, and I spent most of the day following developments. Somehow the Israelis still have the same mentality that they had in the 1940s. When asked about civilian impact in a recent interview, Naftali Bennett said, forcefully:
“Are you seriously asking me about Palestinian civilians? What is wrong with you? Have you not seen what’s happened? We’re fighting Nazis.
There's that “Nazi” word again. What aspect of Naziism do the oppressed people of Gaza fulfil? But there's another term that describes Bennett's opinion of such people: Untermenschen.
That's somehow reflected in Israel's “recommendation” to the inhabitants of the northern Gaza strip to move south within the next 24 hours. That's over a million people! And they have no water and no transport! For once the United Nations agreed. According to this article, António Guterres said “Moving more than one million people across a densely populated war zone to a place with no food, water, or accommodation, when the entire territory is under siege, is extremely dangerous - and in some cases, simply not possible”.
The reaction, from Israel's U.N. Ambassador Gilad Erdan: he claimed that Israel should be praised for its advance warning to Gaza residents. The United Nations and the Security Council were "facing one of their most pivotal moments," adding: "Will they remain true to their founding values? Or will they empower genocidal terrorists? This shouldn't be a question."
No, that shouldn't be a question. Crimes against humanity have no place in modern conflicts.
He's not alone. In another article I read of Israel Katz, the Israeli energy minister:
“What hypocrisy!” he said in a translation of a post on X, the site formerly known as Twitter. “We will not provide an ounce of water and electricity to those who do not evacuate. We will work with all our might to eliminate the Hamas-Daesh murders and restore security. Whatever was will not be.”
Are the Israelis all blind? I'm reminded of the issue where the children of abusive parents become abusive themselves. And when will the USA finally stop giving them a carte blanche?
While following up these developments, came across this article from The Atlantic. It contains, addressing the Israelis:
Don’t poison your national unity, as Bush did ours, by using the crisis for partisan advantage; Israel’s new unity government is a good sign. Don’t squander your moment of global legitimacy, or assume that the world’s support will last a day longer if news emerges of mass civilian deaths in Gaza, or believe that its loss wouldn’t matter. It matters that democratic countries, which have criticized the Jewish state but know the difference between Israel and Hamas, are now expressing outrage, just as the same countries’ support mattered when Ukraine was brutally invaded by Russia. “It was very important not to be alone,” President Volodymyr Zelensky said yesterday, extending his solidarity to Israel. This is more than the Netanyahu government, which has been carefully neutral on Russian aggression, deserves from Ukraine. Vladimir Putin is holding his cards close on Israel and Hamas. Zelensky understood, as Netanyahu didn’t, that Russia, Iran, and Hamas will land on one side, and Ukraine and Israel on the other.
America should have its friend Israel’s back while conveying unpleasant truths to its face. After Saturday it’s clear that two things, apparent contradictions, have to be accepted at the same time: A group that seeks Israel’s destruction must be destroyed, and Israel’s cruel treatment of Palestinians in the occupied territories only helps that group’s cause. It’s impossible for Israel to live in peace alongside Palestinians who will never accept its right to exist, and it’s impossible for Palestinians to accept a fate of permanent subordination. To address these together will require profound change from both sides. It’s beyond the ability and will of the current Israeli government, and on the West Bank, a sclerotic Palestinian government, weakened by its own corruption and by continued Israeli domination, is just as useless. Perhaps, out of this horror, better alternatives will emerge.
| Saturday, 14 October 2023 | Dereel | Images for 14 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More garden work
|
Topic: gardening, general | Link here |
Jesse Walsh along again today to continue working on the garden. He made quite good progress, and in another couple of weeks the garden should be looking almost presentable. Sadly, though, it looks like the Alyogyne huegelii hasn't survived being blown over:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Only one (the smallest) of the three branches seems to have survived. Should I remove the rest, or just plant a new tree?
|
Login failures
|
Topic: technology, opinion | Link here |
What good is login protection in an unreachable network? I still use passwords, because it “feels right”, but I don't log in very often. And when I do, my password (a very unlikely character combination) is more finger memory than a recollection of the character sequence.
But then I had repeated root login failures. Why? Which words capitalized? Which special characters? I couldn't work it out, and in the end I had to reset the password (from an open xterm) to what I thought it had been. But somehow I had the feeling that something was wrong. Had somebody changed it? How could they access it?
But then there was distress, running Microsoft “Windows”. I connect to it with rdesktop, and the invocation includes user name and password. But recently it has prompted me for the password anyway, though it hasn't changed. Is this some new Microsoft provocation?
|
No to racial discrimination
|
Topic: politics, history, general, opinion | Link here |
Reading my diary of 10 years ago today I was reminded of a very disturbing court decision in Malaysia: it prohibited non-Muslims (in other words, non-Malays) from using the word “Allah” to refer to God.
That's horrible! I don't know if they have repealed it yet, but that's so discriminatory. And since I have been following the history of Palestine, I was aware that Mandatory Palestine had a significant Christian Arab population, round 12%. Their word for God? “Allah”.
Today the Australian Government held an Indigenous Voice referendum. What's that? How can a voice be indigenous? Ah, no, that's not what it means. The text is:
A Proposed Law: To alter the Constitution to recognise the First Peoples of Australia by establishing an Aboriginal and Torres Strait Islander Voice.
That's so vague that it should be rejected just because of that. The intention is good: to lessen the social gap between Aboriginal Australians and Torres Strait Islanders on the one hand and everybody else on the other hand.
But the approach is broken for so many reasons. They're singling out two ethnic groups who are only related to each other by having been here before the British. Why did they leave out the Dutch? And as far as I can tell, nothing mentions those many people who are part one, part the other. And what is a “Voice”? As I understand it, it's a consultative body for matters which pertain to these two ethnic groups. But I can't find a clear definition. The whole concept is so vague that I'm sure it could be open to abuse in the future.
But there's a much better reason to reject it: it's racial discrimination, and that leads to conflicts; Israel is an obvious example, but Malaysia is much worse. The fact that it was suggested with the best of intentions makes no difference to that. As soon as I heard of it, I knew that my answer had to be “no”. There are so many other ways to address the issues without adding this kind of vagueness and complexity.
Off to the Dereel Hall this afternoon to vote. I must have been one of the last: the whole time I was there, nobody else showed up to vote. And the person who looked up my name in the list of voters was so confused by my name that she couldn't find it in the list; I had to help her.
The final results aren't due for 2 weeks, to give time for postal votes to come in. But the answer was clear only minutes after polls closed at 18:00: “No”. Now hopefully the government will come up with better ways to address the issues.
|
Palestinian religious distribution: ChatGPT speaks
|
Topic: technology, history, general, opinion | Link here |
What was the religious distribution of Mandatory Palestine in 1920? Above I wrote that 12% were Christians, and I still think that that's what I read. But there are experts. Bard didn't want to know, but ChatGPT tells me:
In 1920, Mandatory Palestine had a religious distribution that consisted of a majority Muslim population, followed by a significant Jewish minority and smaller Christian and Druze communities. The exact percentages were approximately 78% Muslim, 11% Jewish, 9% Christian, and 2% Druze.
That's close enough, though the 11% Jewish is more than I recall. It probably greatly depends on the exact date and who did the counting.
| Sunday, 15 October 2023 | Dereel | Images for 15 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Why are you pinging me?
|
Topic: technology, opinion | Link here |
For some reason I have a whole lot of ICMP traffic, apparently random scans, resulting in this kind of output:
=== grog@eureka (/dev/pts/4) ~ 120 -> arp -a | grep incomplete
hisen.lemis.com (192.109.197.237) at (incomplete) on em0 expired [ethernet]
taskumatti.lemis.com (192.109.197.224) at (incomplete) on em0 expired [ethernet]
? (192.109.197.196) at (incomplete) on em0 expired [ethernet]
monorchid.lemis.com (192.109.197.133) at (incomplete) on em0 expired [ethernet]
dhcp-249.lemis.com (192.109.197.249) at (incomplete) on em0 expired [ethernet]
brewer.lemis.com (192.109.197.147) at (incomplete) on em0 expired [ethernet]
Why? Is it a preliminary to an attack? In any case, it annoys me, and occasionally I get log messages telling me that more than 200 have arrived in a second:
Oct 13 15:52:21 eureka kernel: Limiting closed port RST response from 728 to 200 packets/sec
Oct 13 15:53:14 eureka kernel: Limiting closed port RST response from 780 to 200 packets/sec
Oct 13 15:57:32 eureka kernel: Limiting closed port RST response from 731 to 200 packets/sec
Over 700 packets a second! How can I limit it further? Off looking at firewall rules, but somehow I couldn't find anything useful.
But wait a while... There is a limit, as the messages above show. How do I set it? More searching—yesterday was the 22nd birthday of my bug report 44034, looking for documentation of sysctls. It's still open!
More searching, and finally found the sysctl: net.inet.icmp.icmplim. And now there's documentation, if that's the word:
=== root@eureka (/dev/pts/8) ~ 62 -> sysctl -d net.inet.icmp.icmplim
net.inet.icmp.icmplim: Maximum number of ICMP responses per second=== root@eureka (/dev/pts/9) ~ 62 -> sysctl net.inet.icmp.icmplim=5
net.inet.icmp.icmplim: 200 -> 5
And then things worked as expected:
Oct 15 14:14:02 eureka kernel: Limiting icmp ping response from 11 to 5 packets/sec
Oct 15 14:14:05 eureka kernel: Limiting icmp ping response from 8 to 5 packets/sec
Oct 15 14:14:09 eureka kernel: Limiting icmp ping response from 13 to 5 packets/sec
Oct 15 14:14:12 eureka kernel: Limiting icmp ping response from 13 to 5 packets/sec
Oct 15 14:14:13 eureka kernel: Limiting icmp ping response from 9 to 5 packets/sec
Oct 15 14:14:22 eureka kernel: Limiting icmp ping response from 44 to 5 packets/sec
Oct 15 14:14:23 eureka kernel: Limiting icmp ping response from 90 to 5 packets/sec
Oct 15 14:14:24 eureka kernel: Limiting icmp ping response from 95 to 5 packets/sec
The only issue now is to apply this to all systems. And of course things like Microsoft fall into the “too hard” basket.
|
Israel and Gaza: Coming to their senses?
|
Topic: politics, general, opinion | Link here |
Following the Israel–Hamas war is depressing. It's clear that Israel is overreacting, committing multiple war crimes in the process. But the USA is with them all the way! That's no way to calm things down.
Gradually, though, some people are coming to their senses. The obvious thing is: what happens next? Over the last week I've been trying to get my head around that. One thing seems clear: Israel must leave the occupied territories (including the illegal settlements, but that will have to come later). But Palestine is in no position to rule itself. Somebody more neutral would have to take over. I was tending towards the idea of a combined US-Saudi force, but looking at the US track record on this sort of thing, it doesn't seem to be a recipe for success.
But then I read this article. Very much what I was thinking of, just without the USA. Just Saudi Arabia keeping the peace. That most peaceful of people, Mohammed bin Salman, who had Jamal Khashoggi murdered and dismembered five years ago.
But people change, and if MBS wants to rehabilitate himself, arranging for peace in Palestine would be an excellent way to do it. Could it work?
|
Greg Lehey reconsidered
|
Topic: technology, general, opinion | Link here |
Last month I was surprised when ChatGPT came up with a surprisingly accurate description of me. For some reason I looked again today. I have changed!
Hi! Greg Lehey is a computer scientist and software engineer. He is known for his work in the field of FreeBSD, an open-source operating system. Lehey has made significant contributions to the FreeBSD project, including writing the book "The Complete FreeBSD" and serving as the project's release engineer for several years. He has also been involved in various other open-source projects and has a strong background in Unix-like systems.
Where did that come from? I've never been involved in the FreeBSD release process. Also my membership of the Core Team has disappeared. And the whole thing is shorter than it was last month.
|
Tagine again
|
Topic: food and drink, opinion | Link here |
Somehow our food planning isn't what it used to be. We had decided to eat a tagine today, and Yvonne had bought courgettes and aubergines. But what else do we need? We had some relatively tough beef in the deep freezer, so I chose that. But though there are 7 tagine recipes in my recipe collection, we didn't have all the ingredients for any of them. No apricots. No almonds. No ras el hanout. Not enough chick peas.
So I faked a fake tagine. Instead of apricots I used prunes. I faked a ras-el-hanout with 10 g coriander, 5 g cinnamon, a couple of cloves and some cardamom seeds—something that could equally well have gone into a curry. I was going to use flaked almonds instead of whole almonds, but in the end I forgot. And somehow the recipe seemed too primitive to waste any saffron on.
The result: OK, but no more. The lack of saffron was evident. But it gave me time to think about times and proportions. I ended up cooking the meat for a good 5 hours, and it wasn't disintegrating. And for a general rule, the same weight of meat and vegetables seems a good choice. The potatoes and carrots got 2 hours cooking and weren't overly soft, and the aubergines and courgettes did well with 45 minutes, also much more than in the recipes. And some sweetness is right; the 80 g of honey seemed to be the right amount.
| Monday, 16 October 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Monitor dropout again
|
Topic: technology, opinion | Link here |
Yet another short dropout on monitor 2 this morning, this time earlier than on other days. What is it that makes it happen only once a day, and that relatively early?
|
ICMP limits: shut up!
|
Topic: technology, opinion | Link here |
Yesterday's rate limiting for ICMP packets works. And it has reduced the number of packets by over 95%. But why does the system have to crow about it so much? /var/log/messages is full of messages like
Oct 16 16:07:10 eureka kernel: Limiting icmp ping response from 10 to 5 packets/sec
Oct 16 16:07:12 eureka kernel: Limiting icmp ping response from 6 to 5 packets/sec
Oct 16 16:07:13 eureka kernel: Limiting icmp ping response from 7 to 5 packets/sec
Time to find a way to mute this message. But it's interesting to see who is doing the pinging. Nearly all the packets I saw were like these:
11:45:09.560748 IP kimchi.lemis.com > 603203865.sof.cdn77.com: ICMP echo reply, id 47690, seq 260, length 64
11:45:09.787330 IP kimchi.lemis.com > 925802276.buc.cdn77.com: ICMP echo reply, id 28932, seq 74, length 64
11:45:10.852431 IP kimchi.lemis.com > ec2-3-28-123-39.me-central-1.compute.amazonaws.com: ICMP echo reply, id 2, seq 3054, length 16
11:45:10.862328 IP kimchi.lemis.com > ec2-3-29-130-161.me-central-1.compute.amazonaws.com: ICMP echo reply, id 2, seq 3054, length 16
11:45:11.043500 IP kimchi.lemis.com > ec2-3-28-40-215.me-central-1.compute.amazonaws.com: ICMP echo reply, id 2, seq 3054, length 16
11:45:11.179465 IP kimchi.lemis.com > ec2-157-175-151-52.me-south-1.compute.amazonaws.com: ICMP echo reply, id 6, seq 6161, length 16
kimchi runs NetBSD, and I haven't found a way to rate limit that yet. But are do so many systems involved?
|
New computer on its way
|
Topic: technology | Link here |
Call from Seth at Leading Edge Computers this afternoon. My new computer, the first for nearly 10 years, is ready for pickup.
Joy? No, not really. Now there's a lot of work to do to get the thing working the way I want it. And that will be fun, at least partially because I've decided to once again run two machines: eureka will stay roughly as it is and be the gateway machine, and the new machine will get the name hydra from the HP Z800 that I inherited from Bruce Evans last year. I had chosen the name because I thought it had 24 CPUs, but that proved incorrect: there are only 12. But the new machine has 32 CPUs, so that should fit well. I'll then use it as my main machine, and that raises questions about how to split the disk storage.
Today I spent some time preparing more mundane things: how do I set up X while maintaining the status quo on eureka? That requires 8 monitors, and I only have 7:
|
|||||||||||||||||||||||
But I can use monitor 0 on eureka as monitor 3 on hydra until things are up and running. And with the laptop on the right I could join them all together with x2x to get a display with a total of 17,406,976 pixels, only marginally more than the 17,382,400 pixels I had 17 years ago.
| Tuesday, 17 October 2023 | Dereel → Ballarat → Dereel | Images for 17 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Paying for hydra
|
Topic: technology, general, opinion | Link here |
In preparation to pick up my new machine this afternoon, I checked my bank accounts. Yes, I had enough in my ANZ account, but only just. What about Bank of Melbourne? “Please match the requested format”!
That's been going on for years! When are they finally going to fix it? Spent some time messing around, noting that it happened on all browser/computer combinations I had. Called up the help line, but then decided that I couldn't face the annoyance of talking to somebody whose brain is stored in a script. Finally, somehow, I managed to access the system, after they first tried to send me a “secure code” to my phone, after which I was given the opportunity to choose to send to my mobile phone instead. Another 30 minutes waste and lots of annoyance.
|
More amateur scammers
|
Topic: technology, general, opinion | Link here |
Seen in mail today:
Subject: Australian Taxation Office: Refund Notification
That's stupid enough that it could almost be legitimate. What's in the message?
From MGstatter@t-online.de Tue Oct 17 08:06:11 2023
...
Received: from mailout02.t-online.de (mailout02.t-online.de [194.25.134.17])
by lax.lemis.com (Postfix) with ESMTP id 64A252811D
Subject: Australian Taxation Office: Refund Notification
Dear Taxpayer,
After the last annual tax calculation of your fiscal activity, we have
discovered that you are eligible to receive a tax refund balance of
3,850.50 AUD. Kindly complete the tax refund request and allow 5-10
working days to process it.
Get Started <https://freezing-zesty-nannyberry.glitch.me/at0hs7hs.html>
I suppose that's enough to confuse a number of people. But part of the issue is that the interface software is so hard to understand.
|
The new hydra
|
Topic: technology, opinion | Link here |
Into town in the afternoon to pick up the new machine. Everything went smoothly, though it's clear that times have changed. Detailed manuals for the coolers, just a wordless sheet of paper with diagrams of how to put things onto the motherboard. A couple of CDs, but nothing relating to the motherboard. To make up for it, there were lots of silly stickers:
|
|||||||||||||||||||||||
It's interesting to note that some of the images bear a marginal resemblance to Beastie:
|
|||||||||||||||||||||||
Back home, spent some time looking for documentation, and finally found the manual. The motherboard has an amazing number of interfaces, and they're somehow all different. Enough for today; I'll continue tomorrow.
|
distress wants attention too
|
Topic: technology, opinion | Link here |
Recently I've had various issues with distress.lemis.com, my Microsoft box. For a while it wanted me to enter a password every time I connected, but that went away. Today, though, things hung, notably DxO PhotoLab. OK, Microsoft solution: reboot.
After 5 minutes it still hadn't rebooted. Time for the Big Red Button. And yes, after that it rebooted, still with the old windows open. Stop DxO and... it rebooted again, and then performed updates that I didn't ask for!
How I hate Microsoft!
| Wednesday, 18 October 2023 | Dereel | Images for 18 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Looking at hydra
|
Topic: technology, opinion | Link here |
Before setting up the new iteration of hydra.lemis.com, it's interesting to take a detailed look inside. Things have changed in the last 10 years. Between the first IBM Personal Computer 42 years ago and when I built eureka, changes seemed to be relatively small. CPU on a motherboard, slots for expansion cards, power supply at one side. About the biggest difference was that the original PC was oriented horizontally, while most of my machines in the last 30 years have been oriented vertically (“tower” housing).
But hydra is different. Open the case and it's mainly fans:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
I identified the RAM at the right, noticeable because of its violent nature:
|
|||||||||||||||||||||||
But where's the SSD? Based on the motherboard specs, it should be nearby, but I haven't been able to find it.
Another thing of interest is that there's more on the other side, “under” the motherboard:
|
|||||||||||||||||||||||
About the only thing there that I recognize (because Seth pointed them out) are two slots for SATA SSDs.
The back of the machine is about the most recognizable:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
But the devil's in the detail, and the detail's in the motherboard manual that I had to go searching for, though the markings give some clues. Not all the USB slots are the same: speeds differ, and so do the connectors.
And the front? Nothing at all!
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Well, the DVD drive, of course, but that has to be there. The normal buttons you find on the front of a case are on top, where dirt could fall in if they weren't plugged:
|
|||||||||||||||||||||||
The case is quite well made, and the covers come off easily. Putting them back is more of a challenge: there are no rails. You have to align the two screws in cutouts at the front (here rotated 90° to save space):
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
It's not really difficult, just fiddly, but it detracts from the quality of the case.
|
Setting up hydra
|
Topic: technology, opinion | Link here |
I had already put monitors in place to set up X, but that's jumping the gun. First we need an operating system. In principle I have that, from my work in March. So what do I do? Clearly I need to copy the system image, but first, what's already on the disk? Boot a live system from a FreeBSD 13.2-RELEASE memory stick and go looking for the disks. gpart show should tell me what's there. No geoms found! After some searching, discovered that the M.2 disks have names like /dev/nvd0, and sure enough, there's one, but no subdisks. Is it completely empty? What does hexdump say? Nothing: it's not on the memory stick. OK, mount a file system on eureka and copy the first couple of megabytes there. Yes, hexdump on eureka confirms, disk completely empty, all binary zeroes.
OK, the easiest way to partition it is with an installation. 50 GB root file system, no swap and the rest a /home file system. And that went smoothly enough, though I didn't need the installation at all.
Next, copy the image, currently on eureka and unused: it's only there for the swap partition (the only SSD I have on my FreeBSD systems). But how do I copy it? I should have installed the temporary 13.2 system on the second partition so that I could use it to copy the real image to the final root file system.
I can still do that. Boot the memory stick again. Select the boot menu. It doesn't make any sense at all, and it only shows the current boot sequence, without offering any obvious way of changing it. The only other option was to enter the BIOS setup screen, which is garish and equally confusing:
|
|||||||||||||||||||||||
First, how do I navigate it? The arrow keys provide some function, but not what I would expect. I can go up and down from the current position, CPU, but not sideways. Ah, go up to GAME BOOST and then somehow to the right to Boot Priority.
|
|||||||||||||||||||||||
And it highlights the individual meaningless images with almost as meaningless names, all starting with “UEFI”. Clearly one was the memory stick. But I couldn't find a way to move it ahead of the disk.
Off searching on the web. Yes, of course there are pages explaining what to do, notably from the maker. But though I selected “MSI CLICK BIOS 5” (as at the top of the screen), what I found looked very different:
|
|||||||||||||||||||||||
And the instructions are so vague that they're almost useless. In particular, they don't tell you how to navigate the screen!
More searching and came across this video, conveniently in Tamil:
It uses a mouse! OK, I can do that. Plug in a mouse. Nothing happens. Reset and reenter the setup screen. Yes, it works, and I can select the devices:
|
|||||||||||||||||||||||
OK, save:
|
|||||||||||||||||||||||
What's this Boot Option stuff? How do I set that? Ah, they're too polite that to assume that I don't know. Clearly it's intuitive.
But there are other pages, like this one. It tells me to go looking for the version for my motherboard on the MSI home page! And for some reason, that page hung!
Somehow I managed to find a way to boot from the memory stick; it's not clear how, but it's repeatable. If there's a bootable USB stick, boot from it, otherwise from SSD. Set up the installation to install on /dev/nvd0p3, and reboot.
And of course it came up with the system on /dev/nvd0p2: I had forgotten to wipe it. So, all over again. This time I didn't bother with a new installation. Just newfs /dev/nvd0p2 and copy the file system on eureka, incongruously mounted on /photowork. But I couldn't mount it. Why not? It was in /etc/exports. I could mount other file systems, so I tarred it to /home and untarred it from there. And I was then able to boot it with no further problems.
Looking at the running system was a certain feeling of satisfaction:
FreeBSD 13.1-STABLE #0: Mon Jan 9 12:57:21 AEDT 2023
root@hydra:/usr/obj/eureka/home/src/FreeBSD/git/stable-13/amd64.amd64/sys/GENERIC amd64
FreeBSD clang version 14.0.5 (https://github.com/llvm/llvm-project.git llvmorg-14.0.5-0-gc12386ae247c)
VT(efifb): resolution 1024x768
CPU: AMD Ryzen 9 7950X 16-Core Processor (4500.10-MHz K8-class CPU)
Origin="AuthenticAMD" Id=0xa60f12 Family=0x19 Model=0x61 Stepping=2
...
real memory = 206561083392 (196992 MB)
avail memory = 199861567488 (190602 MB)
206,561,083,392 bytes of memory! That's 806,879,232 times as much as on my first computer in 1978. And top shows:
Mem: 120M Active, 2176K Inact, 501M Wired, 37M Buf, 186G Free
OK, next to bring the system up to date. make -j 40 buildworld kernel. Somehow it's satisfying to see this kind of output from top:
last pid: 69482; load averages: 23.76, 8.11, 3.16 up 0+00:08:26 13:51:46
138 processes: 40 running, 98 sleeping
CPU: 97.3% user, 0.0% nice, 2.4% system, 0.3% interrupt, 0.0% idle
Mem: 4209M Active, 1618M Inact, 2333M Wired, 1574M Buf, 178G Free
PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
69286 root 1 90 0 364M 298M CPU26 26 0:09 100.00% c++
69314 root 1 87 0 542M 420M CPU19 19 0:07 99.98% c++
69320 root 1 87 0 469M 368M CPU25 25 0:07 99.97% c++
69336 root 1 87 0 533M 422M CPU9 9 0:07 99.79% c++
69220 root 1 93 0 412M 333M CPU30 30 0:12 99.63% c++
69258 root 1 91 0 363M 298M CPU4 4 0:10 99.58% c++
That clearly excited the processors, and the fans ran up to a dull roar, though not that much more than some laptops make. It was over almost before I noticed:
>>> World build started on Wed Oct 18 13:49:32 AEDT 2023
>>> World build completed on Wed Oct 18 13:59:00 AEDT 2023
>>> World built in 568 seconds, ncpu: 32, make -j40
...
>>> Kernel build for GENERIC started on Wed Oct 18 13:59:00 AEDT 2023
>>> Kernel(s) GENERIC built in 70 seconds, ncpu: 32, make -j40
...
>>> Installing kernel GENERIC on Wed Oct 18 14:00:10 AEDT 2023
>>> Installing kernel GENERIC completed on Wed Oct 18 14:00:13 AEDT 2023
Last time I built a new system was last month with dereel. The times:
>>> World build completed on Thu Sep 28 15:44:37 AEST 2023
>>> World built in 8728 seconds, ncpu: 4, make -j6
...
>>> Kernel build for GENERIC completed on Thu Sep 28 15:58:48 AEST 2023
>>> Kernel(s) GENERIC built in 851 seconds, ncpu: 4, make -j6
...
>>> Installing kernel GENERIC on Thu Sep 28 15:58:48 AEST 2023
>>> Installing kernel GENERIC completed on Thu Sep 28 15:59:04 AEST 2023
Years ago I recall people (Anton Blanchard?) having competitions to build Linux kernels in under 10 minutes. That was years ago, of course, but in general software, particularly compilers, evolve to take advantage of higher processor power. But one way or another, that's quite a pleasing result. buildworld 15 times as fast, kernel 12 times as fast.
I followed this up. Yes, it was Anton, but it wasn't a 10 minute kernel build, it was a 8 second kernel build. For 2002, that's not bad, but he had to hack things to do so, and of course, my comment about compilers stands.
Next, upgrade packages. Surprise, surprise, everything Just Worked. Well, almost everything. For some reason Emacs had an interface problem when displaying on an X server on eureka: all control character input got a c-s in front (for example, c-f became c-s-f. The problem was related to X, and didn't happen on an xterm. That makes it useless, of course, but there's every reason to believe that it's not a problem with hydra. Hopefully it relates only to eureka, so it will soon not be an issue.
And, apart from discussions on IRC, that was as much as I felt like doing today. The next step will be X, and I'm gathering courage.
| Thursday, 19 October 2023 | Dereel | Images for 19 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Bruno the destroyer
|
Topic: animals, general | Link here |
Cats love to scratch, and we have some wicker baskets for dirty linen. Over the years, many have decorated them. But Bruno is the most effective:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Finally it was too much, and we've replaced it with a plastic basket.
|
More hydra investigations
|
Topic: technology, opinion | Link here |
Spent much of the day documenting my fun with hydra. One thing that became evident was on the BIOS display screen:
|
|||||||||||||||||||||||
“DDR” (they mean memory, not Deutsche Demokratische Republik) speed 3600 MHz? I paid for 5600 MHz components. Is this a tuneable? How can I tell? Where's the BIOS manual? Followed up with the MSI web site, which insisted on changing www.msi.com to au.msi.com and then hanging. More reading the documentation, which contained a link to the BIOS: https://download.msi.com/archive/mnu_exe/mb/AMDAM5BIOS.pdf. Is that even valid? Yes!
It's still not good. No information about how to navigate the screen, but an important detail: the F7 link at the top of the page isn't just useful, it's essential. Only then do I get the displays that all the web pages have been talking about! And with a bit of effort I managed to set the memory speed to 5600 MHz. Reboot. Hang!
Dammit, have I bricked the configuration? Powered down, powered up, and it came up showing DDR speeds of 3600 and 5600 MHz, depending on where I looked. It also showed a processor speed of 5.5 GHz instead of the previous 4.5 GHz, and told me that I had set “Precision Boost Overdrive”, whatever that may mean. More looking in the BIOS manual and managed to turn it off. More delays, and then it booted normally.
So what speed is my memory running at? It looks as if further investigation is needed.
Also checked displaying an Emacs window on tiwi. It works! So it seems that whatever the issue I had yesterday, it's related to eureka. Hopefully that's the end of that issue.
|
Historical problems building new machines
|
Topic: technology, history, opinion | Link here |
I built eureka, at least the current iteration, nearly 10 years ago. And the one before that, by coincidence, exactly 15 years ago. And in each case I grumbled about the appalling documentation. Still, there's reason to believe that it's getting even worse.
|
Who bombed the Gaza hospital? Who cares?
|
Topic: politics, opinion | Link here |
A hospital in Gaza has been bombed, killing in the order of 500 people seeking refuge there. Many people, including Yvonne, assumed that it was the Israelis. And Israel denies responsibility, not for the first time: they're chronically irresponsible.
Still, there's a lot of confusion about the matter. The media can't even decide on the name of the hospital. Al-Ahli Arab Hospital? That's the one that Al Jazeera prefers. Al-Ahli Baptist Hospital? The Western news groups used that, though Wikipedia now redirects to “al-Ahli Arab Hospital”. So it's no surprise that nobody has so far been able to convincingly establish who did it. The Wikipedia page is continuously being updated.
Yes, there was serious loss of life, currently 471 deaths. But that's only a fraction of the 3,478 dead so far in the current 2023 Israel–Hamas war, and almost nothing in comparison with the 31,227 fatalities since 1947. And Israel can't deny responsibility there.
Still, it's particularly disappointing that Joe Biden expressed such unconditional support for the Israelis that “the other team” (his words) won't consider him an honest broker, especially since the USA vetoed a UN draft resolution aimed at lessening the suffering in Gaza because it didn't stress Israel's right to self-defence—and that after saying that the right to self-defence goes without saying. I fear that the USA will continue to be the Bad Guys for most of the Muslim world.
| Friday, 20 October 2023 | Dereel | Images for 20 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
hydra: progress?
|
Topic: technology, opinion | Link here |
For some reason I'm hesitating with setting up X on hydra. Part of it is simply getting all the hardware physically in place. There are no fewer than 14 cables and 24 connectors involved, and they need to be placed on a desk top that already has 9 computers on it. But I got started today, and it looks as if it will all fit roughly as planned.
The next step is configuration. The display card is an GeForce RTX™ 3060 GAMING OC 8G card, based on the NVIDIA GeForce RTX 3060. Where do I find the correct driver? By searching their driver list; they don't have a direct link. And then I have to configure the thing; I've already established that NVIDIA's nvidia-xconfig program is seriously wanting, maybe to the point that it's useless. But that might be the missing link that I need to get the thing working.
|
Help configuring X
|
Topic: technology, history, opinion | Link here |
By chance, saw this in my mail inbox today:
Setting up X11: A no-tears guide to XFree86 configuration.
Just what I want. Oh, wait. Something's not right. XFree86? That's not new. Wikipedia tells me that the last release was in December 2008, nearly 15 years ago. And the title of the article somehow sounds familiar. Of course: I wrote it, in July 1995, and the mail was just part of Academia's daily spam: “Claim ... for your profile”.
Reading it is interesting. It is almost only concerned with mode lines, something that has become completely irrelevant since the introduction of digital displays. Now it's painless unless you're stupid enough to want more than one screen. And then it's at least as bad as it has ever been. But my article is completely obsolete; it's only of historical interest.
|
More garden work
|
Topic: gardening | Link here |
Jesse Walsh along to do some more garden work today, somewhat hampered by mechanical problems: his whipper-snipper gave up on him.
|
Goodbye, Canon M
|
Topic: photography, opinion | Link here |
Round 10 years ago, the writing was on the wall for serious cameras: DSLRs were on their way out, but the only serious “mirrorless” products were from smaller makers. Canon and Nikon offered models that were more alibis than anything serious. The Canon EOS M series seemed better than the Nikon 1, but that wasn't saying much, and I was continually expecting them to discontinue the system, including the mount.
It took a long time, so long that I was beginning to doubt my estimate, but now it has happened. I suppose the best thing about it is that I was right after all; I feel a little sorry for people who have invested in EOS M glass.
| Saturday, 21 October 2023 | Dereel | Images for 21 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
An unbiased history of the Palestine conflict
|
Topic: politics, history | Link here |
The current situation in Gaza is capturing the world's attention, and just about everybody I see is pro-Israel and anti-Gaza and Palestinians in general. Few appear to understand this history. Where do you find an unbiased history? The United Nations have published one. It makes sense to me, and clearly it has been through much more scrutiny and fact checking than anything I could write. It also shows how many UN resolutions Israel has ignored.
|
tiwi panic
|
Topic: technology, opinion | Link here |
tiwi.lemis.com, my TV computer, has far too little memory. I should have replaced it with another machine months ago, but every time I've played around with it I have run into connector hell, which even resulted in me buying a new TV. So for the moment I put up with web browsers swapping for 10 or 20 seconds at a time.
But today was different. The whole display froze, though the mouse cursor showed that the system was still alive. Into the office and checked: firefox 100% CPU, X 100% CPU. kill -9 on the firefox seemed to do nothing, but in fact it only took about 30 seconds. And then another firefox process took its place. When they were all gone, a chrome came in, again 100%.
Finally it was done. top showed almost no CPU usage, also no X. Back to the lounge room to find tiwi booting; it must have just started. Why?
Oct 21 15:37:18 tiwi kernel: Fatal trap 12: page fault while in kernel mode
Oh. I wonder where that happened, especially since the issues seemed to have been over. The reboot process conveniently included a completely useless backtrace:
Oct 21 15:37:18 tiwi kernel: KDB: stack backtrace:
Oct 21 15:37:18 tiwi kernel: #0 0xffffffff80c50235 at kdb_backtrace+0x65
Oct 21 15:37:18 tiwi kernel: #1 0xffffffff80c02f01 at vpanic+0x151
Oct 21 15:37:18 tiwi kernel: #2 0xffffffff80c02da3 at panic+0x43
Oct 21 15:37:18 tiwi kernel: #3 0xffffffff8109fd57 at trap_fatal+0x387
Oct 21 15:37:18 tiwi kernel: #4 0xffffffff8109fdaf at trap_pfault+0x4f
Oct 21 15:37:18 tiwi kernel: #5 0xffffffff81077548 at calltrap+0x8
That shows how control passed to the panic() routines, and says nothing whatsoever about the cause of the trap. But then there's the information about the dump itself:
=== root@tiwi (/dev/pts/9) /var/crash 4 -> cat core.txt.2
Unable to find a kernel debugger.
Please install the devel/gdb port or gdb package.
Somehow this is more work than it should be.
| Sunday, 22 October 2023 | Dereel | Images for 22 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Earthquake!
|
Topic: general | Link here |
Lying in bed just after 2:00 this morning, I felt the earth tremble. Earthquake? Yes! Only a couple of seconds, and not very strong, but clearly it couldn't have been anything else.
When I got up, the news media confirmed it. Magnitude 4.8, north-west of Apollo Bay. No, magnitude 5.0, just south of Colac. No, Apollo Bay was right, but it was magnitude 4.7.
Which should I believe? Do I care? USGS isn't Australian, and they haven't changed their story, but do they care that much? Still, the media didn't give any good links to the other claims, so I'll take their version.
Interestingly, especially considering the time I spent in California, I have only ever experienced two earthquakes. The other was in November 1977, near Bendigo, not that far from here.
|
Bruno: no pellets?
|
Topic: animals | Link here |
For some time now we've been feeding Bruno 30 g of pellets in the morning, 60 g of minced beef and about 20 g of chicken neck in the evening.
But he's becoming fussy. Yesterday morning he didn't touch his pellets. That's not good, because they could contain trace foodstuffs that he needs. So, Bruno, eat your pellets.
He didn't. By this afternoon, he had eaten less than 10 g of yesterday morning's ration. From now on, he doesn't get anything else until he eats at least 10 g a day.
|
No wildflower show
|
Topic: gardening | Link here |
Today they had planned a wildflower show in the Brisbane Ranges, east of Steiglitz (pronounced „Stieglitz“). But the weather wasn't suitable, so we decided not to go, leaving me more time to look at hydra.
|
CAPTCHAs: How to insult people
|
Topic: technology, gardening, general, opinion | Link here |
Should I inform people that we weren't going to the wildflower show? Probably it was canceled anyway, but it can't do any harm to follow the convoluted “contact the organiser” on the web site, so I did that.
“Prove that you're human”. Another bloody CAPTCHA! And when I had done that, another. They seem to be getting more common lately. A few days ago, Bendigo Discount Pet Supplies asked for one, after I had entered entered my password. People, are you really trying to chase off your customers? I didn't complete it, but found that I was able to complete my purchase without logging in and without a CAPTCHA. What kind of brain-dead “webmaster” thought that one out?
|
More Microsoft distress
|
Topic: technology | Link here |
Another random strangeness from distress today:
|
|
|||||||||
That was on not one, but two windows after I tried to start rdesktop. Closed both of them and started again, normally.
Why is Microsoft so inscrutable? Arguably, of course, this is an rdesktop issue.
|
hydra: progress
|
Topic: technology, opinion | Link here |
Cancelling the excursion meant that I could finally start the dreaded task of setting up X on hydra.lemis.com.
The whole thing is complicated enough that I've started (another) web page describing what to do.
First, what driver do I need? I had the link https://www.nvidia.com/en-us/drivers/unix/ to find the drivers, and once again I had to search the drivers to find which supported my card. Fortunately it was the newest one, 535.113.01. Load from packages? Not there, only nvidia-driver-535.104.05_1. OK, load that, note the important parts of the pkg-message:
To use these drivers, make sure that you have loaded the NVidia
kernel module, by running
# kldload nvidia-modeset
OK, do that, also remove the old driver for whatever I was using in March. Also reinstall nvidia-settings, but that proved to be the same version, 470.86_1.
Load the kernel module, run nvidia-xconfig on an empty /etc/X11 directory:
=== root@hydra (/dev/pts/0) /etc 31 -> nvidia-xconfig
WARNING: Unable to locate/open X configuration file.
New X configuration file written to '/etc/X11/xorg.conf'
And as I expected, it was an absolute bare-bones config file, only one display. Still, I have something. Run startx. Flashing red lines on the display, panic! I don't have a dump partition on this machine (it would take up 5% of my expensive SSD, and I shouldn't need a swap partition), so not much to do there.
Reboot, try again. It works! Well, I got one blank screen with a large mouse cursor, interestingly at the other end of my row of monitors. But it seemed to be dead in the water.
=== grog@hydra (/dev/pts/4) ~ 1 -> xterm -display :0
Authorization required, but no authorization protocol specified
xterm: Xt error: Can't open display: :0
I haven't seen that one before, and ps tells me
root 2239 0.0 0.0 25283568 79332 0 I< 14:28 0:01.07 /usr/local/libexec/Xorg :0 -auth /root/.serverauth.2226
/root/.serverauth.2226 exists. So what's the problem? When I tried as root, I didn't get the error message, just a hang. Somehow got xdpyinfo to run, showing a resolution of 640x480.
But then I looked at my list of kernel modules. No nvidia-settings. Where did it go? I'm 100% sure that I loaded it. Load it (again) and things look very different: now it started a single display apparently across all four monitors. That's not what I want here, but exactly what I was looking for 10 years ago for eureka server 1. There I had to use the Xinerama option, maybe because more than one display card was involved. Here it seems unnecessary.
Well, maybe it was all 4 monitors. The one on the left came on too, but the ones in the middle didn't. With a bit of fiddling (why does every monitor have a different way to select inputs?) got monitor 1 (second) working, but when I powered on monitor 2, it turned off after about 5 seconds. It's my oldest, and I suspect that in the last 6 months it gave up the ghost.
Damn. Those were all the monitors I had. Oh, no, feel the world. I've had a 7" Feelworld FW760 portable display for my cameras for nearly 6 years. It has an HDMI input and a resolution of 1920x1080, just what I need except for the size. And that worked:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Interestingly, when I stopped X, I returned cleanly to a vty display, something that isn't always the case. And it was the same display on all 4 monitors!
That was relatively painless, but we still have a long way to go:
After that, I need to get my mail client working properly on hydra, and then I should be able to swap eureka and hydra and continue migrating at my pace.
| Monday, 23 October 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
hydra consolidation
|
Topic: technology, opinion | Link here |
Somehow I don't have enough time to invest in setting up hydra. Today, apart from documentation, I did relatively little. I was able to get my mail client and firefox to run on eureka's X server, however. That's much faster than with dereel, which was continually running into memory constraints. And that went surprisingly smoothly; soon I'll be able to power dereel down.
|
Olympus image quality
|
Topic: photography, opinion | Link here |
Yesterday I took some photos of the first 4-screen display on hydra. It was convenient to use the Olympus E-PM2, which has pretty much the same sensor as the Olympus OM-D E-M1 Mark I.
Or so I thought. Yes, the shades were down, and so I took this photo at 1/60 s, f/1.7 and 3200/30° ISO:
|
|||||||||||||||||||||||
Despite noise reduction with DxO PhotoLab, there's a lot of noise. I should check how it compares with the E-M1s.
|
Bloody weather station!
|
Topic: technology, general, opinion | Link here |
One thing that seems to have changed with the arrival in position of hydra is my weather station. It's the most unreliable thing I know, and it has an only marginal wireless connection with the outside station. And it seems that the presence of hydra has so insulted it that it didn't want to communicate at all.
Tried various alternatives. tiwi, in the lounge room, seemed a good choice, since it has a direct view of the outside station about 10 m away. But I couldn't start the program on. First there was library hell.
ld-elf.so.1: Shared object "libmysqlclient.so.20" not found, required by "wh1080"
Oh, how about that, I never installed it on tiwi. Dammit, what's the command for re-scanning the library paths. ldd? No, that just shows what libraries a program needs, and where (if anywhere) it finds them. ldconfig! OK, ran that. And yes, it found the library, but:
=== grog@tiwi (/dev/pts/4) ~/src/weather/WH-1080-tiwi 39 -> wh1080 Stones-road grog
wh1080 starting
Can't connect to databaase: Can't connect to local MySQL server through socket '/tmp/mysql.sock' (2) (2002)
What's that? A permission? Normally MySQL has exemplary error reporting. Is it even from MySQL? Normally it spells “database” correctly. Before chasing up the error number, how about checking whether it works with
=== grog@tiwi (/dev/pts/8) ~/src/weather/WH-1080-tiwi 9 -> su
Password:
ld-elf.so.1: Shared object "libreadline.so.8" not found, required by "bash"
Huh? Have I messed up the ldconfig call? RTFM time, but no, it should have worked. It appears that there's some issues with libraries in the parent directory (/usr/local/lib) and a subdirectory (/usr/local/lib/mysql). Finally got the libraries to work again, but I still couldn't get the program to work.
Dammit, we have other machines. Finally installed it in the window of Yvonne's office, again in full view of the outside station about 15 m away. And yes, it received a signal. More library pain, and finally it ran. Well, at least it hung during initialization. gdb time, sort of; in fact, I didn't have any time, so I postponed.
How I hate this device!
|
Bruno
|
Topic: animals, opinion | Link here |
Bruno is still not eating much. He doesn't look unhealthy; on the contrary, he has lost a little weight, and that suits him. But we still want to get him to eat his pellets, and today we finally got him to eat about 15 g, which I think is enough. We'll have to ensure that he eats at least 10 g a day.
| Tuesday, 24 October 2023 | Dereel → Ballarat → Dereel | Images for 24 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Feeding Bruno: success?
|
Topic: animals, opinion | Link here |
Bruno is still not happy with his pellets, but today hunger got the better of him, and he finally finished the second round of pellets that we had given him.
But clearly he doesn't like them—at the moment, anyway. Maybe we should give him canned food instead.
|
Garden flowers in mid-spring
|
Topic: gardening | Link here |
It's a couple of days over a month since the September equinox, but the weather has only now allowed me to take the monthly garden photos.
At this time of the year, the garden should be at its best. It isn't. Multiple issues have come to bear that seem to make each year worse than the one before, at least since we moved to Stones Road. At least part of it is clearly due to the fact that I no longer have the energy or inclination to do it myself.
One of the few new things is the Maple “Chantilly Lace” in front of the house:
|
|||||||||||||||||||||||
At a distance it looks good, but even in this relatively protected position the leaves are suffering:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The Abutilon outside the laundry that looked so sickly a year ago has recovered with a vengeance. Here a year ago and now:
|
||||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
But the other one, on Piccola's grave, is still not looking good:
|
|||||||||||||||||||||||
The nearby Acacia is now enormous and looking very pleased with itself:
|
|||||||||||||||||||||||
It's gradually coming into flower:
|
|||||||||||||||||||||||
The roses are doing surprisingly well, considering that they haven't had much attention:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The remaining Clematis is flowering, but there are almost no leaves:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
But that's about all. The Hibiscus rosa-sinensis has recovered from last year, but there's still no evidence of buds:
|
|||||||||||||||||||||||
Probably an issue of fertilizer.
It's fairly clear that the Alyogyne huegelii is 9 parts dead:
|
|||||||||||||||||||||||
Only a small stem is still alive:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
I've been watching with some pleasure how the seed spikes of this wild grass have grown over the years. They don't seem to go away by themselves:
|
|||||||||||||||||||||||
But that was last month. Today, to my horror, I saw:
|
|||||||||||||||||||||||
A misunderstood “please remove the dead flowers from the beds”. That took 8 years to grow, and now it's on the compost heap:
|
|||||||||||||||||||||||
Sometimes I wonder why I bother.
|
More weather station woes
|
Topic: technology, general, opinion | Link here |
Why isn't my weather station software running properly on lagoon? It looks like it's hanging trying to read the configuration, which is in a MySQL database on eureka. But there are no error messages.
OK, gdb time:
=== root@lagoon (/dev/pts/2) /home/grog/src/weather/WH-1080-lagoon 40 -> gdb wh1080
bash: gdb: command not found
What? No debugger? Ah, I recall that people were having difficulty keeping it in sync with upstream development. OK,
=== root@lagoon (/dev/pts/2) /home/grog/src/weather/WH-1080-lagoon 41 -> pkg install gdb
Newer FreeBSD version for package zziplib:
To ignore this error set IGNORE_OSVERSION=yes
- package: 1204000
- running kernel: 1201503
Ignore the mismatch and continue? [y/N]: y
The following 109 package(s) will be affected (of 0 checked):
Installed packages to be REMOVED:
avidemux-plugins: 2.7.4
chromium: 79.0.3945.130
gstreamer-plugins-vp8: 0.10.23_7,3
opera: 12.16_6
spidermonkey52: 52.9.0_5
tesseract: 4.1.0_3
Ports hell again! OK, build from source. But it hangs, and of course it wants to install hundreds of support ports, finally running into problems wanting to replace an installed port of something like readline. No, sorry, this is Yvonne's machine. I don't want to risk breaking it for her.
OK, how about the version of gdb on eureka?
=== root@lagoon (/dev/pts/2) /usr/ports/devel/gdb 49 -> /eureka/usr/bin/gdb
ld-elf.so.1: Undefined symbol "readline_echoing_p" referenced from COPY relocation in /eureka/usr/bin/gdb
Nope, that's enough. Once again the Ports Collection has beaten me. What alternatives do I have? Finally found a USB extension cable and put the station in a shelf behind the cameras:
|
|||||||||||||||||||||||
And how about that, there it was able to communicate with the outside station (on the other
side of the wall). So apart from the cable lying on the ground, the problem has
been solved worked around.
|
Dentist again
|
Topic: health | Link here |
Into town to see Mario Cordioli, the dentist. He had wanted to check how my gums had healed since the last visit, and was happy with what he saw. But looking back to that occasion, I'm not sure that he checked everything that he wanted to. Presumably we'll see next time.
|
hydra and X: the next step
|
Topic: technology, opinion | Link here |
The functions that I'm using on hydra are now working well. The next step is to continue configuring X. And there the confusion starts.
What fonts? I seem to have most fonts that I need on eureka, but not on tiwi. And how do fonts work, anyway? If I run a web browser on hydra and displaying on eureka, some fonts, notably Chinese, Korean and Devanagari, are missing. Clearly the client is looking for the fonts on hydra, but where? And the configuration file generated by nvidia-xconfig doesn't have any mention of fonts at all. On eureka, /etc/X11/xorg-0.conf contains:
FontPath "/usr/local/lib/X11/fonts/misc/:unscaled"
FontPath "/usr/local/lib/X11/fonts/100dpi/:unscaled"
...
But on hydra the directory /usr/local/lib/X11/ doesn't exist! Clearly there are fonts somewhere, but where? What does pkg info say?
=== root@hydra (/dev/pts/4) ~ 122 -> pkg list xorg-fonts
/usr/local/share/licenses/xorg-fonts-7.7_1/LICENSE
/usr/local/share/licenses/xorg-fonts-7.7_1/NA
/usr/local/share/licenses/xorg-fonts-7.7_1/catalog.mk
And that's all. That's rather less than I expected. Much searching brought me to the hierarchy /usr/local/share/fonts/. And yes, there are plenty of fonts there. In fact, all the packages whose names start with xorg-fonts are already installed.
So what am I even looking for? Off searching the web, not for the first time. This page seems to be a good start, but it takes me back to /usr/local/lib/X11. Recommended Xorg Font Configuration Method, from the FreeBSD forums, seems to be the obvious choice. But it refers to fonts that I can't find.
It does, however, link to the FreeBSD handbook, which contains a lot of stuff that contradicts what I recall. And nothing so far addresses clients such as firefox or xterm.
I've been fighting X for a third of a century now. I'm making progress, but the goalposts keep shifting.
| Wednesday, 25 October 2023 | Dereel | Images for 25 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Fixing the irrigation
|
Topic: gardening | Link here |
Devin Lafranchi along today to take a look at the irrigation system. It took him 2½ hours, and he only managed two of the four circuits. He'll be back, probably on Friday, to do the other two. At least that's an indication that I was right not to try to do it myself.
|
More X font investigations
|
Topic: technology | Link here |
What's the most reliable description of X fonts? Clearly it should be what x.org publishes. But somehow it wasn't on the list of options that I got during my search.
But it's there: Fonts in X11R7.7. And it's worth reading. In particular, on page 8:
If no font path is specified in a config file, the server uses a default value specified when it was built.
And that's how the current server on hydra manages to find things.
But it's not enough. A lot of stuff has changed in the last third of a century, including how fonts are specified. I still need to get my head around this statement (on page 7):
You can specify a special kind of font path directory in the form catalogue:<dir>. The directory specified after the catalogue: prefix will be scanned for symlinks and each symlink destination will be added as a localfont path entry
And together they explain what I found in /var/log/Xorg.0.log on hydra:
[ 8999.408] (==) FontPath set to:
/usr/local/share/fonts/misc/,
/usr/local/share/fonts/TTF/,
/usr/local/share/fonts/OTF/,
/usr/local/share/fonts/Type1/,
/usr/local/share/fonts/100dpi/,
/usr/local/share/fonts/75dpi/,
catalogue:/usr/local/etc/X11/fontpath.d
On the other hand, comparing the font packages installed on eureka and hydra shows that yes, all 7 xorg-fonts packages are installed on both. But why restrict yourself to xorg-fonts when you can have just font-*? Looking on eureka I find a whole lot of things like:
font-cronyx-cyrillic-1.0.3_4 X.Org Cronyx Cyrillic font
font-cursor-misc-1.0.3_4 X.Org miscellaneous Cursor fonts
font-daewoo-misc-1.0.3_4 X.Org miscellaneous Daewoo fonts
font-ibm-type1-1.0.3_4 X.Org IBM Type1 font
font-isas-misc-1.0.3_4 X.Org miscellaneous ISAS fonts
font-jis-misc-1.0.3_4 X.Org miscellaneous JIS fonts
font-micro-misc-1.0.3_4 X.Org miscellaneous Micro fonts
font-misc-cyrillic-1.0.3_4 X.Org miscellaneous Cyrillic font
font-misc-ethiopic-1.0.4 X.Org miscellaneous Ethiopic font
font-misc-meltho-1.0.3_4 X.Org miscellaneous Meltho font
font-mutt-misc-1.0.3_4 X.Org miscellaneous Mutt fonts
font-schumacher-misc-1.1.2_4 X.Org miscellaneous Schumacher fonts
What are they all? “Daewoo” looks Korean. Is it? pkg is keeping it very close to its chest:
=== root@eureka (/dev/pts/18) ~ 111 -> pkg info font-daewoo-misc-1.0.3_4
Name : font-daewoo-misc
Version : 1.0.3_4
Installed on : Sun Jan 17 13:54:01 2021 AEDT
Origin : x11-fonts/font-daewoo-misc
Architecture : FreeBSD:12:amd64
Prefix : /usr/local
Categories : x11-fonts
Licenses :
Maintainer : x11@FreeBSD.org
WWW : https://www.x.org/
Comment : X.Org miscellaneous Daewoo fonts
Annotations :
FreeBSD_version: 1201000
repo_type : binary
repository : FreeBSD
Flat size : 795KiB
Description :
This package contains miscellaneous X.Org Daewoo fonts.
That's the complete entry, and it doesn't say much. How about font-mutt-misc?
=== root@eureka (/dev/pts/18) ~ 112 -> pkg info font-mutt-misc-1.0.3_4
Description :
This package contains miscellaneous X.Org Mutt fonts.
And that's the only difference. And this?
=== root@eureka (/dev/pts/18) ~ 113 -> pkg info font-schumacher-misc-1.1.2_4
Description :
This package contains miscellaneous X.Org Schumacher fonts.
Clearly the info is less than helpful.
So it looks as if I will still need to fight my way through the list of fonts to find out what they are and which are needed. On eureka I already have:
=== root@eureka (/dev/pts/18) ~ 118 -> pkg info | grep font | grep X | wc -l
53
And there are probably more. One font set that stood out in the X document was ClearlyU, which seems to be intended for Unicode. But there's nothing of that name in the list of packages.
More searching...
|
Processing photos on hydra
|
Topic: technology, photography | Link here |
hydra is something like 8 times as fast as eureka, and its disks are even faster. Clearly photo processing is one of the things I should be doing next.
And how about that, it almost worked! The only issue was ~/public_html/photos/RCS, a symlink to a file on eureka. But I still need to investigate how to tweak parallelism so that hydra can convert up to 40 photos at once.
And then there's mail. Decades ago (2003/04/17 04:07:38) I started modifying /etc/aliases to create fantasy emails. Nowadays this is handled by /usr/local/etc/postfix/virtual, and I thought that /etc/aliases had done its dash. But no, /usr/bin/mail still uses it, so I need the one entry
+# One Ring to rule them all
+root: grog
+
Also decades ago I ran into trouble with mutt: I couldn't get to the end of the folder with the specified character, so I changed it to G. And that config file isn't on hydra, so it doesn't work. That's one to look at when I have time.
|
More artificial stupidity
|
Topic: technology, opinion | Link here |
Mail from Michael James today, pointing at the latest feats of “Artificial Intelligence”. Here an answer from Bard to the question “if it takes 3 hours to dry 3 towels on a line, how long does it take to dry 9 towels on the line?”:
If it takes 3 hours to dry 3 towels on a line, then it will take 9 hours to dry 9 towels on the line.Assuming that the drying time is directly proportional to the number of towels, we can use the following equation...
ChatGPT gets it half right:
Hi! If it takes 3 hours to dry 3 towels on a line, it would take the same amount of time to dry 9 towels on the line. The time it takes to dry the towels is directly proportional to the number of towels being dried.
If that's intelligence, I certainly don't want to see stupidity.
| Thursday, 26 October 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Another monitor glitch
|
Topic: technology | Link here |
My old Matrix monitor came up with a new trick today: distorted display, again only a few seconds. Can it be that the problem is with the display card and not the monitor?
|
Next firefox trick
|
Topic: technology, opinion | Link here |
I run my :0 X server with 4 X screens, one per monitor, as opposed to a single X screen across 4 monitors for server :1, something that Callum Gibson doesn't understand. That means that I can place a different browser window on each screen (monitor).
But firefox has always made it difficult. Two different firefox instances on the same screen refuse to work, even if they're started from different machines: the second one hands over to the first, meaning that I get all the settings of the first instance. That's one of the reasons why I have 4 screens in the first place.
But today I discovered that firefox has gone one step further. Clearly it still finds more than one instance an abomination, and now if I have an instance running on eureka:0.2 (third screen) and I try to start another on eureka:0.1 (second screen), it simply associates with the third screen, and I get no display at all where I asked for it. It seems related to the server: if I start an instance on eureka:1, it works.
Well, sort of. Clearly I need a different profile for each instance (one of the reasons I still don't use Chrome). But today, after starting the second instance and then trying to display a mail message on the first one, it couldn't find the running instance! It tried to start a new one (hah!) on the same screen. I had to tell the profile selector what it should already have known and select the profile name of the running instance.
Why does firefox do this? Is there maybe a way to work around it? Or is it just another instance of this horrible “There can only be one” mentality that seems to pervade the Microsoft space? And is that the reason why it now crashes several times a day?
|
Alternative video access methods
|
Topic: multimedia, technology, opinion | Link here |
For a couple of years now I have been using youtube-dl to download videos. Apart from the finer-grained approach to displaying the videos (in particular single frame forwards and backwards with mpv and the lack of network issues), it also seems to avoid the advertisements that are becoming ever more invasive online.
But they're not the only game in town. Today I got some other points, incongruously from the Unix Heritage Society mailing list. Here a selection:
You can enjoy non-chopped up videos by replacing youtube.com by yewtu.be in related URLs (easiest way to remember).Or you can paste a youtube URL in the search box of any Invidious instance, like https://invidious.fdn.fr.
Other instances listed here in case one of them is down: https://docs.invidious.io/instances/
|
More font investigations
|
Topic: technology, opinion | Link here |
So, do I have the ClearlyU fonts? Yes, sort of. Part of an OpenBSD CVS repository, where I found:
Version: 1.8
6 December 2000
Clearly that's an old, worn-out magic font.
Callum Gibson came up with another link, https://en.wikibooks.org/wiki/Guide_to_X11/Fonts
But then I came up with this document, explaining how to set fonts specifically for xterm. The specifications are in the .Xdefaults file.
And of course I (once) knew that. But why has it taken me such a long time to find it again? And why has hydra ended up with such inappropriate fonts in .Xdefaults? Tomorrow I may find out.
There's also a mention of xfontsel, another program I once knew. Do I need it?
| Friday, 27 October 2023 | Dereel | Images for 27 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Bruno's progress
|
Topic: animals, opinion | Link here |
Bruno was formally adult at the end of last month, at least according to the Governing Council of the Cat Fancy. According to another metric, he'll be adult at the end of this month. But he's still sleeping in a dog cage in Yvonne's bedroom. Time for him to enter the time-honoured tradition of spending the night on top of the freezer:
|
|||||||||||||||||||||||
That's his favourite basket, and it has clearly seen better days. Yvonne bought a new one on Wednesday, but he didn't want to know about it. We've been putting this one on the freezer in the daytime to see how he comes to terms with it. Yesterday he threw it on the ground, but today he went into it voluntarily.
Later in the day I found him scraping the side of the wash basin in the bathroom, but I didn't get time to take a photo. Only later did we discover that the cage (with litter box) was closed, and he held on for several hours before finally relieving himself in Yvonne's basin.
|
Anke Hawke again
|
Topic: animals, health, opinion | Link here |
Anke Hawke along today for another clinic. That's becoming less and less of a factor for me: she's now staying at Chris Bahlo's place, and we won't be having a dinner. Yvonne picked her up and brought her back to our place so that she could give Yvonne a riding lesson, so at least I saw her.
And she told me that I had lost weight. Not so: my weight has barely changed since the last time she was here. But maybe my improved posture has given that impression.
|
More font pain
|
Topic: technology, opinion | Link here |
Why does xterm render different font subsets on eureka and elsewhere? Clearly eureka has more fonts, but which? I've been saving email messages, nearly all spam, which have non-renderable characters in the Subject: line, like this:
|
|||||||||||||||||||||||
|
|
|||||||||
The images can be enlarged, but the second is a detail starting at the subject field of message 16. Clearly there's not much to be found there.
More reading, notably this one, which explains the .Xresources file that I've been seeing. There's nothing special there: it's just a new name for what we've been calling .Xdefaults all our lives.
And then there's this page, which suggests that xfontsel only works for old-style fonts. It also gives an idea of how to test fonts, somewhat Linux-centric:
$ xterm -fa <truetype font> fs 12
$ xterm -fa "DejaVu Sans Mono"
$ xterm -fa "Ubuntu Mono" -fs 10
But somehow there's a disconnect here. X server fonts are defined in a list of directories; the fonts for xterm are defined either in /usr/local/lib/X11/app-defaults/XTerm or a local .Xresources or whatever. But there are several of them, here from /usr/local/lib/X11/app-defaults/XTerm on hydra:
*fontMenu.Label: VT Fonts
*fontMenu*fontdefault*Label: Default
*fontMenu*font1*Label: Unreadable
*VT100.font1: nil2
*IconFont: nil2
*fontMenu*font2*Label: Tiny
*VT100.font2: 5x7
*fontMenu*font3*Label: Small
*VT100.font3: 6x10
*fontMenu*font4*Label: Medium
*VT100.font4: 7x13
*fontMenu*font5*Label: Large
*VT100.font5: 9x15
*fontMenu*font6*Label: Huge
*VT100.font6: 10x20
*fontMenu*font7*Label: Enormous
!*VT100.font7: 12x24
*VT100.font7: -adobe-courier-medium-r-normal--24-240-75-75-m-150-iso10646-1
There's no specification for Default; presumably that's what it gets started with, in my case (from the window manager configuration file):
/usr/local/bin/xterm -name "xterm" -bg BlanchedAlmond -s -sl 2048 -sb -ls -j -rw -display :0.1 -geometry 90x50+53+0 -fn 10x20 -e /usr/local/bin/bash &
And yes, the commented out font for Enormous appears to be in the distribution. This file also contains the comments:
! xterm can switch at runtime between bitmap (default) and TrueType fonts.
! The "faceSize" resource controls the size of the latter. However, it was
! originally given with a size that makes the two types of fonts different
! sizes. Uncomment this line to use the same size as "fixed".
!*faceSize: 8
I still need to consider what that means.
So: what are these fonts? Aliases, if I recall correctly. But that suggests just one font, which was fine in the days of ASCII. But now? I don't seem to have any font set that covers all the characters I want to display.
But there are 7 different fonts in the description above. Do any of them do better? Yes!
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|
|
|||||||||
|
|
|||||||||
The first is the Default font, and the second is the Huge font (run the cursor over an image to compare it with its neighbour). But they should be the same! The default from the window manager is 10x20, and that's what's in the config file for Huge. Still, now the fonts on hydra are as good as on eureka, but not good enough. Still, for the time I'll put up with that. I don't really need Chinese display on an xterm, though I do on my web browser.
So back to my search for East Asian fonts for the server. And yes, it wasn't until later that I realized that that's not the same thing as the browser. Somewhere I found a suggestion to install urwfonts. OK, did that, and got the message:
You'll have to add /usr/local/share/fonts/urwfonts
to your X font path by either:
$ xset fp+ /usr/local/share/fonts/urwfonts
$ xset fp rehash
or by adding it to your X-server configuration file (usually
/etc/X11/xorg.conf) and restarting the X-server.
Or, of course, both. But let's test them first:
=== root@hydra (/dev/pts/4) /usr/local/share/fonts/100dpi 156 -> xset fp+ /usr/local/share/fonts/urwfonts
xset: bad font path element (#17), possible causes are:
Directory does not exist or has wrong permissions
Directory missing fonts.dir
Incorrect font server address or syntax=== root@hydra (/dev/pts/4) /usr/local/share/fonts/100dpi 157 -> xset fp rehash=== root@hydra (/dev/pts/4) /usr/local/share/fonts/100dpi 158 ->
That doesn't look good, and the error message doesn't tell me what the problem really is. By inspection, at least it wasn't a missing fonts.dir. In any case, after that I saw no difference. Moving on, this page, from the FreeBSD forums, tells me to set the fonts from firefox preferences. And the fonts I was looking for were already there! But selecting them made no difference.
But this page, also from the FreeBSD forums, says:
$ pkg search noto | grep -i fonts
noto-2.0 Google Noto Fonts family (meta port)
noto-basic-2.0_1 Google Noto Fonts family (Basic)
noto-emoji-2.020 Google Noto Fonts family (Emoji)
noto-extra-2.0 Google Noto Fonts family (Extra)
noto-hk-2.001 Google Noto Fonts family (Traditional Chinese Hong Kong)
noto-jp-2.001 Google Noto Fonts family (Japanese)
noto-kr-2.001 Google Noto Fonts family (Korean)
noto-sc-2.001 Google Noto Fonts family (Simplified Chinese)
noto-tc-2.001 Google Noto Fonts family (Traditional Chinese)
Of course! Why search for fonts when you can search for noto? OK, try that:
=== root@hydra (/dev/pts/4) /usr/local/share/fonts/urwfonts 168 -> pkg install noto-sc-2.004
New packages to be INSTALLED:
noto-sans-sc: 2.004
noto-sc: 2.004
noto-serif-sc: 2.001
The process will require 435 MiB more space.
292 MiB to be downloaded.
[1/3] Fetching noto-sans-sc-2.004.pkg: 17% 22 MiB 1.4MB/s 01:34 ETA
435 MB! What's in there? The urwfonts took up 2 MB. Still, we have the space, even if it seems ridiculously bloated. And how about that, it worked! So clearly firefox was lying when it claimed to have Chinese fonts.
But wait, there's more! I only installed the noto-sc (Simplified Chinese) fonts. But I also got Japanese and Korean as well. They're supposed to be separate font sets.
And Devanagari? Somewhere I found a page pointing me to lohit, another obvious word meaning “font” (in fact, it's Sanskrit for “red”, and I don't understand the connection). OK, install them:
=== root@hydra (/dev/pts/4) /usr/local/share/fonts/urwfonts 178 -> pkg search lohit
Lohit-20140220_3 Lohit fonts=== root@hydra (/dev/pts/4) /usr/local/share/fonts/urwfonts 179 -> pkg install lohit
New packages to be INSTALLED:
Lohit: 20140220_3
The process will require 1 MiB more space.
292 KiB to be downloaded.
...
[1/1] Extracting Lohit-20140220_3: 100%
=====
Message from Lohit-20140220_3:
To use these fonts, add the following line to the "Files" section of
xorg.conf:
FontPath "/usr/local/share/fonts/Lohit/"
And how about that, that also worked out of the box. Went through all the pages that I could find in Wikipedia and found that just about all rendered correctly. The ones that didn't are:
And I don't even know what those are.
So what do I do with xterm? The difference between Default and Huge puzzled me. Are there any more surprises?
Yes! The Large font is clearly smaller than Huge, but it has more characters:
|
|||||||||||||||||||||||
Here the difference:
|
|
|
|||||||||
|
|
|||||||||
And that's about all I need.
But why is this so difficult? This should work out of the box, but I've spent days trying to find the Right Way to set fonts, and all I have is something that more or less works. Is xterm maybe finally ready for the scrap heap? Is there something more modern that doesn't have stupidity like big borders and forced colours?
| Saturday, 28 October 2023 | Dereel | Images for 28 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Mee goreng mamak again
|
Topic: food and drink, opinion | Link here |
Mee goreng mamak for breakfast today, the first time for over 18 months.
Why? It's not overly interesting, and somehow the preparations, in particular the boiled potato, seem excessive. Today only the cooking was excessive:
|
|||||||||||||||||||||||
And that was only 30 minutes. I wonder why it disintegrated so quickly.
Minor tweaks to the recipe: add garlic, more sambal. The result shows the disintegrated potatoes:
|
|||||||||||||||||||||||
Once again, boring. But that as at least in part because I had forgotten the limes. I don't know how much that would have helped.
|
Fonts: Microsoft beats X!
|
Topic: technology, opinion | Link here |
Spent much of today consolidating and documenting my experience with X fonts. I had wondered whether I had messed something up with my ~/.Xdefaults file, but there is none on hydra. So what I got is standard X. What a catastrophe!
While writing it up, accidentally executed (view-hello-file), which brings up the word or phrase for “Hello” in multiple languages. And some of them didn't render correctly:
|
|
|||||||||
That could match the web sites that I couldn't display yesterday.
So: I can display the fonts that I want with both xterm (as long as I choose a font size that I don't really want) and with firefox. Where do I go from here?
First, are there more modern replacements for xterm? Are there! 10? 25? 50? Just looking at the descriptions turns my stomach. But I'll have to look at the issue—later.
And then, how does Microsoft handle them? Perfectly, out of the box! Clearly there's a lot of work to be done to get X (on FreeBSD at any rate) even up to Microsoft's standards!
|
Bruno
|
Topic: animals, opinion | Link here |
Bruno is obviously not happy with being moved to the laundry, and over the course of the day he brought his basket and the base cushion back as far as he could manage towards the bedroom:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
He's also eating more, maybe to make up for his fast a few days ago. Today he even ate his pellets.
|
Salmon “Wellington”
|
Topic: food and drink, opinion | Link here |
ALDI had a “salmon Wellington” on special this week, so Yvonne bought one:
|
|||||||||||||||||||||||
What does it taste like? Lots of slime (“cheese sauce”) inside, not enough salt, somewhat boring.
|
|||||||||||||||||||||||
Is that what the typical Australian customer likes?
| Sunday, 29 October 2023 | Dereel | Images for 29 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Bloody Bruno!
|
Topic: animals, Stones Road house, general, opinion | Link here |
First thing this morning, Yvonne told me that she had a problem: Bruno had pulled the fly screen out of the window of the toilet, and it was now lying on the ground and blocking the door:
|
|||||||||||||||||||||||
The window only opens a little bit (enough for me to take the photo above):
|
|||||||||||||||||||||||
It also enabled us to get Bruno out.
So: how do we get it out? Spent some time considering. Ah! When the house was built, I complained about the excessive gap above the doors in the toilets:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
They told me that it was a building regulation: if somebody collapsed inside the toilet and blocked the door, all you needed to do was lift the door off the hinges. Almost exactly our situation. So I tried that. No go. The gap was too small! What a waste of effort!
More thoughts, and ultimately we came up with a “fishing line”, a cable tie strapped to a curtain rail:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
With that I was able to lift it up enough for Yvonne to open the door and grab it, then lift it all the way.
What a pain!
|
7 months of Bruno
|
Topic: animals, opinion | Link here |
We've had cats for as long as I can remember. But somehow Bruno is different:
And somehow he's quite a character. Are we having more experience because he's not allowed outside? It's quite possible.
Apart from his antics in the toilet this morning, we're gradually moving him to the laundry to eat and sleep. And he seems hungrier. He's even eating the pellets that he rejected only a few days ago. It looks like we have more to expect.
|
xterm replacement?
|
Topic: technology, opinion | Link here |
Didn't do much work on the computers today, but I did try a couple of sites looking for a replacement for xterm. As I feared, not much help. I'll try again some other time.
|
Photos the “normal” way
|
Topic: technology, animals, photography, opinion | Link here |
Yesterday Yvonne had a riding lesson at Anke Hawke's clinic at Chris Bahlo's place. Yvonne asked Chris to take photos of her with her (Yvonne's) Olympus OM-D E-M5 Mark III. But no, Chris, who herself has a Olympus OM-D E-M10 Mark II, chose to take the photos with her mobile phone!
Why? I still have no idea. One of the important things about this kind of photo is the distance, and mobile phones don't have zoom lenses. Still, she took them. Now how to access them? Yvonne received information via Facebook, but fortunately it was just a URL for a collection at Google Drive, also not my favourite system. There were 64 images. How do you select them all? There was nothing obvious, and I ended up having to drag the cursor over multiple pages.
Then press “Download”, up near top left. Nothing happens. Try again. Ah, there, at bottom right, about as far from the Download button as can be: “Zipping files”.
Finally they came, with file names reflecting yesterday's date, not once, but twice. And they weren't (quite) the same size!
-rw-r--r-- 1 grog lemis 336,494,133 29 Oct 13:19 drive-download-20231029T021344Z-001.zip
-rw-r--r-- 1 grog lemis 336,507,004 29 Oct 13:19 drive-download-20231029T021347Z-001.zip
What's the difference?
=== grog@hydra (/dev/pts/15) ~/Downloads 24 -> unzip -l drive-download-20231029T021347Z-001.zip > foo2=== grog@hydra (/dev/pts/15) ~/Downloads 25 -> unzip -l drive-download-20231029T021344Z-001.zip > foo1=== grog@hydra (/dev/pts/15) ~/Downloads 25 -> diff -wu foo*
--- foo1 2023-10-29 13:38:08.437605000 +1100
+++ foo2 2023-10-29 13:38:04.533937000 +1100
@@ -1,67 +1,67 @@
-Archive: drive-download-20231029T021344Z-001.zip
+Archive: drive-download-20231029T021347Z-001.zip
Length Date Time Name
-------- ---- ---- ----
- 4197505 10-27-23 21:00 20231028_150054.jpg
- 3214693 10-27-23 20:51 20231028_145125.jpg
- 3316805 10-27-23 20:52 20231028_145212.jpg
- 3167043 10-27-23 20:51 20231028_145127.jpg
+ 3925406 10-27-23 21:00 20231028_150037.jpg
2559290 10-27-23 21:00 20231028_150024.jpg
+ 3310907 10-27-23 20:52 20231028_145221.jpg
+ 3167043 10-27-23 20:51 20231028_145127.jpg
+ 3316805 10-27-23 20:52 20231028_145212.jpg
+ 3214693 10-27-23 20:51 20231028_145125.jpg
+ 2794740 10-27-23 21:00 20231028_150029.jpg
What happened there? It seems that Google Drive zips in random order. Tried the same with sorted lists and it seems that they're both the same, though that doesn't explain the difference in size of the archives.
Finally Yvonne processed the images. For a mobile phone, they're quite acceptable. But looking at the crops tells a different story:
|
|
||||||||||
That's 0.55 out of 12 MP, 4.6% of the original image. In general, the images are much smaller than the ones Yvonne took with her camera. And her camera was available. I still don't understand.
|
Mongolian lamb, take 2
|
Topic: food and drink, opinion | Link here |
Last weekend we ate a pre-prepared “Mongolian lamb” from Coles. It was edible, but not spectacular. But I've had a couple of sachets of ALDI “Mongolian lamb” spice paste long enough to let them expire. Today we tried that, somewhat hampered by the lack of capsicum asked for in the recipe:
|
|||||||||||||||||||||||
Better. but not spectacular. Also something that we won't eat too often.
| Monday, 30 October 2023 | Dereel | Images for 30 October 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Ceramic frying pans
|
Topic: food and drink, opinion | Link here |
Two years ago we bought a ceramic-coated frying pan, and I was quickly convinced by its real non-stick properties, allowing me to fry eggs without oil:
|
|||||||||||||||||||||
But that was two years ago. Sadly, like ceramic kitchen knives, it didn't last. Now it's no better than conventional non-stick pans, and it could be worse.
|
Missing photos!
|
Topic: photography, technology, opinion | Link here |
Juha Kupiainen told me today that not all the photos in this month's diary were also displaying. Checking, I discovered that I was behind in syncing no fewer than 5 days' photos. How did that happen? In any case, time to resync all of them:
=== grog@eureka (/dev/pts/23) ~/Photos 187 -> for i in [12]*; do (cd $i && make sync); done
That will take a while.
|
Fonts: still not done
|
Topic: technology, opinion | Link here |
Checking my diary today, I had a link to Burmese cats. Oh. For that, I need “Burmese” and Thai fonts. I'm still not done.
Have I missed something about TrueType fonts?
|
Replace xterm with rxvt?
|
Topic: technology, opinion | Link here |
One of the names that cropped up relatively frequently in my search for an xterm replacement was rxvt. Do we have one in the Ports Collection? Yes, x11/rxvt-unicode, and it promises:
rxvt-unicode is a clone of the well known terminal emulator rxvt, modified to store text in Unicode (either UCS-2 or UCS-4) and to use locale-correct input and output. It also supports mixing multiple fonts at the same time, including Xft fonts.
Well, if rxvt is so well-known, why don't we have a port? Anyway, installed it:
...
[3/5] Installing rxvt-unicode-9.31...
[3/5] Extracting rxvt-unicode-9.31: 100%
...=== root@hydra (/dev/pts/4) /usr/local/share/fonts/urwfonts 183 -> rxvt
bash: rxvt: command not found=== root@hydra (/dev/pts/4) /usr/local/share/fonts/urwfonts 184 -> man rxvt
No manual entry for rxvt=== root@hydra (/dev/pts/4) /usr/local/share/fonts/urwfonts 186 -> ls -lcrt /usr/local/bin/*xv*
-rws--x--x 1 root wheel 1,712,384 30 Oct 11:41 /usr/local/bin/urxvt
-rwxr-xr-x 1 root wheel 10,808 30 Oct 11:41 /usr/local/bin/urxvtc
-rwxr-xr-x 1 root wheel 89 30 Oct 11:41 /usr/local/bin/urxvtcd
-rws--x--x 1 root wheel 1,721,296 30 Oct 11:41 /usr/local/bin/urxvtd
Why do people do that? At the very least you'd expect a message to tell you the name of the executable, not to mention the purpose of the other 3 files. Started it and noted that it looked quite similar to xterm, but there was little evidence of Unicode support. That might be partially because I couldn't find a way to change the font size. Here first xterm with the Large font, then urxvt with the only (much smaller) font that I could get:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|
|||||||||
|
|
|||||||||
It looks as if I didn't read carefully enough. Unicode, not UTF-8. Can it be modified for UTF-8? I don't know if I care enough to find out.
| Tuesday, 31 October 2023 | Dereel | Images for 31 October 2023 |
| Top of page | ||
| previous day |
|
Yet another xterm font attempt
|
Topic: technology, opinion | Link here |
I still haven't found a good solution to my xterm font problem. More searching brought me to how to use ttf fonts in xterm, from the NetBSD project. That looked like just what I was looking for, especially since it was probably related in environment.
It's simple: install the fonts/liberation-ttf package and enter:
$ xterm -fa 'Liberation Mono' -fs 10
Oh. FreeBSD doesn't have a fonts/liberation-ttf port. A bit of searching showed that it's called x11-fonts/liberation-fonts-ttf, with the information
Liberation fonts from Red Hat to replace MS TTF fonts
OK, try that. It works, and the -fs parameter allows me to set the size. Only: no non-European fonts. In this form it's worse than what I already have.
So: what I really need is to find what characters each font can represent. I'm still not much closer to my goal.
In passing, wouldn't it be nice if the projects could agree on package/port names?
|
Bruno's latest exploits
|
Topic: animals | Link here |
Bruno seems to be at least as interested in heights as Rani. Today he found a new goal, the display cupboard to the left of the TV. It's not ideal—even less than elsewhere—because of the studio flash on top of it:
|
|||||||||||||||||||||||
How can I keep him away? Put an obstacle in his path (here a waste paper basket with the obligatory “Designed for Microsoft Windows” sticker on it)? No, surprisingly that didn't stop him:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Fire the flash to scare him? No, he's made of sterner stuff:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
I really don't know what to do. Probably it would be a good idea to remove the flash when it's not needed.
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 |
