|
| Greg's diary |
| June 2020 |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Monday, 1 June 2020 | Dereel → Ballarat → Dereel | Images for 1 June 2020 |
| Top of page | ||
| next day | ||
| last day |
|
Winter arrives
|
Topic: general, Stones Road house, opinion | Link here |
First day of winter today. It wasn't cold, but it was wet, windy and miserable. In our state-of-the-(Australian)-art house we have fogged up windows and drafts through the enormous gaps under the doors:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
All, we are assured by the builder and DBDRV, correct. May they rot in hell!
|
More mouse experiments
|
Topic: technology, opinion | Link here |
The mouse issue with X on teevee is still not resolved. I've found that the side button remapping on the new mouse (Logitech M705) works, but it's unbearably slow, and xset m seems to have no effect. That in itself is an issue: I currently have three mice connected to the machine, but there's no way to tell xset which mouse you're talking about. The small mice have acceptable cursor speeds, but the M705 is glacial. In addition, it, and only it, seems to have issues such as marking text when deiconfying xterms.
So: for the moment I'm using one of the old mice, and I have the M705 next to it for when I want a button 2 event. Callum Gibson showed some interest and addressed the relationship between X and moused. Yes, stopping moused while X is running caused loss of mouse. But starting X without moused gave a functional mouse! So it seems that moused is only part of the interface. I wish I had better documentation. Callum sent me the logs and his X configuration files, but there was nothing in the mouse section that would have made a difference.
So now my list of things to do is:
|
Finally a haircut!
|
Topic: general, food and drink, opinion | Link here |
Into town today for a number of things, including my first haircut in 5 months. Kerry tells me that he was shut down during the “lockdown”, and only opened again “a couple of weeks ago”. But I had seen the shop open over a month ago. Had he just left the signs on? Or (more likely) not realized how long it had been?
On to look for some pots and pans. This has been on my list for a while: Yvonne has been using old pans that are not suited to induction, and they're all getting tatty. Today a large part of the PTFE coating on her usual frying pan peeled off. And we really don't have many induction-suited saucepans either, so my list had included a 26 cm frying pan, a 30 cm frying pan and a couple of saucepans. To Harris Scarfe, who, I swear, have a permanent sale (“50% off” prices higher than I have ever seen). But they did have a set of, as it happens, a 26 cm frying pan, a 28 cm “sautepan” and two saucepans.
|
|
|||||||||
The “sautepan” (on the right) isn't quite what I wanted, but the combination (“83% off!”) only cost as much as a normal frying pan. And maybe 30 cm wasn't enough for the larger pan anyway; how about 32 cm? Found one there, unfortunately at normal prices (“40% off!”) and thus more than the entire set.
Then on to The Good Guys to look for a replacement for the steam cleaner that failed a couple of weeks ago. They didn't have any! Well, I did find two, with specifications and prices that I could refuse. Why? The shop assistant said that they would be having them in again soon, but then she only saw one of them. Could this be a result of the current restrictions? Certainly they seem to have issues, as this nearby part of the showroom shows:
|
||||||||||||||||||||||||
But somehow I get the feeling that steam cleaners are last year's fad.
Spent a bit of time looking round. What's this?
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
After some examination, I'm fairly sure that it's a vacuum cleaner (“vacuum”), but it's not mentioned anywhere on the packaging. Why do people do this? Clearly this one is for domestic animals, and it's interesting to note that there's a separate one for wild felines:
|
||||||||||||||||||||||||
Another badly thought out advertisement was this:
|
||||||||||||||||||||||||
“Switch from Telstra”! Good idea, but probably not their intention.
Off, without buying anything, to Bunnings, where I spent 45 minutes buying a press button switch, a drill bit and a nut and bolt (nuts carefully hidden). The gate opener has a pair of contacts that will operate the gate when shorted, just what I need an outdoor switch for. But no, couldn't find them anywhere in electrical. Asked an assistant, who didn't understand the concept of a switch that is only on when you press it, so I compared it to a doorbell. Ah, that's down the other end of the shop. At least they had the aisle number right. There I found plenty of door chimes, but no doorbells. Asked an assistant, who told me that the press buttons were only sold together with the chimes, and what brand was it, anyway? Tried to explain the concept of a switch, but gave up and found two different buttons. Took the cheaper, which may be a mistake: it's not waterproof, and it's illuminated, so I'd have to find a way of disabling the illumination.
What a day!
| Tuesday, 2 June 2020 | Dereel | Images for 2 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More mouse insights
|
Topic: technology | Link here |
More discussion about mice with Callum Gibson on IRC today. It seems that he's almost more interested in the issue than I. But our approaches are different: he was looking for (and found) a solution, while I'm trying to understand the architecture so that I can make a choice.
Today he came up with this page, which looks interesting, but which will need some careful reading. This page contains similar information, but also points out that xev takes parameters, and xev -event mouse should stem the voluminous output that it produces by default.
And then xmodmap has an -pp option, which prints the mapping between logical and physical pointer buttons, but doesn't seem to address the issues of chords like the Emulate3Buttons option of mouse(4x). And none of these address the issue of multiple pointers. But I should do something soon: currently I can't paste (button 2), even with the otherwise functional mapping on the M705 mouse.
|
Gate opener: more failures
|
Topic: Stones Road house, general, opinion | Link here |
Planned to insert my new, correctly sized 8 mm bolt into the gate opening mechanism today. Yesterday I had measured the hole, and it was exactly 8 mm. But that was with calipers, and it seems I got it wrong. Yes, the 8 mm bolt fits well, but the hole is more like 8.7 or 9 mm, and that difference translates to about 30 cm in the “closed” position of the gate. Somehow the mechanism is poorly thought out. So it looks like I'll have to position it overclosed and adjust the limit switch. I can only set the position by trial and error, something I don't like.
And then the press button (doorbell) switch. It's illuminated and comes with the helpful information that this can drain the (doorbell) battery, but without any documented or obvious way to disable the illumination. Now clearly the gate opener battery has at least one order of magnitude more charge, but I still don't like the idea. And of course it's 48 V, probably much more than a doorbell (9 V?), so maybe it will just burn out the illumination LED. But maybe I should look for a more appropriate switch.
|
New pans
|
Topic: food and drink, opinion | Link here |
We tried out the new frying pans for breakfast this morning. What features are important for cooking? I can buy el-cheapo pans for $15, or I can buy expensive ones like the new Tefal one for $70. What's the difference?
It doesn't seem to be the anti-stick coating any more. Once upon a time it was PTFE, and particularly in the early days it was very sensitive to damage. But nowadays they use vitreous coatings that are much more scratch resistant, though in the course of time they lose their anti-stick properties, and I think that they're never as good in that respect as the old PTFE coatings.
But there are two other issues that I've seen: evenness of heating and flatness. The evenness is clearly related to the construction of the base. In particular, when frying sausage slices in the pans I have, I've found that they brown better in the middle than on the edges. But this one (here the “sautepan”, which I won't use again for this purpose) does much better:
|
|||||||||||||||||||||||
The other one is the flatness of the surface. Here the pans don't do as well. But I was able to confirm that they flex with temperature. When cold (first photo) an oil film covers the bottom evenly, but when warm the centre rises, leaving the middle dry:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
I think I can live with that. The new expensive one, which I haven't used yet, should not behave like that, if my experience with previous big Tefal pans is anything to go by.
|
More flash investigations
|
Topic: photography, opinion | Link here |
The flash test photos that I took with the Olympus OM-D E-M1 Mark I on Sunday didn't produce the results I expected: at 1/800 s the image was completely black, while the images I took years ago with the E-510 showed a clearly delineated black band. Indication of shutter problems?
The responses I got on the Facebook M43 Tech Talk group were generally useless. They pointed out that the specified sync speed for the E-M1 Mark I is generally not 1/320 s: it's 1/250 s, and it always has been. The 1/320 s is an exception only for the toy flash supplied with the camera. So most of the answers were “you should be using High Speed Sync”. One even suggested that it was operator error, or that the flash didn't cost enough.
Today I took similar photos with 4 of the 6 Olympus cameras at my disposal: the E-30, the E-PM2, the E-M1 Mark II and the E-M5 Mark III. The results? All pretty similar:
So where does that leave me? Clearly from now on I'll set the E-M1 Mark I to 1/250 s for flash, but I don't have the slightest expectation that the problem won't show up again.
|
PayPal pain
|
Topic: technology, general, opinion | Link here |
Yvonne doesn't buy much online, but she tried today. Special socks. Where do they come from? She didn't know. She knew the price, but not the currency. And she wanted the password for PayPal.
OK, went and found that, typed it in. Failure. Go to some URL unrelated to PayPal to try again.
DAMN! Have we fallen prey to a scammer? Quick, in something of a panic, off to change my PayPal password!
How do you do that? I was able to log in with no trouble, but where's the “reset password” tab? Must be behind the calendula icon. But there I find ways to update my photo, my time zone, how to close the account, how to add physical or email addresses. What about my password? Finally I found a link SECURITY at the bottom of the page. Must be that. No, only joking, that's just information about security.
Off to ask Google. Ah, up there above the profile. OK, select that.
|
|
|||||||||
Damn you, PayPal, why do you want me to update my browser in an emergency? Off to find another browser and change password. It promised to send me a code to my primary email address, and I needed that to continue. But it didn't arrive. OK, alternative method: enter the complete number of one of my credit cards (the one ending in 11). That worked, sort of:
|
|
|||||||||
What does that mean? First password wrong, second one right? At any rate I can't save it. Click on the red “other vehicle danger” icon? Yes! The British Highway Code says of this symbol:
Other danger; plate indicates nature of danger
But PayPal was just too polite to tell me what I should have known:
|
|
|||||||||
We wouldn't want passwords to be too secure now, would we? OK, finally got my password changed. Logged in, confirmed that nobody had performed any transactions. But then I got email from PayPal:
44 + 01-06-2020 service@paypal.com.a To Greg Lehey ( 100) + Welcome to PayPal One Touch ???
50 + 01-06-2020 service@paypal.com.a To grogpal@lemi ( 567) + Verification code to reset PayPal password
52 + 01-06-2020 service@paypal.com To Greg Lehey ( 88) + You've just changed your password
What's that One Touch? The email was in only broekn plain text:
You've chosen to skip login with PayPal One Touch TM. From now on you won't be asked for your email or password when paying with PayPal on this device: Desktop Firefox FreeBSD.
You can now breeze through checkout on millions of eligible apps and merchant websites. As long as you actively use PayPal, we'll automatically log you in unless you turn this setting off. For your security, we'll always ask you to log in whenever you update your personal or financial information, or if you haven't used this feature in a while.
This feature isn't recommended for shared devices. If you share this device with a family member or friend or are using a public computer, we recommend turning this feature off to avoid unauthorised purchases:
- Log in to your PayPal account
https://www.paypal.com/au/webapps/mpp/home?utm_source=unp&utm_medium=email&utm_campaign=PPC040707&utm_unptid=a1f3aafb-a495-1e1a-9c4f-b8750c3966bb&ppid=PPC000397&cnac=AU&rsta=en_AU&cust=WU77GKRAXFZXQ&unptid=a1f3aafb-a495-11ea-9c4f-b875c0396b6b&calc=a8858777991bf&unp_tpcid=remember-me-enabled-notification&page=main%3Aemail%3APPC004711%3A%3A%3A&pgrp=main%3Aemail&e=cl&mchn=em&s=
+ci&mail=sys&xt=104037 .
- Click the "Settings" icon next to "Log out".
- Click "Security".
- Click "Update" next to "One Touch for auto login at checkout".
- Click "Opt out" per device or "Turn off One Touch??? on all devices".
...
From this I can at least deduce:
How serious is this? I had something like this set up already. But is it the same? I don't know, but it's concerning, especially since it was sent before I got round to changing anything. But another similar message arrived later.
But back to Yvonne's socks. Off to fill out the form and finally pay for them. Hang:
|
|
|||||||||
What does CLAMP mean? Some kind of error, possibly an inappropriate reference to a wheel clamp? And what about this at the bottom left?
|
|
|||||||||
What's a keidson? But after a minute or two I got:
|
|
|||||||||
Is this URL kosher? It looks very much as if we were transferred to an external site, though the URL starts with paypal.com: https://www.paypal.com/webapps/hermes?flow=1-P&ulReturn=true&token=EC-82P64788W31433305&useraction=commit&country.x=AU&locale.x=en_AU#/checkout/login
So: did my password get compromised again? Off to change it again (and yes, that was really the password I used at the time). Send email to the seller, but that didn't work either:
I'm sorry to have to inform you that your message could not
be delivered to one or more recipients. It's attached below.
<ncservice1@buyeasyservice.com>: host mxbiz1.qq.com[203.205.232.191] said: 550
Ip frequency limited.
http://service.exmail.qq.com/cgi-bin/help?subtype=1&&id=20022&&no=1000725
(in reply to end of DATA command)
I've never seen that before.
So what's going on? My guess is that the payment was passed to the seller's system, and that it's broken in more than one way. In all probability there was no compromise, but how can you tell? Certainly it's an indication that this whole setup is more complicated than is acceptable: how can you tell what has happened? PayPal can't even write their documentation intelligibly.
| Wednesday, 3 June 2020 | Dereel | Images for 3 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Completing the sock purchase
|
Topic: technology, general, opinion | Link here |
So what went wrong with Yvonne's attempt to order socks yesterday? Her mail to the seller bounced, but their system had obviously noted the problem and sent her email reminding her of the purchase—along with a $6 rebate voucher.
OK, try again. Yesterday's URL was clearly old and worn out, and that was something they hadn't expected:
|
|
|||||||||
But retrying brought me further, including additional inapplicable but required information:
|
|
|||||||||
|
|
|||||||||
And the voucher (“only valid for today”) was still applicable, so in the end we saved the $5.99 postage. It worked! And the receipt from PayPal explained another mystery:
Subject: Receipt for your payment to keidson.
So the company variously refers to itself as keidson, ACCFC-COM, buyeasyservice.com and stromore.com. No wonder we're confused: they seem to be too. Once again, it seems, an application of Hanlon's Razor.
|
Rice in new saucepans
|
Topic: food and drink, opinion | Link here |
I cook rice in bulk and freeze it in portions. Today was a day, not helped by the fact that the stainless steel saucepan I use for rice was already in use. OK, what are the new saucepans for? Not for that; at least that was my intention. But why not?
It proved to be very successful. Rice does have a tendency to stick to the bottom of a saucepan, and the non-stick coating proved very effective. I wonder how it will go with nasi lemak.
But one thing hit me. The projecting handles:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
They're not in the way on the stove, but they'll take up significantly more space dish washer. Something to think about next time we buy cookware.
In passing, this time the rice:water ratio was 1:1.45. I think I can go to 1:1.5 next time.
| Thursday, 4 June 2020 | Dereel | Images for 4 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Winter flower bouquet
|
Topic: gardening, general | Link here |
Sadly, only 3 days after the beginning of winter, we have no roses worth picking any more; hopefully some will come. But we have a few gladioli, valerian and lavender:
|
|||||||||||||||||||||||
|
Dishwasher issues confirmed
|
Topic: food and drink, general | Link here |
Somehow I've been doing a lot of cooking lately. Today I filled two dishwashers and still had things left over. And as suspected, the handles of the new saucepan took up mre space than they should have:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
Next teevee hang
|
Topic: technology | Link here |
Once again teevee hung this evening, again with firefox; why can't it do it during the day when I have more time? And maybe it's time to stop using firefox until that issue is solved. So I took the opportunity to explicitly add to xorg.conf:
Section "InputDevice"
Identifier "Mouse0"
Driver "mouse"
Option "Protocol" "auto"
Option "Device" "/dev/sysmouse"
Option "Emulate3Buttons"
Option "ZAxisMapping" "4 5 6 7"
EndSection
According to the man page that's the default, so it shouldn't be necessary and shouldn't make any difference. Certainly it made no difference, and there was no mention of it in the log file.
FOOL! That's a Boolean, so it requires an argument. I should have written:
Option "Emulate3Buttons" "True"
Where “True” can also be spelt “1”, “on” or “yes”. Next time. Maybe I should set up a second machine to test with without interfering with teevee. And in the meantime I should probably use chrome instead.
| Friday, 5 June 2020 | Dereel | Images for 5 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Chilis ready for harvest?
|
Topic: gardening, food and drink, opinion | Link here |
The chilis I planted in October got off to a rocky start, but they've grown well and borne many fruit:
|
|||||||||||||||||||||||
Time to harvest them? I had thought so, but why? Once I harvest them they'll start to wither, or I'll have to freeze them. They'll keep longer on the plant, and it seems that even the mild frosts of the last few days haven't reached their relatively protected position.
A real question, though, that applies to nearly all garden vegetables: why bother? This plant probably cost me $5, required attention, and has now borne about $5 worth of chilis that are hotter than I really like. If I'm going to grow chilis (and tomatoes) they should be varieties that I can't easily get in the shops.
|
Still more teevee mouse problems
|
Topic: technology, opinion | Link here |
OK, I misentered the Emulate3Buttons entry in my X configuration yesterday. Time to fix that, and also try running X without moused. While I was at it, also worked around the bug in the loader scripts:
--- /boot/lua/drawer.lua 2020/06/05 05:11:37 1.1
+++ /boot/lua/drawer.lua 2020/06/05 05:11:43
@@ -145,9 +145,6 @@
end
local function defaultframe()
- if core.isSerialConsole() then
- return "ascii"
- end
return "double"
end
Also some research into starting processes at boot time. Years ago I simply put commands into /etc/rc.local, but somewhere along the line I found out that that was an old, worn-out magic word. The alternative seems to be to embed them in the rc startup scripts. The man page gives a minimal example, from which I've removed the empty lines:
#!/bin/sh
#
# PROVIDE: foo
# REQUIRE: bar_service_required_to_precede_foo
. /etc/rc.subr
name="foo"
rcvar=foo_enable
command="/usr/local/bin/foo"
load_rc_config $name
run_rc_command "$1"
Somehow that's a whole lot more effort than what I put in /etc/rc.local:
(cd /home/grog/src/weather/WH-1080-teevee && ./dorun)
Fortunately that should still work, according to the man page.
Next, kill X and all three mouseds, and restart from an xterm on eureka. Blank screen! Tried to put an xterm on it. It didn't fail, but it didn't display either. What does ps say?
=== root@teevee (/dev/pts/6) /var/log 24 -> ps aux | grep Xorg
root 64152 0.0 1.2 6398656 46108 7 S 15:16 0:01.45 /usr/local/bin/Xorg :0 -listen tcp -auth /home/grog/.serverauth.64135
root 64256 0.0 1.2 6394268 45636 7 I 15:17 0:00.56 /usr/local/bin/Xorg :1 -listen tcp -auth /home/grog/.serverauth.64239
How could I possibly have started two servers? I just ran the same command as in the startup scripts:
DISPLAY=:0 startx -- -listen tcp
How can that start 2 servers? But it did, and what I saw was server 1. Server 0 was running normally.
As if that wasn't strange enough, round this time I somehow had finger trouble on eureka and typed X into a shell prompt. The screens all went blank. Panic? Crash? After some personal panic (eureka has been up for 268 days, something of a record in recent times) discovered that I could still contact it from lagoon, and that it had started a new server. :2? I already had :0 and :1 running. No, it seemed to somehow have usurped the :0 server number, and when I tried to start a client, it went there—until a little later, when the server spontaneously stopped and server :0 returned to normal. I've been using X for over 30 years now, but I've never seen anything like that.
Back to teevee. Switched to server 0. Yes, the mouse works without moused. And /var/log/Xorg.0.log now confirms that it has enabled Eumulate3Buttons:
[1697249.116] (II) Using input driver 'mouse' for 'sysmouse'
[1697249.116] (**) sysmouse: always reports core events
[1697249.116] (**) Option "Device" "/dev/sysmouse"
[1697249.116] (==) sysmouse: Protocol: "Auto"
[1697249.116] (**) sysmouse: always reports core events
[1697249.116] (==) sysmouse: Emulate3Buttons, Emulate3Timeout: 50
[1697249.116] (**) sysmouse: ZAxisMapping: buttons 4 and 5
[1697249.116] (**) sysmouse: Buttons: 5
[1697249.116] (**) Option "config_info" "devd:sysmouse"
[1697249.116] (II) XINPUT: Adding extended input device "sysmouse" (type: MOUSE, id 7)
[1697249.116] (**) sysmouse: (accel) keeping acceleration scheme 1
[1697249.116] (**) sysmouse: (accel) acceleration profile 0
[1697249.116] (**) sysmouse: (accel) acceleration factor: 2.000
[1697249.116] (**) sysmouse: (accel) acceleration threshold: 4
[1697249.117] (II) sysmouse: SetupAuto: hw.iftype is 4, hw.model is 0
[1697249.117] (II) sysmouse: SetupAuto: protocol is SysMouse
[1697249.117] (II) config/devd: device /dev/ums0 already opened
[1697249.117] (II) config/devd: device /dev/ums1 already opened
I had started X from an xterm on eureka, not exactly the normal way, and it had pointed out that some of the path names in the startup scripts assumed that I was in my home directory. OK, stop X, try again the way I had intended (login on /dev/vty0). Still no Emulate3Buttons. Damn.
OK, I have another trick up my sleeve. Let's try xmodmap. It can't enable Emulate3Buttons, but it could reassign, say, button 8 on the Logitech M705.
=== grog@teevee (/dev/pts/8) ~ 13 -> xmodmap -e "pointer = 1 2 3 2 2 2 2 2"
Warning: Only changing the first 8 of 20 buttons.
X Error of failed request: BadValue (integer parameter out of range for operation)
Major opcode of failed request: 116 (X_SetPointerMapping)
Value in failed request: 0x2
Serial number of failed request: 9
Current serial number in output stream: 9
Huh? That's not very clear, but is it maybe complaining about the fact that I have assigned logical button 2 to more than one physical button? I'm sure I've seen that that should be possible. Still, try it out...
=== grog@teevee (/dev/pts/8) ~ 15 -> xmodmap -e "pointer = 1 0 3 2"
Warning: Only changing the first 4 of 20 buttons.=== grog@teevee (/dev/pts/8) ~ 16 -> xmodmap -pp
There are 20 pointer buttons defined.
Physical Button
Button Code
1 1
2 0
3 3
4 2
5 5
6 6
7 7
...
OK, that looks more like what I'm looking for. But when I tried it out, the scroll wheel went crazy; presumably button 4 maps to scroll wheel up. Certainly when I turned the wheel in that direction, it pasted junk all over the place. And scroll wheel down (5?) did nothing. But I didn't think of that until later.
That's enough for one day. Stopped X, unplugged and replugged the USB receivers for the mice to restart moused, and back to the last level of pain.
What next? Back to the old kernel to see if that makes a difference? Or could the 50 ms Emulate3Timeout be an issue? I could increase it, of course.
In passing, reading the logs I saw:
[fvwm][menustyle_parse_old_style]: <<DEPRECATED>> The old MenuStyle snytax has been deprecated.
Is that a snytax or a semantics error?
| Saturday, 6 June 2020 | Dereel | Images for 6 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Avant Stellar?
|
Topic: technology, opinion | Link here |
Eleven years ago I finally gave up trying to maintain my 20-year-old Northgate OmniKey keyboards. Some undocumented time previously I had bought a couple of Avant Stellar keyboards but found them hard to use, and 8 years ago I replaced them with the keyboard I still use, a Sun Type 7.
I hadn't thought much more of the Avant Stellar keyboards until I received mail from Kari Eveli last year, offering software for the keyboard. What can I do with it? Clearly it's a good idea to put it online somewhere, but all the software I have (in my src directory, and deliberately not displayable) is my own. No objection to putting up other software, but where? That ended up in the “too hard” basket.
And then I received email from Roberto Jordan, who is looking for an Avant Stellar keyboard. OK, finally got round to looking for them, and how about, I still have both of them:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
What are they worth? How would I know? Make me an offer.
Only later, via a Google search, did I (re)discover that I had written a web page on the subject. So that's why people contacted me. I also discovered that one of the keyboards (who knows which one?) had a defective Q key. But which one?
I suppose the real question is: why Avant Stellar? Reading back, it's surprising how much I disliked the keyboard. Apart from its similarity to the OmniKey, the biggest attraction appears to be the row of function keys to the left of the main keyboard. But the Sun Type 7 has that too.
In passing, it's interesting to note that a company called Northgate Keyboard Repair repairs both OmniKey and Avant Stellar keyboards.
| Sunday, 7 June 2020 | Dereel | Images for 7 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
New Asian foods
|
Topic: food and drink, opinion | Link here |
Over the course of time I have bought a number of spice mixes for Indian and South-East Asian foods. I've left many of them to expire because they proved to fit in the “too hard” basket. Today I tried a couple: Shan Lahori Fish (“Shan Lahori Fish Mix has a perfect blend of rich and aromatic spices which brings you the exotic taste of Fried Fish from the streets of Lahore”) and one of four different Rendang preparations that I had.
The Lahori fish is apparently a batter. “Mix contents of package with ½ cup water” for 1 kg of fish. OK, nothing like being careful, apart from the fact that there was no way that we were going to consume 1 kg of fish. Choose 200 g flathead and try to make 20% of the content. The powder was simple: 20% of 100 g are 20 g. And 20% of a half cup? Let's say 24 g water. But that was too watery. In the end I ended up with 40 g of each, and it was still far too watery to put in a friteuse, not helped by the lumpy nature of the “batter”.
|
|||||||||||||||||||||||
Nothing for it. Mix the stuff together and fry in a pan:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That worked, but the result was disappointing. Only later did I find a video showing what to do:
That doesn't look much better.
The rendang was from Dollee. It was mainly interesting because it was so simple: fry 1.2 kg meat with paste for 10 minutes, add 100 ml water and cook for another 15 minutes, add 100 ml coconut milk and warm up again. That's barely enough to cook beef. And this for a dish that owes its flavour to hours-long slow cooking! Tried it out, using 500 g/40% of the paste, and to my surprise it produced what looked like quite a good, consistent gravy after the first 10 minutes:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
It certainly didn't need any water, though I added 50 g of coconut milk later on. It's interesting to note that Dollee's recipe online now specifies 50 ml of water instead of 100 ml. But both are too much. Cooked very slowly for 2 hours, and it was still firm. Why?
Somehow the whole dinner today, which took me most of the afternoon, was disappointing.
|
More flash problems
|
Topic: photography, opinion | Link here |
Taking photos with the flash triggers is always a bit of touch-and-go. Today I took 14 photos of the cooking, of which only 10 were correctly exposed. On 3 occasions the flash didn't fire at all (wobble trigger in shoe), and on one occasion I got the dreaded half-photo:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Why does that happen? The flash finished before the shutter was completely open, which means that it must have started firing before the shutter opened. On the other hand, the following image (not shown here, for obvious reasons) was one of the shots where the flash didn't fire at all. And of course, this time the shots were at 1/250 s, thus disproving the answers of some people to my question that I got on the Facebook M43 Tech Talk. But is there any significance in that relationship? I still can't think of one.
| Monday, 8 June 2020 | Dereel | Images for 8 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Phở?
|
Topic: food and drink, opinion | Link here |
Another paste that I've been meaning to try for some time is labeled “Phở; Vietnamese beef flavor paste”. OK, the devil is in the detail: “flavor” [sic] suggests that there's no beef in it at all, though they don't claim that it's vegetarian. But the other thing, possibly more important, was “Made in Thailand”.
|
|||||||||||||||||||||||
OK, today's the day. “Put one spoonful of paste in 3 cups of water”. How much is that? Somewhere between 3 and 30 ml of paste in between 70 and 285 ml of water, a range of 40:1.
|
|||||||||||||||||||||||
In the end I put 30 g of paste into 400 ml of water, and that seemed acceptable, possibly a little too much. But it has chili in it! Why do all these pastes have more chili in them than necessary? My understanding is that Phở has none at all.
It didn't taste too bad. I've come to the conclusion that with all these pastes you shouldn't look so much for a faithful fulfilment of their claims as something interesting that you can use in a dish. But it would be nice to find a way to make real Phở without too much difficulty.
|
Trump gets his wall
|
Topic: politics, general, opinion | Link here |
One of the most stupid things that Donald Trump set in motion was a wall around the USA. Well, maybe a fence. Well, maybe not the entire USA, just the parts that prove dangerous.
Finally it's there:
|
|||||||||||||||||||||||
Protecting the leader of the land of the free and the home of the brave.
|
Nele Kömle in town
|
Topic: general, opinion | Link here |
We haven't seen much of Nele Kömle in the last few years, since she moved much further away. But she visited Chris Bahlo today, so we went by to say hello. She had both children with her—Nelson must now be over 5, and he was clearly impressed by the mediaeval atmosphere:
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
|
Walking to the Big Oak
|
Topic: animals, gardening | Link here |
When we lived in Kleins Road, we frequently walked our dogs down the extension of Kleins Road through the forest to the junction of Swansons Road and Cahills Road. The other end of Cahills Road is on the Rokewood Junction Road, just opposite Chris Bahlo's property. So when we were done with our visit, we walked the dogs the length of Cahills Road to the junction, which we had christened „Dicke Eiche“ (“Big Oak”); in reality there are three straggly Eucalypts there. Here a photo from last year:
|
||||||||||||||||||||||||
On the way, saw what might be the first Epacris impressa of the season:
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
|
Pearl mushrooms
|
Topic: food and drink, opinion | Link here |
Yvonne found some pearl mushrooms, a kind of oyster mushroom, in town last week and brought them home. They're interesting looking things:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Unfortunately, they don't have a very pronounced taste. I used about half of them in an improvised noodle dish, but we hardly noticed them. I'll have to consider what to do them if we buy them again.
| Tuesday, 9 June 2020 | Dereel | Images for 9 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Understanding flash problems
|
Topic: photography, opinion | Link here |
Why am I continually having misfiring flash triggers? I've been assuming that there were contact problems, but that bears testing. The cameras make contact between a pin in the middle of the flash shoe. There may be others for camera-specific flash functions, but there's always the one in the middle, and it's the most important. Here the foot of a flash trigger and of the Meike MK-300:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
One reason might be low contact voltage. How do the triggers compare with my flash units? I had already had difficulties with the MK-300, which seemed to work, but which wouldn't trigger. Tried it out with Nickel-zinc batteries instead of the usual NiMH batteries, and bingo! It worked. So there's clearly something to the contact voltage hypothesis. The other thing is the nature of the contacts themselves. On the flash units, the metal plate is the negative pole of the contact, but there is none on the triggers. They're hidden on the side:
|
|||||||||||||||||||||||
Could that be part of the problem? I thought so, but then I discovered that the Olympus STF-8 also has contacts only on the side, like on the triggers. OK, there's two halves to a contact. What do the cameras do? Here the OM-D E-M1 Mark I, the E-M1 Mark II and the E-M5 Mark III:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The E-M5 Mark III is the clearest, since it's the newest and also has hardly been used with flash at all. The contacts are on the side of the shoe. So any poor contact with the triggers would be due to inadequate pressure between the shoe and the contact on the trigger, or due to dirty contacts on the side; clearly cleaning the base of the shoe isn't enough.
Brought out all my triggers and modern flashes—the mecablitz 40 CT 4 didn't really count, and we know it has a voltage round 150 V—and measured them. The voltages ranged between 3.05 V and 4.9 V, the flash triggers being the lowest.
OK, that's reasonable. How de you get even 3.2 V out of a 3 V Lithium battery? Use an inaccurate multimeter? I have two different brands of trigger, iShoot and Wansen. They're compatible and look identical from the outside:
|
|||||||||||||||||||||||
Are they the same inside? I was pretty sure that they were going to be, but in fact they have quite different circuit boards:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
In retrospect, that makes sense: the Wansen triggers have an on-off switch on the side, clearly an advantage. But the batteries are 12 V! So why is the trigger only generating 3 V? Measured them all:
| Unit | Battery | Contact | ||||
| voltage | voltage | |||||
| 1 | iShoot | 3.5 | ||||
| 2 | iShoot | 11.2 | 3.05 | |||
| 3 | Wansen | 3.2 | ||||
| 4 | Wansen | 3.2 | ||||
| 5 | iShoot | 3.2 | ||||
| 6 | Wansen | 0.4 | 0 | |||
| 7 | Wansen | 0 | ||||
| Meike MK-300 | 3.8 | |||||
| mecablitz 15 MS-1 | 4.26 | |||||
| Viltrox JY-670 | 4.9 | |||||
| Olympus STF-8 | ||||||
The batteries in triggers 6 and 7 were non-functional. The obvious benchmark is the STF-8, but I couldn't test it: the (rechargeable) batteries were also flat, all of them. Three were so completely discharged that I couldn't recharge them, voltages under 100 mV. How could that happen? Maybe I forgot to turn off the power the last time I used it, but it should have stopped drawing power long before destroying the batteries. Clearly it had misbehaved; maybe I didn't forget to turn it off, and that it's a bug in its power management. It wasn't my favourite flash unit before this event, but that's another black mark.
|
Panic!
|
Topic: technology, opinion | Link here |
I've been wondering what to do about teevee, and making do with the mess that it's in at the moment. One thing was obvious: the hangs always occurred when I was doing something with firefox. Well don't do that, then: use Chromium. So I've been doing that for about a week, without making friends with it greatly. It reluctantly accepted my settings to make it behave like a normal X client and not a window manager, but I still can't get it to fill the screen correctly. That's something that people don't do in the Microsoft space, of course, but that's no reason for it to misbehave under X.
Today, however, things were different: in a potentially similar situation to where firefox hung, the system paniced! That's probably not Chrome's fault: I suspect that the problem is in the nvidia driver. But it shows that Chrome can't work around the problem.
Took a look at the dump, but I couldn't even get a backtrace. All values appeared to be 0. Is that an issue with the dump or with the tools? Looking at gdb(4), the only significant difference made since 2004 was to replace a quotation which the committer didn't like. I suspect that more and more parts of FreeBSD are atrophying.
| Wednesday, 10 June 2020 | Dereel | Images for 10 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Bloody dogs!
|
Topic: animals, general, Stones Road house, opinion | Link here |
Out as usual for a walk with the dogs today, Nikolai on the line, Leonid running free. And he saw a dog in the distance, towards Progress Road, and ran off in pursuit, ears set to passthrough.
Damn! What do I do now? First, get rid of Niko. Put him back inside the property, shut the gate, and off to see what I could do. Niko too: he pushed his way between the gate and the gatepost. I didn't think he could do that. So then I had the worst case scenario, two unleashed big dogs chasing a small black dog too far away for me to do anything about it.
Somehow things untangled, and the dogs shot past me. I managed to hit Niko with his line roll, but it didn't stop him. The owner managed to stop a car and a couple of other people joined in the fun, but in the end the dogs came back by themselves, having not quite made it to Bliss Road. I off to talk to the owner before it occurred to me that I should get rid of the dogs first. Did that, off in the car in the direction the dog and owner went, but couldn't find them. Tried asking at the house on the corner of Bliss and Stones Road—the gate was open, unusually—but there was nobody home.
Damn. Once again it's “keep the dogs on the line at all times”. But I would at least have wanted to apologize and see if the dog was hurt. Leo was, slightly:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That proved to be a barely noticeable wound on a hind paw. Did the other dog do that? Serves him right if she did, but there could be other reasons.
So: how do we find the owners? In the end they found us, just as I was about to set off looking for the collar chain that Niko lost during the chase. Husband not amused, not prepared to talk, threatened multiple times to shoot my fucking dogs. I tried to talk sense to him, offered to take the dog for a vet check—they said that she was limping—but he kept repeating his threats. I warned him that if he continued, I'd report the matter to the police. They were in a car, so it would be easy to track them. On the third occasion I gave up and left, telling them that I would call the police.
That in itself was interesting. How do you contact the police? The emergency number is 000, but this wasn't an emergency. There's no police station in Dereel, of course. Google told me that the Ballarat police station had the number (03) 5336 6000, so called there, and after negotiating their hard-to-understand menus, was connected with Joanne, who was sympathetic, took some notes, and told me that they weren't responsible for our area. That was Rokewood, (03) 5346 1342, who would probably be closed, but the phone would redirect. She also gave me a reference number for the call.
That was the case; it redirected to Bacchus Marsh, 90 km away (three times the distance of Ballarat). Got connected to a Constable Petri, who took typical details (what did the man look like? I didn't pay much attention) and said that somebody would sort things out. No reference number, just his name.
Then another car arrived. Two men, one of whom proved to be the owner of the dog, now very apologetic, and his partner had told him that maybe I misunderstood his final threat: it was really to say that he would not shoot the fucking dog. Had a much more level-headed discussion, in which it eventuated that they're the people who built the new house in Harrisons Road, and their names are Daniel and Alicia. And I lost my bet: they had moved in in late February, not Anzac Day as I had guessed. All in all a much better end to the matter, though clearly it would have been better if it hadn't happened.
OK, call back to the police. I didn't have the number, since I had been redirected. https://www.police.vic.gov.au/local-police-stations helps: Address: 117 Main St, Bacchus Marsh VIC 3340, Phone: (03) 5366 4500. There another person (who didn't identify himself) took my notes and said he would pass it on to Constable Petri, who would call me back if he needed any further information. Somehow I was left with the feeling that they weren't very organized.
So why did this happen in the first place? Our dogs have never been aggressive. Is it the lack of social interaction? Clearly they'll stay on the leash at all times now.
|
Fixing the gate
|
Topic: Stones Road house, general | Link here |
I've been trying different things with the closing angle of the gate, but today's experience with Nikolai made it clear that we need to do something soon. There isn't much hope of getting a clean fit for the adjustments on the mounting bracket. So let's adjust it to the next step towards closed (which would make it over-closed) and adjust the end switch.
Out to do that. Yes, though there's not much space, it's possible to adjust the end switch relatively well. But there's a very big step between the adjustments on the bracket, and the next one wouldn't open sufficiently.
What do I do? The real problem was that the installation instructions are wrong. I should have drilled the holes in the gate with the opener almost open. Then I could have adjusted the switch to get it almost right. Now I have the choice of drilling another hole in the adjustment bracket, or repositioning the screws holding the opener to the gate. Which is better? To be considered.
|
Wearing braces again
|
Topic: general, opinion | Link here |
When I went to school in England, I was required to wear braces for my trousers, not these newfangled belts that the US Americans wore. Elsewhere, of course, I used belts on my pants. But lately I've noted that I'm continually pulling my pants up. That wouldn't happen with braces. So today Yvonne bought me a pair of braces, and indeed so far they seem to be doing a better job. Another indication that I have become an old fogey?
|
Spring meets autumn
|
Topic: gardening, opinion | Link here |
Our first (very) timid Jonquil of winter is flowering:
|
|||||||||||||||||||||||
And the outdoor Hibiscus rosa-sinensis “Uncle Max” is still flowering and producing buds:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
We've had a number of frosts over the last week, though my claim is that they didn't get into the bed where the hibiscus is located. So I'll keep a particular eye on this one.
|
teevee: reinstate old kernel
|
Topic: technology, opinion | Link here |
So I've tried just about everything to get the mouse working correctly on teevee after last month's kernel build. Was it really the kernel that made the difference? Reinstate the old kernel and compare.
Yes! Now the mouse works normally again. But what caused the difference? Where do I even start looking? It may be related to /dev/input, but even there I can't find any useful documentation.
My experience with upgrading teevee in the last three months has been horrible. There were so many problems that I can't even start describing them. I'll have to go back and write them all up. But where do I go from here?
| Thursday, 11 June 2020 | Dereel | Images for 11 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Revisiting ancient X clients
|
Topic: technology, opinion | Link here |
Now that Facebook has supplanted email, Yvonne doesn't look at her mail much any more, and missed a message I had sent her earlier today.
There's a solution for that, an X client that displays a US-American style mailbox with a flag, and lifts the flag:
|
|
|||||||||
|
|
|||||||||
I had installed it 30 years ago when email was something that didn't come too frequently, but I removed it again after I was getting more than 200 messages a day. But maybe times have changed enough that it might work for Yvonne.
But what was it called? I recalled bip, but that's an (old, worn-out) IRC proxy. It took the combined efforts of several people on IRC and a couple of Google searches to rediscover it: xbiff, apparently part of X itself that doesn't get installed automatically (at least not on FreeBSD). OK, pkg install should do the trick.
Oh:
New packages to be INSTALLED:
avidemux-qt5: 2.7.4
xbiff: 1.0.4
Why should xbiff depend on avidemux? My guess is that it doesn't, but pkg is still upset that I rebuilt avidemux with non-standard dependencies.
OK, let's build it from source. That worked fine. And Juha Kupiainen came up with the objection:
* juha gets notifications with xfce4 !!
Ah, the Unix way: do one thing, and do it well. xfce is a “Desktop Environment”, and at the very least it would require relearning the habits of decades. It's also small and lightweight, which corresponds to:
# pkg install xfce
The process will require 178 MiB more space.
31 MiB to be downloaded.
178 MB! That's 5 times the size of the (already bloated) kernel, and over 5,000 times the size of the Research Unix kernels. By contrast, xbiff is:
-rw-r--r-- 1 root wheel 10,940 16 May 07:49 xbiff-1.0.4-1f285059cb.txz
O tempora! O mores!
|
Update your diary for you
|
Topic: technology, general, opinion | Link here |
I'm not getting as many requests for links from my diary as I once did, but this one got me today:
Date: Wed, 10 Jun 2020 23:13:28 +0000
From: Tara Worsham <tara@onlinesafetykids.com>
To: "groggyhimself@lemis.com" <groggyhimself@lemis.com>
Subject: Something is missing on lemis.com
My name is Tara and I want to say that your page
http://www.lemis.com/grog/diary-jul2017.php?subtitle=Still more mouse
problems helped me a lot!
So first of all - Thank you! When I tried to work with a VPN I got
TONS of errors (It was annoying!).
I asked on several Facebook groups for a solution and got this great
guide:
https://www.wizcase.com/blog/avoid-netflix-proxy-error-and-streaming-error/
https://www.wizcase.com/blog/avoid-netflix-proxy-error-and-streaming-error/
- This was the only solution that worked.
OK, what's that URL? It's truncated, but clearly referring to this article. But that's about problems with my mouse (which had been ongoing for 16 years at the time, and now, 3 years on, have still not been resolved). How can that help Tara? It doesn't even help me.
But she was talking about VPNs, and the link was to a VPN page. What earthly relevance is that? I found it strange enough that I replied and asked, but for some reason I didn't get an answer.
And yes, I checked the entire month's diary, and there was no reference that would have explained the message.
|
Making amends for the dogs
|
Topic: animals, general | Link here |
Yvonne thought we should bring a little something to Daniel and Alicia to say sorry for yesterday's incident, so off to their house, where we found Daniel and Pippa, the dog. Pippa came running up to greet me, but then she recognized me and retreated to the safety of Daniel. Clearly it'll take a while for her to recover.
| Friday, 12 June 2020 | Dereel → Ballarat → Dereel | Images for 12 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More shopping
|
Topic: Stones Road house, general, food and drink, opinion | Link here |
How do I approach fixing the gate opener problems? For the time being I have a rubbish bin in front of the gate to keep the dogs in:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
But clearly that's not the solution. I have two basic approaches: reposition the gate end of the opener on the gate, or drill a new hole in the adjustment ring. Either way I need tools, and since moving to Stones Road I can no longer find my old drills and spanners.
Never mind, they were old and flaky anyway. New tools don't cost that much money. So off to Ballarat to buy some.
On the way dropped in at Harris Scarfe to look for more kitchen utensils. I was quite happy with the pans I bought last week, but I hadn't thought to look for something smaller. Currently I'm still using pots that look like this:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That's a 14 cm pot in the first image, just what I need for a single portion of laksa. And this morning I had finally found an application for the steel scouring pads that I found last month: cleaning the remains of poached eggs off the second one. It took me 5 minutes! Replacements can't be that expensive.
Spent something like 30 minutes at Harris Scarfe and left with nothing. They had absolutely no 14 cm saucepans. And the cheapest 16 cm saucepan that I could find cost $35, half the price of the four pots that I bought last week. After looking around the whole sales area, I found neither the kitchen knives I was also thinking of, nor anything else of use. It seems that the one offer that I bought was really about the only really good one. Here a comparison:
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
|
|
||||||||||
There are more pots in the second special offer, but they're much smaller, ranging from 16 cm (two of them) to 26 cm, finishing where the other one started. And the price (“$699.95”) was still $279.95, four times the price of the set that I bought. Clearly I was misled by their offer last week.
On the way, went through Central Square food court, where the signs of the “lockdown” are still evident:
|
||||||||||||||||||||||||
Of the 7 food suppliers, only 4 were open, and judging by the clientele, they're probably wondering why they're open.
On to Bunnings, where I managed to fight my way past strange units to find the drill bits and spanners that I needed:
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
Then, on Petra Gietz' recommendation, to the Kmart in the same area (Delacombe Town Centre). I hadn't been there since they opened nearly 3 years ago—so long, in fact, that I had forgotten that I had been to Kmart (and, as it happens, bought a frying pan). Once again I had to marvel at the appalling traffic flow. Getting out of the Bunnings area requires you to find an unmarked, very narrow exit, which leads you out away from town. How do you turn around? There's no obvious place. And on the other side, they still route all traffic through the middle, over the pedestrian crossings. There are parking areas, but they're not marked.
At Kmart I found some pots that were cheaper, and ultimately ended up with a 16 cm saucepan that cost $15. In proportion to the set I bought at Harris Scarfe last week, that's still relatively expensive. I think that says more about the price of the set than about the price of the saucepan. And they still haven't improved their signage. You need good luck to find the aisle leading to the cash registers.
|
Hot water failure!
|
Topic: general | Link here |
The nights have been cold for the last week. My weather station reported temperatures as low as -0.5°. And Petra Gietz has had her problems as a result: at least one of her hot water solar panels has been damaged by frost. We've seen that before only shortly after moving into the house. They solved it at the time by replacing the panels with evacuated tubes. And since the failure and subsequent replacement of the hot water system itself, they're now hanging around on the roof, unused. Can she use them?
On the way home, dropped in at Petra's to take a look. Yes, the hot water system looked pretty much like our old one:
|
||||||||||||||||||||||||
The panels were also leaking in a similar manner. Interestingly, Petra told me that she, too, had frequently had to remove the panel on the heater and reset the circuit breaker, like I had had to do. Under the circumstances it's worth wondering whether to replace the panels or the whole system.
|
Panic!
|
Topic: technology, opinion | Link here |
Another panic on teevee this evening, again related to web browsing. This is becoming all the more common. Is there maybe something wrong with the hardware? Time for a backup system, anyway. But that will require rearranging the storage. Currently teevee has 8 TB of video storage. Should I maybe really be thinking about a NAS? Somehow that's the diametric opposite of fast SSDs. But for this application it would be more then sufficient.
| Saturday, 13 June 2020 | Dereel | Images for 13 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Burning books
|
Topic: history, music, opinion | Link here |
Fifty years ago today I had an interesting experience: I walked in to Colston Hall in Bristol with my girlfriend Sue Fortescue to listen to a rehearsal of the London Symphony Orchestra playing Beethoven 9 under Colin Davis. No admission checks, nothing. Probably we were entitled to enter—Sue was a member of some orchestra or another, probably the National Youth Orchestra, and an ex-boyfriend of hers was singing in the choir, but nobody even paid attention to our presence.
Colston Hall is located in Colston Street in the heart of Bristol. Both the hall and the street are named after one of the founders of Bristol's wealth, Edward Colston. From the Wikipedia page, he was “a merchant, philanthropist, and Tory Member of Parliament”. A great man indeed. Over 170 years after his death he was still held in such high esteem that a statue of him was erected in the centre of Bristol, not far from Colston Avenue.
But Colston was also a slave trader. In the current politically correct atmosphere that outweighs all other aspects, and earlier this week protesters toppled the statue and threw it into Bristol Harbour.
Is that appropriate? You could rid Bristol of everything that reminds of Colston. As the usage of the name above shows, that would take some effort. But what about the wealth that it brought to Bristol? That, too, was tainted by the slave trade. Are people proposing to sell off parts of the city and return the money to the descendents of the slaves?
300 years ago morals were different. Colston was probably well-known as a slave trader because he saw no reason to hide the fact. Mercifully, things have changed, but there's no reason to forget the past. Rather than remove the statue, it would make more sense to make it part of an information centre explaining how it could happen that a slave trader could have come to be a respected part of Bristol society.
Things are even more interesting 100 km east in Oxford, where the façade of Oriel College contains a bust of Cecil Rhodes, an even better known figure with dubious background. He, too, should go, and the college has obtained permission to remove the bust. But Rhodes established the Rhodes Scholarship for people from territories under British rule, including “black” people. Remove all traces of Rhodes and you're not doing anybody a favour.
But one way or another, expunging references to people of dubious character reminds me of the book burning of extremist groups. If you don't agree with it, remove it. That's not a way to understand history. If you don't agree with it, document your disagreement. In the present cases that should be easy and effective. Oriel College has attempted it with this page, made no easier to read by the low contrast rendering.
|
Yet another ThinkCentre
|
Topic: technology, opinion | Link here |
In principle it should be easy enough to get replacement hardware for teevee: there are lots of ThinkCentres on sale on eBay. But as on other occasions, it's really difficult to compare what's on offer.
OK, I want a fast processor, which was the criterion when I bought the last two machines. I also want a PCIe slot for the graphics card, and on comparison I discovered that dischord, the Microsoft machine, doesn't have one. Will the eBay listings tell me? Not a hope.
So off, looking first at the processors. The ones that came into question were all Intel Core i5 processors: -2400, -3470, -4570 and -4570T. On the face of it, you'd think that this was an order of increasing throughput, but it proves that the 4570T (2.9 GHz) has a CPU mark of 3,293, compared to the 5,153 of the -4570 (admittedly at 3.2 GHz). And I had already established that the -2400 had a CPU Mark of 5,827.
But that's not what they're saying now. Not 5,827, only 3,786! Did they make a mistake? It seems so. What they're telling me now is:
| CPU Name | CPU Mark | Rank | “CPU Value” | Price (USD) | ||||
| Intel Core i5-2400 @ 3.10GHz | 3,786 | 944 | 126.25 | $29.99 | ||||
| Intel Core i5-3470 @ 3.20GHz | 4,667 | 825 | 116.67 | $40.00 | ||||
| Intel Core i3-4160 @ 3.60GHz | 3,461 | 1011 | 80.52 | $42.99 | ||||
| Intel Core i5-4570 @ 3.20GHz | 5,153 | 769 | 71.58 | $71.99 | ||||
| Intel Core i5-4570T @ 2.90GHz | 3,293 | 1048 | 16.47 | $199.95 |
OK, -4570 it is. There were two of interest, both model M93p. What does that mean? Off to the Lenovo site, where I found voluminous documentation, including a 186 page User Guide with this reassuring diagram:
|
|
|||||||||
Not one PCIe slot, but three, two of them 16x. Just what I need. So I ordered the cheaper one, not just because it was cheaper, but because it had better specs.
And the processor and processor speed? Ah, sorry, can't tell you that. This manual only has 186 pages, 50 of them safety information. No space to mention the microprocessor.
While I was at it, put the serial numbers of two of the existing ThinkCentres into the web site. How about that, they convert the gibberish on the labels into something informative:
It's about time I started keeping better records of this stuff. What I have today is:
| System name | Label | Serial | Model | |||
| dischord | MT-M 3132 A8M | PBKEX93 | M71e | |||
| lagoon | MT-M 7033 FE-1 | R8RRK4V | M91p | |||
|
Summer in winter
|
Topic: gardening | Link here |
Our Corymbia ficifolia is still flowering, a week before the winter solstice!
|
|||||||||||||||||||||||
And those buds on the left suggest it will carry on doing so for a while.
|
Two more grid power failures?
|
Topic: Stones Road house, general | Link here |
Another two grid power failures today, the first 7 seconds, the second 2 seconds.
Or were they? The inverter waits a minute after any grid anomaly before reconnecting, and the second failure occurred before the minute was up.
| Sunday, 14 June 2020 | Dereel | Images for 14 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Black lives matter?
|
Topic: politics, opinion | Link here |
I'm quite surprised how much traction the reaction to the death of George Floyd has been and continues to be. It's not as if he was the first person to die at the hands of an overly violent police force—the Washington Post has documented round 1000 shooting deaths per year at the hands of the US Police. Certainly high time for reform.
But why now? Why “Black lives matter”? Don't white lives matter too? Emphasizing black lives (the lives of people with not completely white skin) is discrimination, something that people should be trying to eliminate. Lives matter. The skin colour of the person embodying that life is completely irrelevant.
Yes, a disproportionate number of those killed were “African Americans” or “people of color” or some other euphemism. But they weren't all. George Floyd was killed by Derek Chauvin, a white officer of the Minneapolis Police Department. Three years ago a Minneapolis police officer shot Justine Damond dead for no obvious reason.
What's the difference? More than meets the eye:
None of this detracts in the slightest from the atrocity of the Floyd incident. And I probably wouldn't have known about Damond either had she not been Australian. But there are so many such deaths! In the ensuing demonstrations, several further people have been shot and injured, some on camera, such as the case of Martin Gugino.
So why now? It's not too soon. The US police has transformed itself into a paramilitary organization. My guess is that Mohamed Noor shot Damond because she didn't show the deference that US police expect. I had a similar incident in Oakland years ago, where a policeman pulled a gun on me because I approached him. It seems that the way I did it was a typical ambush situation.
So really we have two separate issues that are getting confused: US police violence and racial discrimination. We can't get rid of either, but it's good to understand that the two are different. And the solutions are different too.
One of the issues is clearly the impunity with which US policemen can use their weapons. On the few occasions where police officers are charged with causing death, they're usually acquitted. Otherwise, it seems, there is little
In passing, it's interesting to note what Lorraine Caranza said recently. She grew up in Manhattan in the 1940s and 1950s. And as she put it, policemen were people of authority and worthy of respect—as the Germans would say, my friend and helper. Why have things changed?
|
ThinkCentre serial numbers
|
Topic: technology, opinion | Link here |
Yesterday I found the details for two of my four ThinkCentres based on their serial numbers. They're written in a relatively small font, and my eyes are no longer the best. After one failure, decided to take photos of them:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
It turned out that I had transcribed them correctly, but the second is so old (about 12 years) that their system ran into trouble converting the serial number. That's the one that seems to be flaky, so it's good to know that it's the oldest. Also, as planned, updated my configuration page:
| System name | Label | Serial | Model | CPU | Speed | Memory | ||||||
| dischord | MT-M 3132 A8M | PBKEX93 | M71e | Core i5-2400 | 3.1 GHz | 16 GB | ||||||
| lagoon | MT-M 7033 FE-1 | R8RRK4V | M91p | Core i5-2400 | 3.1 GHz | 4 GB | ||||||
| teevee | MT-M 7483 P33 | R81HZK3 | M58p | Core 2 9505 | 2.83 GHz | 4 GB | ||||||
| MT-M 6073 BMM | L3AAX4D | M57p | ? | ? |
|
Still more flash problems
|
Topic: photography, opinion | Link here |
I took most of the ThinkCentre serial number photos with the mecablitz 15 MS-1 ring flash, and it seemed to work well. But for the last machine, which was not connected to anything, it seemed easier to use the studio flash in my office.
Easier yes, but:
|
|||||||||||||||||||||||
Why so underexposed? I had used my Sekonic L-308S exposure meter, and it showed the correct reading after that photo. Could it be because of the closeness? At 1:1 you need to open up 2 stops to compensate for the lens extension. But that's too much pain. Just use the mecablitz again and set it to f/6.3:
|
|||||||||||||||||||||||
And the exposure meter showed “f/5.6.5”, or about f/6.7. Pretty much exactly what I would expect.
Well. Looking at the Exif data, the mecablitz is set to +1 EV compensation. That was clearly necessary here. But that wouldn't explain the difference in exposure. What caused it? The exposure meter told me that the flash intensity was correct in each case, but the difference in exposure is obvious.
|
More debug problems
|
Topic: technology, general, opinion | Link here |
For some reason, since moving my weather station to teevee, I've been getting bogus rain readings, only on startup. It looked like an uninitialized variable, but after several attempts I still can't work out where. OK, put in some debug code:
/* XXX horrible kludge, fix! */
if ((readings->rain > 1.0)
&& just_started)
{
int *doom = 0;
*doom = 0; /* boom! */
}
just_started = 0;
That's a sure-fire way to get a core dump: deference a NULL pointer. I'm sure that there are other ways, but nothing like as simple as the CALL ABEND that we had in TAL. And sure enough, I got a core dump. And another. And another. It seems that it happened every time, tough I can't see how that can be. OK, replace with the old version and look at the dump.
=== root@teevee (/dev/pts/7) /home/grog/src/weather/WH-1080-teevee 22 -> gdb wh1080 wh1080.core
Core was generated by `./wh1080 Stones-Road grog eureka'.
Program terminated with signal 11, Segmentation fault.
Reading symbols from /lib/libm.so.5...Reading symbols from /usr/lib/debug//lib/libm.so.5.debug...Error while reading shared library symbols:
Dwarf Error: wrong version in compilation unit header (is 4, should be 2) [in module /usr/lib/debug/lib/libm.so.5.debug]
...
Reading symbols from /libexec/ld-elf.so.1...Reading symbols from /usr/lib/debug//libexec/ld-elf.so.1.debug...Error while reading shared library symbols:
Dwarf Error: wrong version in compilation unit header (is 4, should be 2) [in module /usr/lib/debug/libexec/ld-elf.so.1.debug]
#0 0x00000000002041e4 in ?? ()
(gdb) bt
#0 0x00000000002041e4 in ?? ()
#1 0x00007fffffffece0 in ?? ()
#2 0x0000000000000000 in ?? ()
(gdb)
Huh? What does that mean? And why didn't it read symbols from the executable?
=== root@teevee (/dev/pts/7) /home/grog/src/weather/WH-1080-teevee 23 -> file wh1080
wh1080: ELF 64-bit LSB executable, x86-64, version 1 (FreeBSD), dynamically linked, interpreter /libexec/ld-elf.so.1, for FreeBSD 12.1 (1201517), FreeBSD-style, with debug_info, not stripped=== root@teevee (/dev/pts/7) /home/grog/src/weather/WH-1080-teevee 24 -> gdb wh1080 wh1080.core
...
(gdb) file wh1080
Reading symbols from wh1080...Dwarf Error: wrong version in compilation unit header (is 4, should be 2) [in module /eureka/home/grog/src/weather/WH-1080-teevee/wh1080]
Off looking for answers. This page seems relevant, though it comes up with two different answers. But it seems that somehow gdb no longer works. All the object files were generated for the same version of FreeBSD—and isn't it one of the purposes of installing everything at once to specifically avoid this kind of compatibility problem?
|
Next steak and kidney pie
|
Topic: food and drink, opinion | Link here |
I've been trying for months to find the right cooking parameters for steak and kidney pie. Two months ago I came to the conclusion that I should bake them at a higher temperature (210°) without heat from above. Also put them as low as possible in the oven to ensure that they got cooked from below.
The result was still not what I expected. After 20 minutes the crust was still pale, so I turned on the top element and continued for another 5 minutes, after which I had:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Is that good enough? At least it was acceptable, but I would like to have one setting that I can leave running the whole time.
| Monday, 15 June 2020 | Dereel | Images for 15 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Chili harvest
|
Topic: gardening, food and drink, opinion | Link here |
Finally got round to harvesting my chilis today. The idea of leaving them on the bush until needed was not good:
|
|||||||||||||||||||||||
How much? 65 g! I could assume that if I had harvested them all a week or two ago, they might have weighed a little more, and there were 2 or 3 that I had harvested earlier. But that would still be way shy of 100 g, and I can buy them for about $1. In addition, this particular kind is too hot for my liking. I could try something like Poblanos next year, assuming that they will grow well here.
|
Ate my puppy?
|
Topic: animals, opinion | Link here |
While walking the dogs today, a middle-aged woman on a bicycle came by. We greeted her, but she didn't answer until she was almost past, and we had difficulty understanding her. But what I reconstructed was “That's the beast who ate my puppy last year”.
Huh? We've only had two incidents round here: three years ago they were attacked by another (foolhardy) dog, and they defended themselves in no uncertain manner. And last week they dogs attacked another dog while walking. There's no excuse for the second incident, but in each case the other dogs were not significantly injured. Did I imagine what the woman said?
|
Cleaning up old and mouldy computers
|
Topic: technology, opinion | Link here |
So now a new ThinkCentre is on its way. Time to move on with my network restructure. The idea is to keep all the videos on a central machine, probably eureka, which we can then access from any computer, and have teevee (existing ThinkCentre M58p) and tiwi (new ThinkCentre M93p) as redundant machines to drive the TV.
But I still haven't made up my mind about what to do in the office. One or two machines? Two has the great advantage that I can gradually upgrade. And what about the old M57p ThinkCentre? It failed early last year. I had some recollection of it having difficulties after a power failure (now a thing of the past), but I can't find any indication in my diary.
And then there's the question of monitors. I had had difficulties with the new dereel, which seemed to fail after some time. It wasn't helped by being plugged into a monitor that didn't want to stay connected to a non-responsive dereel while another input (eureka) was active. I had another Acer monitor in the corner, but it has an external power supply (19 V, no less), and I couldn't find it. Finally I found it today, so plug it in, connect the ThinkCentre and see what it says. Nothing. No response on the power button. Why that? It wasn't until much later that I noticed that I had had issues with a CMOS jumper on the motherboard, and that it wasn't clear how to replace it. Could it be that that would stop it from powering up? The good news is that if it is defective, it's the oldest of my ThinkCentres, so it wouldn't be much of a loss.
OK, the mains power cord was functional, right? Move the connectors to dereel, power on, all good. And dereel carried on working with no issues, though I didn't leave it on very long. First I need to decide how I want to spread my disks.
|
Fixing gdb
|
Topic: technology, opinion | Link here |
Yesterday's problems with gdb irritated me. Somewhere I heard a whisper “install gdb from ports”. OK, I can do that.
Aaargh!
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
People, what is this abomination? Yes, I don't have great objections to the sparing use of colour in output, but half colour? Only foreground, with no regard for background? I just plain can't read it.
Yes, it can be disabled. But it must be disabled. Why not allow it to be enabled?
| Tuesday, 16 June 2020 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
gdb pain
|
Topic: technology, opinion | Link here |
So I had to rebuild gdb to get rid of this horrible illegible coloured output. First problem: it didn't like my GNU make, the one I installed only a month or so ago. I had to upgrade.
Why? Why are things so intimately tied to revisions of standard software? Sure, upgrading didn't take anywhere as long as discovering the problem, but is that really necessary?
So, run make config and deselect anything that looks like an attempt to vomit colour over my screen. Build. Reinstall. Try out. Damn, it still displays things in colour!
Dammit, this must have hurt somebody else. Yes, indeed, a bug report: New gdb terminal colors are unreadable on GNOME Terminal "dark" theme:
That's surprising. I had thought that the colours were chosen for a black background, but it doesn't work there either. What were people thinking of? In any case, the solution should apply.
And what was it? “CLOSED ERRATA”. Long discussion, showing that in the case of Fedora (against which the bug was reported), this was default in each case. At least it's not only FreeBSD that allows such a violation of POLA. They did bring out some patch, but that was over a year ago, so it probably doesn't fix the POLA issue. Somehow that didn't come into discussion. From the response to the bug report:
I understand the frustration, but in my experience it is sometimes necessary to adjust colour schemes here and there when you're dealing with styled applications. The fact that GDB (and GNOME terminal, for that matter) provides ways to customize and disable the feature makes me see this as NOTABUG.
Not in my book. In the case of this bug, if the default installation results in illegible displays, something's wrong. And from my viewpoint, why is this a “styled application”?
They did, however, come out with a reasonable solution: there's a command
(gdb) set style enabled off
And that turns off the whole thing. I can put it in ~/.gdbinit and it will be applied automatically on startup. But why should I have to spend two days trying to get a program to work that I have been using for the past 30 years?
| Wednesday, 17 June 2020 | Dereel | Images for 17 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Crash! Crash! Crash!
|
Topic: technology, opinion | Link here |
Now I have tidied up my desktop (upper surface of a desk) sufficiently to be able to run a second system conveniently. A couple of days ago I had set up what proved to be an AMD Phenom™ 9550 with 6 GB of memory and noted that it didn't fail.
But it wasn't on for long, so today I took a look and considered how to proceed. The disk in the machine identified as dereel.lemis.com, but seemed unconfigured. Here the relevant boot information:
Jun 17 14:22:48 dereel kernel: FreeBSD 12.1-RELEASE r354233 GENERIC amd64
Jun 17 14:22:48 dereel kernel: FreeBSD clang version 8.0.1 (tags/RELEASE_801/final 366581) (based on LLVM 8.0.1)
Jun 17 14:22:48 dereel kernel: CPU: AMD Phenom(tm) 9550 Quad-Core Processor (2210.11-MHz K8-class CPU)
Jun 17 14:22:48 dereel kernel: Origin="AuthenticAMD" Id=0x100f23 Family=0x10 Model=0x2 Stepping=3
Jun 17 14:22:48 dereel kernel: Features=0x178bfbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,CLFLUSH,MMX,FXSR,SSE,SSE2,HTT>
Jun 17 14:22:48 dereel kernel: Features2=0x802009<SSE3,MON,CX16,POPCNT>
Jun 17 14:22:48 dereel kernel: AMD Features=0xee500800<SYSCALL,NX,MMX+,FFXSR,Page1GB,RDTSCP,LM,3DNow!+,3DNow!>
Jun 17 14:22:48 dereel kernel: AMD Features2=0x7ff<LAHF,CMP,SVM,ExtAPIC,CR8,ABM,SSE4A,MAS,Prefetch,OSVW,IBS>
Jun 17 14:22:48 dereel kernel: SVM: NP,NAsids=64
Jun 17 14:22:48 dereel kernel: TSC: P-state invariant
Jun 17 14:22:48 dereel kernel: real memory = 6442450944 (6144 MB)
Jun 17 14:22:48 dereel kernel: avail memory = 6189461504 (5902 MB)
I later gave this the number 2 in my (not yet published) lists. Number 1 was the defective ThinkCentre M57p.
OK, first the disks. How do I divide the load between eureka (to become my personal machine) and dereel (to become the server box)? Clearly things like mail, web server, mail server, firewall and NAT belong in dereel. But after some consideration, I think that at least some of the disks belong on eureka.
Which ones? Currently we have:
| eureka:/home | Greg's home, source files, web files, mail | |
| eureka:/Photos | Photos | |
| lagoon:/home | Yvonne's stuff | |
| teevee:/spool | Videos | |
| eureka:/dump | backups for other machines (external) | |
| eureka:/backup | backups for eureka (external) | |
| eureka:/photobackup | backup for eureka:/Photos (external) | |
| eureka:/videobackup | backup for teevee:/spool (external) | |
And apart from the root file systems, that's about all. The more I think of it, the more it seems that all shared data should be on eureka. It would certainly be the easiest to achieve: just move the current disk in teevee to eureka.
While pondering this, the monitor on dereel went blank. That's funny: it was a vty, and they don't have screen savers. But no, it was dead in the water. Can't ping, can't get any response from the keyboard. In the end pressed the reset button and... the system powered down! I've never seen that before.
More attempts showed that I couldn't get it to power on again. Power supply? Who knows? But I had two further machines to try out.
The next one (number 3) powered up and hung in the self-test, displaying 0078 in the bottom right-hand corner. That's reminiscent of my very first experience with an IBM clone, some time round 1986: the motherboard or processor was defective when I got it, and it displayed 1710 (from memory), indicating that it couldn't communicate with the disk. But I didn't think that any BIOS made in this millennium would show such primitive codes. But it seems so: this page suggests that there's a cable problem.
Got called away, so powered it down. Came back later. Couldn't power it on.
Dammit, what's wrong with these boxes? They all worked normally last time I used them. OK, one more box to go, number 4.
This one hung while booting, with the message
can't open '/boot/beastie.4th': device not configured.
Huh? Ah, that's a particularly clear message saying “my view of the disks on this machine doesn't match your view”. To get the message I had to load the kernel, so the disk was working. But it proved that, for some strange reason, this motherboard had a floppy disk configured, so /dev/ada0 corresponded to disk1, not disk0. Disable the floppy and all was well:
Jun 17 15:55:15 dereel kernel: CPU: AMD Athlon(tm) II X2 240 Processor (2812.76-MHz K8-class CPU)
Jun 17 15:55:15 dereel kernel: Origin="AuthenticAMD" Id=0x100f62 Family=0x10 Model=0x6 Stepping=2
Jun 17 15:55:15 dereel kernel: Features=0x178bfbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,CLFLUSH,MMX,FXSR,SSE,SSE2,HTT>
Jun 17 15:55:15 dereel kernel: Features2=0x802009<SSE3,MON,CX16,POPCNT>
Jun 17 15:55:15 dereel kernel: AMD Features=0xee500800<SYSCALL,NX,MMX+,FFXSR,Page1GB,RDTSCP,LM,3DNow!+,3DNow!>
Jun 17 15:55:15 dereel kernel: AMD Features2=0x37ff<LAHF,CMP,SVM,ExtAPIC,CR8,ABM,SSE4A,MAS,Prefetch,OSVW,IBS,SKINIT,WDT>
Jun 17 15:55:15 dereel kernel: SVM: (disabled in BIOS) NP,NRIP,NAsids=64
Jun 17 15:55:15 dereel kernel: TSC: P-state invariant
Jun 17 15:55:15 dereel kernel: real memory = 2147483648 (2048 MB)
Jun 17 15:55:15 dereel kernel: avail memory = 2042023936 (1947 MB)
Of course it had to be the slowest, only 1640 CPU marks. It also has almost no memory, but I can take the memory out of machine 2 and increase that to 8 GB. But I'm left wondering how much life is left in this machine. Should I maybe just throw them all away and start again with yet another ThinkCentre?
Gave up and went to watch TV. And teevee obligingly panicked while I was doing something relatively innocuous with a firefox. OK, now that I have a working (k)gdb, take a look.
warning: the debug information found in "/usr/lib/debug//boot/kernel/kernel.debug" does not match "/boot/kernel/kernel" (CRC mismatch).
What does that mean? It's gdb's way of saying “I don't check return codes”. There is no file /usr/lib/debug//boot/kernel/kernel.debug, of course, but that's clearly incorrect formatting for /boot/kernel/kernel.debug. But that file doesn't exist either, so no debug information was found, and there was also clearly no CRC. OK, where are we?
(No debugging symbols found in /boot/kernel/kernel)
0xffffffff80bf068f in sched_switch ()
(kgdb) bt
#0 0xffffffff80bf068f in sched_switch ()
#1 0xffffffff80bcaba7 in mi_switch ()
#2 0xffffffff80c19b53 in sleepq_catch_signals ()
#3 0xffffffff80c19e44 in sleepq_timedwait_sig ()
#4 0xffffffff80bca5b2 in _sleep ()
#5 0xffffffff80be0a42 in umtxq_sleep ()
#6 0xffffffff80be07f4 in do_wait ()
#7 0xffffffff80bde4ee in __umtx_op_wait_uint_private ()
#8 0xffffffff81091e87 in amd64_syscall ()
#9 <signal handler called>
#10 0x000000080954a60c in ?? ()
Backtrace stopped: Cannot access memory at address 0x7fffffffdc08
(kgdb)
And what was the cause of the crash? gdb is now too polite to mention such horrible things. The 17 lines it displays on starting are mainly propaganda and have absolutely nothing to do with the dump I'm looking at. But savecore is more verbose:
Jun 17 16:35:57 teevee kernel: Fatal trap 12: page fault while in kernel mode
Jun 17 16:35:57 teevee kernel: cpuid = 1; apic id = 01
Jun 17 16:35:57 teevee kernel: fault virtual address = 0x0
Jun 17 16:35:57 teevee kernel: fault code = supervisor read data, page not present
Jun 17 16:35:57 teevee kernel: instruction pointer = 0x20:0xffffffff82a3c132
Jun 17 16:35:57 teevee kernel: stack pointer = 0x28:0xfffffe00004bd880
Jun 17 16:35:57 teevee kernel: frame pointer = 0x28:0xfffffe002daefe10
Jun 17 16:35:57 teevee kernel: code segment = base 0x0, limit 0xfffff, type 0x1b
Jun 17 16:35:57 teevee kernel: = DPL 0, pres 1, long 1, def32 0, gran 1
Jun 17 16:35:57 teevee kernel: processor eflags = interrupt enabled, resume, IOPL = 0
Jun 17 16:35:57 teevee kernel: current process = 12 (swi4: clock (0))
Jun 17 16:35:57 teevee kernel: trap number = 12
Jun 17 16:35:57 teevee kernel: Timeout initializing vt_vga
Jun 17 16:35:57 teevee kernel: panic: page fault
Jun 17 16:35:57 teevee kernel: cpuid = 1
Jun 17 16:35:57 teevee kernel: time = 1592375690
Jun 17 16:35:57 teevee kernel: KDB: stack backtrace:
Jun 17 16:35:57 teevee kernel: #0 0xffffffff80c0b3f5 at kdb_backtrace+0x65
Jun 17 16:35:57 teevee kernel: #1 0xffffffff80bbf4ae at vpanic+0x17e
Jun 17 16:35:57 teevee kernel: #2 0xffffffff80bbf323 at panic+0x43
Jun 17 16:35:57 teevee kernel: #3 0xffffffff81091310 at trap_pfault+0
Jun 17 16:35:57 teevee kernel: #4 0xffffffff8109135f at trap_pfault+0x4f
Jun 17 16:35:57 teevee kernel: #5 0xffffffff810909b8 at trap+0x288
Jun 17 16:35:57 teevee kernel: #6 0xffffffff8106a91c at calltrap+0x8
A page fault in the clock interrupt handler? That doesn't sound like a software bug. What did we have last time?
Jun 12 17:53:24 teevee kernel: Fatal trap 12: page fault while in kernel mode
Jun 12 17:53:24 teevee kernel: cpuid = 1; apic id = 01
Jun 12 17:53:24 teevee kernel: fault virtual address = 0x0
Jun 12 17:53:24 teevee kernel: fault code = supervisor read data, page not present
Jun 12 17:53:24 teevee kernel: instruction pointer = 0x20:0xffffffff82c64132
Jun 12 17:53:24 teevee kernel: stack pointer = 0x28:0xfffffe00004bd880
Jun 12 17:53:24 teevee kernel: frame pointer = 0x28:0xfffffe002dafae10
Jun 12 17:53:24 teevee kernel: code segment = base 0x0, limit 0xfffff, type 0x1b
Jun 12 17:53:24 teevee kernel: = DPL 0, pres 1, long 1, def32 0, gran 1
Jun 12 17:53:24 teevee kernel: processor eflags = interrupt enabled, resume, IOPL = 0
Jun 12 17:53:24 teevee kernel: current process = 12 (swi4: clock (0))
Jun 12 17:53:24 teevee kernel: trap number = 12
Hmm, that's interesting. Exactly the same failure. That doesn't sound like a hardware bug. Maybe I should follow this up after all. But somehow I feel that there's too much work to be done just to get kgdb working again.
| Thursday, 18 June 2020 | Dereel | Images for 18 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Midwinter Hibiscus
|
Topic: gardening | Link here |
As I suspected, the Hibiscus rosa-sinensis “Uncle Max” that I planted outside has new flowers, only a couple of days before the Winter solstice. Here last week and today:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
New ThinkCentre
|
Topic: technology, opinion | Link here |
The ThinkCentre that I ordered on Saturday arrived today, rather more quickly than I had expected. And that from Toll, who haven't impressed me positively in the past
OK, was I right about the expansion slots? Yes, indeed. But a little surprise:
|
|||||||||||||||||||||||
What's that cable fouling the slots? I discovered that I could get a card in anyway, but it seems a strange thing to do.
Fired up the machine, which came up and then presented a blank screen, something that has happened too often in the recent past. Pressed the Shift key on the keyboard, and the display came back:
Getting Windows ready
Don't turn off your computer
Now isn't that reassuring? I let it go for a while, though I didn't have a mouse connected. OK, use the one I had on my test box. No reaction. In the end I used the wired mouse that Yvonne (for some reason) prefers, and that was recognized.
Why? Microsoft is supposed to be plug and play, but I've frequently noticed that it doesn't recognize mice without a “driver”, though FreeBSD has no difficulty with them.
The Microsoft wanted me to configure it, of course. Not now. Removed the disk and went looking for a new one, finding a disk marked tiwi.lemis.com 2018?. That should do. Put it in the machine along with a random display card with no obvious identification that I had found lying around. Oh. It came up as teevee, thus mightily confusing teevee:
arp: 6c:0b:84:04:0a:5c is using my IP address 192.109.197.158 on em0!
nfs server eureka:/: not responding
nfs server eureka:/: is alive again
It turns out that just changing the IP address on the new machine wasn't enough: it had NFS file systems mounted, and it hung.
OK, reboot, frob /etc/rc.conf and try again. And it came up with X running and all. Not quite good enough for a direct replacement for teevee: the environment has changed in 2 years, and in particular I'm now using mpv instead of mplayer. But I had known that.
Looking at the Xorg.0.log, discovered that the red card is a Radeon card, presumably the one I got with my last ThinkCentre (the M91p that has become dischord). And there was some issue with that card that I couldn't recall. Still, it was a good start, and I have a number of other cards (presumably from Nvidia) that I can put in the box.
First, what disk do I use? I discovered that I have no less than 35 disks lying around:
| 16 | PATA | |
| 5 | Defective SATA | |
| 3 | External USB disks | |
| 4 | Microsoft | |
| 5 | Various FreeBSD | |
| 2 | 3½" | |
I don't know why I keep all of them. I'm going to have to ensure that I keep a motherboard with a PATA in case I ever need the data on it.
It's not clear at all why I keep the defective disks. For recovery, perhaps? I should probably just throw them out. And the external disks are also clearly not a choice. I want to keep the Microsoft disks as they are, mainly because I wouldn't know how to reproduce them. There's “Windows” 8 and Vista there, amongst others.
So I'm down to the 5 FreeBSD disks. Which one? In principle I need about 100 GB: two root file systems of 40 GB each and some swap. But the FreeBSD disks start at 1 TB and go up to 6 TB. The one that I had booted was 1 TB, but the root partitions were too small, and I didn't really want to repartition them. The other one was one of my many /src file systems, probably a better choice. Put in the “too hard” basket for today, but clearly I'm going to have to use one of the 1 TB disks. It's also clear that I need some better form of record keeping.
|
Chicken cordon bleu, air fried
|
Topic: food and drink, opinion | Link here |
Chicken cordon bleu for dinner tonight. How do we prepare it? Last time we fried them in butter, and we were happy. But today Yvonne thought that we should do them in the “air fryer”.
OK, what time and temperature? All the recipes I could find on the web were agreed on the temperature: 350 degrees. And that where the “coffee machine”, the “air fryer” that I was using, doesn't go beyond 200°. And the time? They varied between 8 and 16 minutes. And one came up with the helpful information that this was a French dish (like “French Fries”, I suppose).
OK, convert 350 degrees to a sane value (about 177°, so take 180°) and see what happens. After 12 minutes or so the things still hadn't browned, so turned the heat up to 200° and gave a total of 16 minutes. The result? Barely acceptably browned:
|
|||||||||||||||||||||||
But they were overcooked! The moral of the story: don't try to prepare cordon bleu in an “air fryer”.
| Friday, 19 June 2020 | Dereel | Images for 19 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
All Australian Post
|
Topic: general, opinion | Link here |
Mail message from Australia Post today:
33 + 18-06-2020 Australia Post To groggyhims ( 521) + Your delivery has been collected
Yes, I knew that. Chris Bahlo had just brought us 3 crates of wine that she picked up for us at Napoleons (or “Napoleon”, as Auspost prefers to call it) post office. But today was 19 June, not 18. Look at the headers:
From: Australia Post <noreply@notifications.auspost.com.au>
To: groggyhimself@lemis.com
Subject: Your delivery has been collected
Date: Thu, 18 Jun 2020 22:14:08 -0600
Time zone UTC-6? That's fully 16 hours behind the times. Is this Australia Post or America Post?
|
tiwi upgrade
|
Topic: technology, opinion | Link here |
Finally found a disk for tiwi. Despite the surprising number of spare disks that I have, there was really only one choice: the other 1 TB disk, an ancient copy of my /src file system (now relocated to /home/src). OK, copy a disk image of the current root file system on teevee to eureka, mainly because I had space there. Then boot a spare machine with space for two disks, using my chosen disk as the second disk. Partition the disk. That's pretty straightforward. Well, almost. I got this output from the partition commands:
=== root@eureso (/dev/pts/0) /home/grog 14 -> partition-disk
ada1 destroyed
ada1 created
ada1p1 added
gpart: /dev/ada1p1: not enough space
ada1p2 added
ada1p3 added
ada1p4 added
ada1p5 added
=> 40 1953525088 ada1 GPT (932G)
40 128 1 freebsd-boot (64K)
168 83886080 2 freebsd-ufs (40G)
83886248 20971520 3 freebsd-swap (10G)
104857768 83886080 4 freebsd-ufs (40G)
188743848 1764781280 5 freebsd-ufs (842G)
The error message was from this line:
gpart bootcode -b /boot/pmbr -p /boot/gptboot -i 1 $DRIVE
I had left 64 kB for the boot partition, but that was no longer enough. Now the combined boot code is about 68 kB. And to think that in the Bad Old Days our PDP-8 had a loader all of 16 words (24 bytes) long, the RIM loader listed on this reference card. What bloat!
OK, 256 kB it is:
=== root@tiwi (/dev/pts/0) /home/grog 30 -> gpart show
=> 40 1953525088 ada0 GPT (932G)
40 512 1 freebsd-boot (256K)
552 83886080 2 freebsd-ufs (40G)
83886632 20971520 3 freebsd-swap (10G)
104858152 83886080 4 freebsd-ufs (40G)
188744232 1764780896 5 freebsd-ufs (842G)
Then copy the partition image to the root partitions (p2 and p4). The first one had to go over the net, of course, but the second could go locally. Only that wasn't a good idea. I tried again from eureka:
=== root@eureso (/dev/pts/0) /home/grog 35 -> dd if=/dev/ada1p2 of=/dev/ada1p4 bs=128k
327680+0 records in
327680+0 records out
42949672960 bytes transferred in 1096.931075 secs (39154395 bytes/sec)=== root@eureso (/dev/pts/0) /home/grog 36 -> dd if=/eureka/home/var/tmp/teevee-root-image of=/dev/ada1p4 bs=64k
655360+0 records in
655360+0 records out
42949672960 bytes transferred in 519.695614 secs (82643901 bytes/sec)
It took nearly twice as long locally! On reflection, that's not surprising: the disk had to do twice as many transfers.
OK, mount the first root file system and modify /etc/fstab for the new reality. But of course I had copied a live file system, so what I got was:
=== root@eureso (/dev/pts/2) /home/grog 6 -> mount /dev/ada1p2 /mnt
mount: /dev/ada1p2: R/W mount of / denied. Filesystem is not clean - run fsck. Forced mount will invalidate journal contents: Operation not permitted=== root@eureso (/dev/pts/2) /home/grog 7 -> fsck -y /dev/ada1p2
** /dev/ada1p2
USE JOURNAL? yes
** SU+J Recovering /dev/ada1p2
** Reading 33554432 byte journal from inode 4.
RECOVER? yes
** Building recovery table.
** Resolving unreferenced inode list.
** Processing journal entries.
WRITE CHANGES? yes
** 64 journal records in 5632 bytes for 36.36% utilization
** Freed 88 inodes (0 dirs) 1355 blocks, and 94 frags.
***** FILE SYSTEM MARKED CLEAN *****
Next, move the disk to the destination machine as tiwi. Boot. Works fine, but I couldn't start X: this time it had loaded the nvidia driver and failed. Later.
OK, bring the machine up to date. Build new world and kernel:
--------------------------------------------------------------
>>> Installing kernel GENERIC completed on Fri Jun 19 14:17:48 AEST 2020
--------------------------------------------------------------
4994.12 real 17382.53 user 1162.79 sys
1747228 maximum resident set size
48249 average shared memory size
2559 average unshared data size
207 average unshared stack size
275182279 page reclaims
51447 page faults
0 swaps
That compares favourably with teevee last time:
--------------------------------------------------------------
>>> Installing kernel GENERIC completed on Wed May 27 15:47:46 AEST 2020
--------------------------------------------------------------
12530.61 real 43024.19 user 3455.20 sys
2.5 times as fast! You can't complain about that.
|
Groggy shoots himself in foot
|
Topic: technology, opinion | Link here |
Booting tiwi was another matter:
Jun 19 15:22:20 tiwi kernel: KLD linprocfs.ko: depends on kernel - not available or version mismatch
Jun 19 15:22:20 tiwi kernel: linker_load_file: /boot/kernel/linprocfs.ko - unsupported file type
Jun 19 15:22:20 tiwi kernel: KLD linprocfs.ko: depends on kernel - not available or version mismatch
Jun 19 15:22:20 tiwi kernel: linker_load_file: /boot/kernel/linprocfs.ko - unsupported file type
Jun 19 15:22:20 tiwi kernel: KLD if_ed.ko: depends on kernel - not available or version mismatch
Jun 19 15:22:20 tiwi kernel: linker_load_file: /boot/kernel/if_ed.ko - unsupported file type
Jun 19 15:22:20 tiwi kernel: lo0: link state changed to UP
Jun 19 15:22:20 tiwi kernel: KLD if_ed.ko: depends on kernel - not available or version mismatch
Jun 19 15:22:20 tiwi kernel: linker_load_file: /boot/kernel/if_ed.ko - unsupported file type
Jun 19 15:22:20 tiwi kernel: KLD ums.ko: depends on kernel - not available or version mismatch
Jun 19 15:22:20 tiwi kernel: linker_load_file: /boot/kernel/ums.ko - unsupported file type
Jun 19 15:22:20 tiwi root[904]: /etc/rc: WARNING: Unable to load kernel module linux64
Jun 19 15:22:20 tiwi kernel: KLD linux64.ko: depends on kernel - not available or version mismatch
Jun 19 15:22:20 tiwi kernel: linker_load_file: /boot/kernel/linux64.ko - unsupported file type
What's that nonsense? The whole point of a kernel build is to ensure that that kind of issue doesn't occur. Off to take a look. Yes, all the modules mentioned were built today, along with the kernel. Check:
=== root@tiwi (/dev/pts/0) /usr/src 42 -> uname -a
FreeBSD tiwi.lemis.com 12.1-STABLE FreeBSD 12.1-STABLE r359522 GENERIC amd64
That looks right, up to a point. I had specifically asked for real information about the kernel in /etc/src.conf:
# Don't drop build date, dammit!
# What a silly name for the knob.
# It's about as "intuitive" as the name of this file.
WITHOUT_REPRODUCIBLE_BUILD=dammit
Is the syntax wrong? Went reading src.conf(5) and was no wiser. About all it said on the subject was:
The values of variables are ignored regardless of their setting; even if they would be set to "FALSE" or "NO". The presence of an option causes it to be honored by make(1).
OK, does that mean no value at all, just the name? Tried that, and seriously upset make.
===== Fri 19 Jun 2020 15:37:18 AEST on tiwi.lemis.com: Make -j 6 kernel
make: "/etc/src.conf" line 4: Need an operator
make: Fatal errors encountered -- cannot continue
make: stopped in /eureka/home/src/FreeBSD/svn/stable/12
The real question is: is this my kernel? Went through the kernel with strings(1) and found:
=== grog@tiwi (/dev/pts/2) ~ 8 -> strings /boot/kernel/kernel | grep 12.1-STABLE
FreeBSD 12.1-STABLE #0 r362332: Fri Jun 19 14:13:12 AEST 2020
Oh. That looks different. How about:
=== grog@tiwi (/dev/pts/2) ~ 9 -> strings /destdir/boot/kernel/kernel | grep 12.1-STABLE
FreeBSD 12.1-STABLE r359522 GENERIC
Oh. I got the kernel from the wrong partition. How could that happen? We have:
=== grog@teevee (/dev/pts/8) ~ 9 -> df
Filesystem 1048576-blocks Used Avail Capacity Mounted on
/dev/ada0p4 39,662 20,643 15,845 57% /
/dev/ada0p2 39,662 21,446 15,042 59% /destdir
After a little thinking, it dawned on me: I had booted the kernel from /ada0p2 but mounted /dev/ada0p4 as the root file system. A quick check in /destdir/boot/loader.conf confirmed:
# For alternate system
currdev=disk0p2
rootdev=disk0p2
That might just as well have been commented out, since it's the default: load the kernel from disk0p2 (an obfuscatory way of saying /dev/ada0p2). Then mount the file systems according to /etc/fstab:
# Device Mountpoint FStype Options Dump Pass#
/dev/ada0p4 / ufs rw 1 1
/dev/ada0p2 /destdir ufs rw 0 2
So all my fault after all:
|
|||||||||||||||||||||||
Why didn't I get the problem until the second reboot? The first time round the file systems were identical.
That was almost enough for the day. How about trying X again? Removed /etc/X11/xorg.conf and restarted X. Yes, it worked, sort of. But not only did the mouse not emulate 3 buttons, it didn't even display the cursor!
In the meantime I had found out what issues I had had with the Radeon card. One was that I claimed that there was no HDMI output. But there is. How did I get that impression?
The other issue was with GLX. I've seen more of them since then, and I think I have an idea how to fix it. I hope I can: the issues I've had with teevee could be due to the Nvidia graphics, so it would be good to have a different maker to compare with. But not today.
Started off a port upgrade, with the usual pain. It didn't finish by the evening.
In the meantime, shot myself in the other foot. I had deliberately started tiwi as a clone of teevee, so I had the same crontab on each machine. It was there to download news from YouTube, to a destination on /spool. So each system wrote to the same file, not simultaneously enough to avoid problems. The results were useless. I suppose the best choice would be to download outside the /spool file system; if I have to swap machines, I don't have to think about that.
|
Greg's diary: verbal diarrhoea?
|
Topic: general, technology, opinion | Link here |
I've written a lot about my system upgrade pain over the last 6 months. So far this month it's 35 pages of printed output. Too much? Who reads this stuff, anyway?
There's a simple and good answer to that question: I do. I've spent a lot of time looking back in this diary to see what surprises the systems have had for me in the past. And if anything, I didn't go into enough detail. It really is useful having this information available.
| Saturday, 20 June 2020 | Dereel | Images for 20 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More grid outages
|
Topic: Stones Road house, general | Link here |
Another couple of grid power failures this morning, the first one second, the second 5 seconds.
This is close to my worst case scenario. The states of charge were 28% and 26%: in the 8 minutes between the two, the charge dropped by 2%. At 0:54 it was 20% (the set minimum), and a two-hour outage at that time would have gone unnoticed until too late.
|
More tiwi installation
|
Topic: technology, opinion | Link here |
In principle tiwi is now up and running with the Radeon display board. Today, as planned, I looked for GLX and mouse button emulation.
First run X -configure and move the /root/Xorg.conf.new to /etc/X11/xorg.conf. Start that. Yes, now we have a mouse cursor. But still no middle button emulation, still no GLX (here mpv complaining on-screen):
|
|||||||||||||||||||||||
Out looking for solutions to the GLX issue first. Most of the problems people were having were with Nvidia GPUs, and most were old. But this page included an xorg.conf file with this interesting section:
Section "Module"
Load "glx"
EndSection
Why is that needed? In any case, tried it, with no improvement. OK, nothing for it: dive into Xorg.0.log. I've been running X for 30 years now, but it's not getting any easier. As an example:
[ 83934.490] (II) event2 - Power Button: is tagged by udev as: Keyboard
[ 83934.490] (II) event2 - Power Button: device is a keyboard
[ 83934.490] (II) event2 - Power Button: device removed
[ 83934.490] (**) Option "config_info" "udev:/dev/input/event2"
[ 83934.490] (II) XINPUT: Adding extended input device "Power Button" (type: KEYBOARD, id 8)
[ 83934.490] (**) Option "xkb_rules" "evdev"
[ 83934.491] (II) event2 - Power Button: is tagged by udev as: Keyboard
[ 83934.491] (II) event2 - Power Button: device is a keyboard
What on Earth does that mean? But there are plenty of similar messages where devices are identified, removed and (apparently) used anyway.
In the case of GLX, found:
[ 82445.487] (II) "glx" will be loaded. This was enabled by default and also specified in the config file.
At least that confirms that I didn't need this entry after all. But what went wrong?
[ 82445.487] (II) LoadModule: "glx"
[ 82445.487] (II) Loading /usr/local/lib/xorg/modules/extensions/libglx.so
[ 82445.555] (II) Module glx: vendor="NVIDIA Corporation"
[ 82445.555] compiled for 4.0.2, module version = 1.0.0
[ 82445.555] Module class: X.Org Server Extension
[ 82445.555] (II) NVIDIA GLX Module 390.132 Fri Nov 1 03:45:29 PDT 2019
NVIDIA? Did they write the software? Well, clearly they did. But:
[ 82446.284] (II) Initializing extension GLX
[ 82446.292] (EE) Failed to initialize GLX extension (Compatible NVIDIA X driver not found)
It looks as if the nvidia driver installs the module in the middle of the X distribution. What does /usr/ports/x11/nvidia-driver/pkg-plist say?
%%MODULESDIR%%/extensions/.nvidia/libglx.so
%%MODULESDIR%%/extensions/.nvidia/libglx.so.1
%%MODULESDIR%%/extensions/libglxserver_nvidia.so
%%MODULESDIR%%/extensions/libglxserver_nvidia.so.1
Presumably %%MODULESDIR%% expands to /usr/local/lib/xorg/modules/. That looks OK, but it's not where my system has placed it. What do the logs for pkg-plist say?
r303429 | kwm | 2012-09-01 01:44:41 +1000 (Sat, 01 Sep 2012) | 15 lines
Solve libGL.so and libglx.so conflict situation between libGL, xorg-server and
the nvidia-driver. Install the libraries in port specific directories.
Use pkg-install and pkg-deinstall scripts to update the hardlinks to the
default locations of these files.
That's a long time ago. What went wrong?
The other half of the issue was: where's the GLX support for Radeon? Went looking, but couldn't find anything. Maybe it's in base X. The information in the page above includes:
[ 88.527] (II) LoadModule: "glx"
[ 88.527] (II) Loading /usr/local/lib/xorg/modules/extensions/libglx.so
[ 88.530] (II) Module glx: vendor="X.Org Foundation"
[ 88.530] compiled for 1.18.4, module version = 1.0.0
[ 88.530] ABI class: X.Org Server Extension, version 9.0
[ 88.530] (==) AIGLX enabled
OK, more head-scratching. That's enough for today. On with the mouse emulation stuff. It had started working for me on teevee again after I stopped moused. But I had already done that on tiwi. OK, my virgin xorg.conf doesn't contain the all-important Emulate3Buttons. Put that in and try again. Still doesn't work. Off to compare log files between teevee and tiwi. No obvious difference. In particular, no mention in either log file of Emulate3Buttons. Why not? For the fun of it, added another option:
Option "Emulate3Buttons" "True"
Option "Dontrustanything" "True"
Option "ZAxisMapping" "4 5 6 7"
That, too, started with no issues and no mention of the bogus option. Ugh.
And then I found this page, which included:
Section "InputClass"
Identifier "middle button"
MatchIsPointer "on"
MatchDriver "libinput"
Option "MiddleEmulation" "on"
EndSection
What's all that? But it mentioned a program xinput. OK, install, run as mentioned on the page:
=== root@tiwi (/dev/pts/0) /etc/X11 50 -> xinput
â¡ Virtual core pointer id=2 [master pointer (3)]
â â³ Virtual core XTEST pointer id=4 [slave pointer (2)]
â â³ sysmouse id=7 [slave pointer (2)]
⣠Virtual core keyboard id=3 [master keyboard (2)]
â³ Virtual core XTEST keyboard id=5 [slave keyboard (3)]
â³ kbdmux id=6 [slave keyboard (3)]
Ugh! A program that ignores locales. OK, start a uxterm and try again:
=== grog@tiwi (/dev/pts/9) ~ 1 -> xinput
⎡ Virtual core pointer id=2 [master pointer (3)]
⎜ ↳ Virtual core XTEST pointer id=4 [slave pointer (2)]
⎜ ↳ sysmouse id=7 [slave pointer (2)]
⎣ Virtual core keyboard id=3 [master keyboard (2)]
↳ Virtual core XTEST keyboard id=5 [slave keyboard (3)]
↳ kbdmux id=6 [slave keyboard (3)]
OK, what else can I get out of this stuff? RTFM, which doesn't mention the syntax I used. It wants --every --single --input to be preceded by two dashes, except in the examples. Ugh. But with a bit of experimentation I got:
=== grog@tiwi (/dev/pts/9) ~ 2 -> xinput --list --long
...
⎜ ↳ sysmouse id=7 [slave pointer (2)]
Reporting 3 classes:
Class originated from: 7. Type: XIButtonClass
Buttons supported: 5
Button labels: "Button Left" "Button Middle" "Button Right" "Button Wheel Up" "Button Wheel Down"
Button state:
Class originated from: 7. Type: XIValuatorClass
Detail for Valuator 0:
Label: Rel X
Range: -1.000000 - -1.000000
Resolution: 1 units/m
Mode: relative
Class originated from: 7. Type: XIValuatorClass
Detail for Valuator 1:
Label: Rel Y
Range: -1.000000 - -1.000000
Resolution: 1 units/m
Mode: relative
Doubtless this says something to developers, but it doesn't bring me much further beyond mentioning the buttons. And the results on teevee were identical.
OK, back to the Debian page. He tried again with another incorrectly documented command:
=== grog@tiwi (/dev/pts/9) ~ 6 -> xinput list-props 7
Device 'sysmouse':
Device Enabled (144): 1
Coordinate Transformation Matrix (145): 1.000000, 0.000000, 0.000000, 0.000000, 1.000000, 0.000000, 0.000000, 0.000000, 1.000000
Device Accel Profile (257): 0
Device Accel Constant Deceleration (258): 1.000000
Device Accel Adaptive Deceleration (259): 1.000000
Device Accel Velocity Scaling (260): 10.000000
Device Node (261): "/dev/sysmouse"
Button Labels (262): "Button Left" (147), "Button Middle" (148), "Button Right" (149), "Button Wheel Up" (150), "Button Wheel Down" (151)
Mouse Middle Button Emulation (263): 1
Mouse Middle Button Timeout (264): 50
Enabled! Could it be an issue with the middle button timeout? I can try that:
=== grog@tiwi (/dev/pts/9) ~ 10 -> xinput set-prop 7 "Mouse Middle Button Timeout" 500
(no output)=== grog@tiwi (/dev/pts/9) ~ 11 -> xinput list-props 7
...
Mouse Middle Button Timeout (264): 500
500 ms should be plenty. But still no change.
OK, off to teevee (physically in the lounge room) and try things out. It worked, of course. Well, sort of. The Logitech M705 mouse also didn't work with Emulate3Buttons. I had never tried it before because I use the side button on that mouse.
But didn't it work? Well, not always. I tried several times with xev, and once it did work. So there's something very funny going on here. OK, check the information for the devices. What was the name of that program again? libinput? Tried that:
=== root@teevee (/dev/pts/4) ~ 11 -> libinput list-devices
Device: System keyboard multiplexer
Kernel: /dev/input/event0
...
Device: System mouse
Kernel: /dev/input/event1
Group: 2
Seat: seat0, default
Capabilities: pointer
Tap-to-click: n/a
Tap-and-drag: n/a
Tap drag lock: n/a
Left-handed: disabled
Nat.scrolling: disabled
Middle emulation: disabled
Calibration: n/a
Scroll methods: button
Click methods: none
Disable-w-typing: n/a
Accel profiles: flat *adaptive
Rotation: n/a
No, but. The program I was looking for is xinput, not libinput. But that's another program, and this time it tells me that “Middle emulation” is disabled. So we have two different programs, both apparently from the same source, giving conflicting information.
Or are they from the same source? The man page for libinput is appallingly bad, requiring people to look at separate man pages (like libinput-list-devices(1), a non-existent program, for the command I used). At the end there's the tell-tale:
I've never been very keen on Wayland, but this experience is strongly reinforcing my dislike.
I'm still trying to understand the hierarchy of the mouse access in X, but it seems that libinput opens /dev/sysmouse, a FreeBSD driver, which amalgamates all the mice on the system to a single stream. That's not good: I want different button mappings for the M705 and the other two, and the M705 also needs different acceleration parameters. In addition, the man page is not very helpful. I'm gradually coming to the conclusion that I should examine sysmouse more carefully.
X has always been “fun” to configure, but somehow this exceeds anything I have experienced in the past. 28 years ago I installed X on BSD/OS (called BSD/386 in those days). It was on non-standard hardware and a beta copy of the operating system. It took me less than a day.
|
Tiling the verandah
|
Topic: Stones Road house | Link here |
Sean Barbetti, a tiler recommended by Beaumont Tiles a month ago, out today to take a look at the exterior tiling work that I've been pushing in front of me for the past 5 years. Will something finally come of it?
|
More riders visit
|
Topic: general, animals | Link here |
Another visitation today:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
From left to right, Cherilyn Fuhlbohm on Ruby, Craig Sitch on Algernon, Amber Fitzpatrick on Keldan and Chris Bahlo on Pinta.
|
Hoof sizes
|
Topic: animals | Link here |
Craig's horse, Algernon, is a little larger than Keldan. While walking the dogs, we saw some hoofprints which rather confirmed the difference:
|
|||||||||||||||||||||||
|
Fondue again
|
Topic: food and drink, opinion | Link here |
Fondue de fromage for dinner again. How much bread did we need? Last time I established that the bread cubes should weigh about 3.5 to 4 g, but how much bread did we need in total? I guessed 1.4 times as much, but while cutting it it quickly became clear that that was too much. Stopped after about 1.1x, but in the end we still had lots left over. Weighing the remainder suggested that we should be looking at between ⅔ and ¾ as much bread by weight as cheese.
| Sunday, 21 June 2020 | Dereel | Images for 21 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Finding vegetables
|
Topic: food and drink, opinion | Link here |
Yvonne and I share responsibilities in the kitchen, but each of us has different ways of organizing things. That's clear in the fridge. Here the vegetable compartment (Yvonne's domain):
|
|||||||||||||||||||||||
What is that stuff? This morning I was looking for tomatoes. In the bag, they looked like this:
|
|||||||||||||||||||||||
Are those all tomatoes, or are the ones on the right (red) potatoes? I decided yes, but then Yvonne showed me another bag:
|
|||||||||||||||||||||||
There the vegetables on the right are potatoes.
There must be a simpler way.
|
Garden flowers in early winter
|
Topic: gardening | Link here |
Today was the Winter solstice, time for the monthly flower photos. Despite the relatively mild winter so far, there isn't much to show. I had been rather pleased with the Hibiscus rosa-sinensis “Uncle Max”, which is still flowering and has buds on their way:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That's a far cry from what it looked like in October last year:
|
||||||||||||||||||||||||
On the other hand, we had a bud this time last year:
|
||||||||||||||||||||||||
Other plants that appear to be defying the season include Cannas:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The Corymbia ficifolia is also still flowering, though it is not looking as well as it did a couple of weeks ago. I suspect that a number of the buds won't open:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
One Hebe, and only one, is flowering relatively profusely:
|
|||||||||||||||||||||||
And the Strelitzia nicolai looks like it's producing a new flower spike:
|
|||||||||||||||||||||||
On the other hand, some plants are doing less well than I had expected, including the roses:
|
|||||||||||||||||||||||
At this time of year we normally have quite a few still flowering, but this year this is about the only one:
|
|||||||||||||||||||||||
And the Fuchsia triphylla is once again looking effectively dead:
|
|||||||||||||||||||||||
And the Grevillea robusta doesn't seem to be making it. It's now smaller than it was when we bought it (second image, coincidentally a year ago today):
|
|||||||||||||||||||||||
|
||||||||||||||||||||||||
I wonder why. I thought that the name “robusta” was supposed to be indicative.
And surprisingly, the Schinus molle that I've been observing since its transplantation last November is not improving the way I expected. Yes, it has many more leaves than when we planted it there, but now a couple of branches have died off:
|
|||||||||||||||||||||||
The other one, which we planted to the west of the house, might yet overtake it:
|
|||||||||||||||||||||||
The mistreated Epazote plant also looks as if it will make it:
|
|||||||||||||||||||||||
There are the first signs of spring. The Narcissus that looked so sad a couple of weeks ago is now flowering more reliably, though it's small, and there's a second bulb flowering as well:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Indoors, the transplanted Hibiscus rosa-sinensis from Yvonne's bedroom is growing like fury, unfortunately only upwards:
|
|||||||||||||||||||||||
I'll have to cut the main stem off, but I'd like to wait until there's a good chance of the cutting taking.
And the Buddleja × weyeriana cuttings that I salvaged from the dying bush have taken well. They're even producing flowers:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
I still don't understand why the main plant died. But we're going to have to do something with the cuttings: the stems are far too thin, and I think they need planting outside.
|
Still more tiwi fun
|
Topic: technology, opinion | Link here |
Where do I go from here with tiwi? Clearly I have the wrong version of GLX, and based on experience on teevee it's not beyond the bounds of possibility that I can get other mice to work with middle button emulation. OK, more searching in the X directory hierarchy. The nvidia driver claims to install in /usr/local/lib/xorg/modules/extensions/.nvidia. Is that the only dot directory? No!
=== root@tiwi (/dev/pts/0) /usr/local/lib/xorg/modules/extensions 56 -> l -aR
total 29
drwxr-xr-x 4 root wheel 512 19 Jun 22:25 .
drwxr-xr-x 5 root wheel 1,024 19 Jun 22:25 ..
drwxr-xr-x 2 root wheel 512 4 Mar 13:42 .nvidia
drwxr-xr-x 2 root wheel 512 19 Jun 22:25 .xorg
-r--r--r-- 3 root wheel 14,964,128 4 Mar 13:41 libglx.so
-r--r--r-- 3 root wheel 14,964,128 4 Mar 13:41 libglx.so.1
./.nvidia:
total 29
drwxr-xr-x 2 root wheel 512 4 Mar 13:42 .
drwxr-xr-x 4 root wheel 512 19 Jun 22:25 ..
-r--r--r-- 3 root wheel 14,964,128 4 Mar 13:41 libglx.so
-r--r--r-- 3 root wheel 14,964,128 4 Mar 13:41 libglx.so.1
./.xorg:
total 1
drwxr-xr-x 2 root wheel 512 19 Jun 22:25 .
drwxr-xr-x 4 root wheel 512 19 Jun 22:25 ..
-rwxr-xr-x 1 root wheel 340,128 16 Jun 17:21 libglx.so
The file size and date suggest that libglx.so, libglx.so.1, .nvidia/libglx.so and .nvidia/libglx.so.1 are the same file. Checking shows that they're not (they have different inode numbers), but that the content is the same: in other words, they just wasted 42 MB of disk space.
The other thing is the size itself. .xorg/libglx.so is 340 kB in size; the Nvidia version is 44 times the size. I wonder if, like other Nvidia offerings, it contains 30 or 40 individual binaries, each for a different GPU.
OK, replace the libglx.so with the Xorg version and try again.
|
|||||||||||||||||||||||
=== grog@tiwi (/dev/pts/2) ~ 20 -> mpv /spool/Docco/AlJaz-15
(+) Video --vid=1 (*) (h264 1920x1080 30.000fps)
(+) Audio --aid=1 (*) (aac 2ch 48000Hz)
error: XDG_RUNTIME_DIR not set in the environment.
[vo/gpu/x11] X11 error: BadValue (integer parameter out of range for operation)
[vo/gpu/x11] Type: 0, display: 0x80c607000, resourceid: 0, serial: 41
[vo/gpu/x11] Error code: 2, request code: 98, minor code: 3
[vo/gpu] Could not create GLX context!
error: XDG_RUNTIME_DIR not set in the environment.
[vo/gpu/vulkan/libplacebo] EnumeratePhysicalDevices(inst, &num, NULL): VK_ERROR_INITIALIZATION_FAILED
[vo/gpu/vulkan/libplacebo] Found no suitable device, giving up.
[vo/gpu/vulkan/libplacebo] Failed initializing vulkan device
Failed to open VDPAU backend Shared object "libvdpau_r600.so" not found, required by "mpv"
[vo/vdpau] Error when calling vdp_device_create_x11: 1
Failed to open VDPAU backend Shared object "libvdpau_r600.so" not found, required by "mpv"
[vo/vdpau] Error when calling vdp_device_create_x11: 1
[vo/xv] Warning: this legacy VO has bad quality and performance, and will in particular result in blurry OSD and subtitles. You should fix your graphics drivers, or not force the xv VO.
AO: [oss] 48000Hz stereo 2ch s32
VO: [xv] 1920x1080 yuv420p
[vo/xv] X11 can't keep up! Waiting for XShm completion events...
Ugh! Looks like there are other files to replace as well.
And the mouse? This time I brought one of the mice from teevee, after checking that the middle button emulation worked, and plugged it in to tiwi. Restarted X. No emulation.
And while pondering that, discovered that I had a now-you-see-me-now-you-don't situation with the keyboard. xev agreed that I had remapped the keys. xterm didn't want to know.
Sigh. More pondering. Later it occurred to me that the issues with mouse and keyboard might not be as bad as they sounded. I recall having similar issues with the keyboard on teevee, one of the many unsolved mysteries, but with a workaround. And of course when I plugged the mouse in, usbd started a moused. And I already knew that that wouldn't work well. But that was more than enough frustration for the day.
|
Frying trout, revisited
|
Topic: food and drink, opinion | Link here |
Truites meunières for dinner this evening. Now we have a modern frying pan big enough for them. But somehow we have to get used to it. For some reason it heats more quietly than the old oval fish pan. I've also noted that it needs higher heat settings for the same temperature, maybe reflecting its size: a 32 cm diameter pan is more than 50% larger than a more common 26 cm pan. And so it was that one side of the fish was rather darker than planned:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
| Monday, 22 June 2020 | Dereel | Images for 22 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
A new computer for CJ
|
Topic: technology | Link here |
CJ Ellis along this morning with his “new” computer, a Dell Inspiron desktop (huh? I thought Inspirons were laptops), with an Intel i3-3240 @ 3.4 GHz processor and 8 GB of RAM. 2289 The old one has a Core 2 Duo E7400, and certainly no more than 4 GB, though when he bought it it only had 2 GB.
Certainly the CPU power difference is noticeable: CPU Mark 966 for the old machine, 2289 for the new one. And the additional memory can only be good, especially as the machine runs (of course) “Windows” 10.
OK, what was the problem? The passwords didn't work. CJ still hasn't understood the structure of his accesses, and it seems that the big issue was simply deciding when to use what login. That's why I have a well-hidden file with all his login details—except Facebook, for which he signed up after I had created the file.
OK, mess around with this horrible interface, made worse by the fact that Google Chrome was installed. I'm getting to hate that browser more and more. Now the search bar doesn't seem to include URLs, though possibly some other strangeness in the interface might be to blame for that.
OK, can I contact the machine with TeamViewer? No. I have an old, worn-out TeamViewer, and I need to upgrade. Would you like this version or this version? The latter, of course, since presumably it was the most up-to-date. But no, of course it was sorted the wrong way round, and it was the older version. Something I could have seen, of course, if I had read more carefully.
But not only was it the older of the two, it was too old. TeamViewer had detected that my installation was out of date and offered me an update that was also out of date. The other one worked find—I continue to be impressed how well it works—but why do they do things like that?
Finally got things finished and left it to CJ:
|
|||||||||||||||||||||||
Things didn't stay that way, though. In the evening I got a call from him, with problems that sounded like lack of Internet connectivity, particularly since it happened with his old machine as well. Cable problems? No, he had checked them. OK, take two aspirin and call me in the morning.
|
Political correctness
|
Topic: politics, language, technology, opinion | Link here |
A couple of days ago a mail went round the FreeBSD developer lists: we should consider adapting our terminology to the IETF draft terminology (backup here in case of bit rot). I thought about following up, but decided that the topic could quickly produce more correspondence than I wanted to follow.
I was certainly right about that. To date there have been 343 messages on the topic and the problems that grew out of it.
The issue does not seem to be to encourage uniform, descriptive terminology, but to avoid terms that some people (presumably in the USA) might find offensive. To quote the title of the IETF draft:
Terminology, Power and Oppressive Language
It seems that the terms that are most oppressive (or powerful?) are the contrast “master/slave” and “blacklist/whitelist”. You'd think that this was related to the current dual protests about police brutality and racism, notably in the USA, but the draft dates to 22 October 2018, and in fact expired over a year ago on 25 April 2019. I don't know if any of the hundreds of messages addressed that.
The first concrete change to the FreeBSD code was this one, where the name “slave”, including in composites, was replaced by “minion”. So “dominion” was replaced by “doslave”. I wonder what people in the Doslave of Canada think about that. And it begs the question: what does “minion” even mean. As ever, the OED has many definitions, including:
A male or female lover. Also (frequently derogatory): a man or woman kept for sexual favours; a mistress or paramour.
A fastidious or effeminate man; a fop, a dandy.
As a term of endearment or affection: darling, dear one.
As a derogatory term (esp. as a form of address): slave, underling.
The discussion made it clear that the committer knew none of this background, but that's exactly the point: some terms, like “nigger”, really derived from Latin “niger” meaning “black”, are so offensive to some people that they won't even write them, calling them the “N word” or some such euphemism. But again from the OED:
Used by people who are not black as a relatively neutral (or occasionally positive) term, with no specifically hostile intent.
So clearly it's the perception, not the own understanding, that's the point here. And for that people need a much wider perspective than any of us have.
And then the FreeBSD suggestion that was intended to unify terminology had exactly the opposite effect: another developer took it upon himself to rename the term Master Boot Record to Main Boot Record. More complaints, both for and against.
But the list goes on. Explicit pairs of apparently offensive terms include:
- primary / secondary
- leader / follower
- active / standby
- primary / replica
- writer / reader
- coordinator / worker
- parent / helper
- blocklist / allowlist
- block / permit
What kind of brain damage can lead to this kind of confusion? What's offensive about “writer” or “reader”?
I'm not the only one to object. A number of developers have publicly left the project. I won't, but I wish that, in the words of Poul-Henning Kamp, people would shut up and code.
| Tuesday, 23 June 2020 | Dereel | Images for 23 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Repairing CJ's computer
|
Topic: technology, opinion | Link here |
CJ Ellis called me on the dot of 9:30 this morning, as agreed. Still no joy. OK, fire up TeamViewer and connect. No, sorry, no can do. No password generated. Some message about a proxy not functioning.
OK, he's talking to me on his Internet connection, so that's working. His NTD is connected only to the ATA, which also has an output for the computer. Both computers show the same symptoms. Must be the cable, right? No, says CJ, it's connected.
OK, is it connected to a wrong port? Tell me what connections there are on the back of the ATA. (Pause) power in, NTD in, phone out, spare phone out (for a second phone) and the one with the yellow cable (the connection to the computer). Oh, and the yellow cable wasn't pushed all the way in.
Daily stuff for most people, I suppose. But I had to really rub his nose in it. And of course the machine now works as well as you'd expect Microsoft “Windows” 10 to work.
|
Youtube-dl: legal?
|
Topic: multimedia, technology, opinion | Link here |
I've been using youtube-dl for downloading all sorts of videos for some time now. Basically, it's just a tool, one that the GUI-obsessed mainstream seems to consider unnecessary.
I don't. GUIs have their strengths, but they require far too much human intervention, particularly for repetitive tasks. In many cases it's simpler to download videos, not just from YouTube, but also from online services of TV broadcasters. Clearly there's no legal issue: I could just as well look at them with a web browser, if I could stand it.
Or is there? Do some misguided companies distinguish between legalities depending on how you look at them? Went looking, but couldn't find anything clear. This page quotes YouTube's Terms of Service:
You may access Content for your information and personal use solely as intended through the provided functionality of the Service and as permitted under these Terms of Service. You shall not download any Content unless you see a “download” or similar link displayed by YouTube on the Service for that Content.
Or does it? The URL is on archive.org, and I don't know how to easily find the current information. And what's a “download” link? What does it look like? Where might it be? It's far too difficult to find this stuff.
It's also nonsensical. I don't download much from YouTube—the image quality is far too bad—but I don't have time to search for special rules for each site (or potentially page) that I visit. If people don't want me to access their videos, they can tell me so, like here:
=== grog@teevee (/dev/pts/7) /spool/Videos 90 -> youtube-dl https://tvthek.orf.at/profile/AD-Die-Muse-des-Moerders/13889613/AD-Die-Muse-des-Moerders/14055674
[orf:tvthek] 14055674: Downloading webpage
[orf:tvthek] 14715434: Downloading f4m manifest
[orf:tvthek] 14715434: Downloading m3u8 information
ERROR: This video is not available from your location due to geo restriction
You might want to use a VPN or a proxy server (with --proxy) to workaround.
That makes it clear that the reason isn't because of a missing “Download” link, but because my country has a superfluous “al” in its name.
So, content providers: don't hide behind hard-to-find and nebulous regulations. If you don't want people to access your content in specific ways, use the tools at your disposal.
|
youtube-dl timing
|
Topic: multimedia, technology, opinion | Link here |
It seems that a large number of online content is stored in two different files, one considerably larger than the other. It conveniently shows the download speed and the remaining time. Here an example:
[hlsnative] Downloading m3u8 manifest
[hlsnative] Total fragments: 318
[download] Destination: Friends Like These-6067128257001.fhls-2407-3.mp4
[download] 2.6% of ~963.33MiB at 1.15MiB/s ETA 14:10
This is the first of two images. OK, the information seems to make sense. Calculating the download time, the time for total size at that speed is 13.96 min. And for the remainder it would be 13.60 min. That's close enough; it does use a ~ to represent approximation. Later on in the download things still seem to be reasonable:
[download] Destination: Friends Like These-6067128257001.fhls-2407-3.mp4
[download] 69.7% of ~769.57MiB at 1.26MiB/s ETA 03:34
The estimate of the file size has shrunken considerably, but now our times would be 3.08 minutes remaining of a total of 10.18 min. But when it's finished, I have:
[download] Destination: Friends Like These-6067128257001.fhls-2407-3.mp4
[download] 100% of 782.01MiB in 11:34
I assume that the issue here is that the bit rate of the file differs, so it can't guess any better.
But then, in almost every case, I get:
[download] Destination: Friends Like These-6067128257001.fdash-b08a9eed-57fb-4f85-aa43-32c239b86419-3.m4a
[download] 24.9% of ~70.63MiB at 872.22KiB/s ETA 07:47
In passing, what's this extension m4a? A bit of searching brought me to MPEG-4 Part 14: it's a pure audio file. That's interesting because of its size: the video is 720p, but the audio file is about 10% of the size of the video. I would have expected closer to 1%.
But the real issue here is that it's claiming an ETA of 7:47. That doesn't come close to calculating correctly: it should be about 60 of 81 seconds. But it continues like that:
60.8/81s[download] Destination: Friends Like These-6067128257001.fdash-b08a9eed-57fb-4f85-aa43-32c239b86419-3.m4a
[download] 23.1% of ~69.83MiB at 985.80KiB/s ETA 07:59
54.5/70.8
[download] Destination: Friends Like These-6067128257001.fdash-b08a9eed-57fb-4f85-aa43-32c239b86419-3.m4a
[download] 100% of 73.30MiB in 09:29
Why is that? I'd have to assume that the speeds are correct: why should the audio always come at a significantly lower rate than video? And the final time is also correct enough. In this case, it took nearly as long as the video.
| Wednesday, 24 June 2020 | Dereel | Images for 24 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Another PV recalibration
|
Topic: Stones Road house, general, opinion | Link here |
The Photovoltaic system went through another battery calibration overnight, from 20:44 yesterday until 6:00 this morning.
Is that acceptable? Absolutely. As I said in my letter of complaint on 25 March,
It wouldn't be such a problem if it happened during the night.
I am prepared, without obligation, to accept a single calibration cycle per month, but no more.
And that's almost exactly what has happened. Since then, there have been three recalibrations, each a single cycle and spaced a month apart. I can live with that. It's a pity it took them so long.
|
Protection from the elements
|
Topic: animals, photography | Link here |
|
|||||||||||||||||||||||
|
Understanding modern TV
|
Topic: multimedia, technology, opinion | Link here |
Spent some time today exploring the 7plus online video site, including signing up with them. All went relatively smoothly, but then I was offered a choice: display on your TV.
OK, I'll bite. This was displayed on my TV. Select the button and get:
|
|
|||||||||
What code? Nothing else of relevance was displayed on my TV. And I didn't get any email, of course.
Clearly this is Yet Another case where some group with collective tunnel vision have designed Yet Another “standard” that applies only to what they were thinking of, and TV broadcasters that jump on the bandwagon. And I can (at least for the moment) happily live without it. Presumably it relates to Smart TVs, but how? And if that's the case, presumably they're not expecting me to be running the browser on the TV. So where do they send the code? How do they establish the connection? And what advantage would it bring me?
I'm puzzled.
| Thursday, 25 June 2020 | Dereel | Images for 25 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Olympus stops building cameras
|
Topic: photography, opinion | Link here |
The news of the year: Olympus Corporation is selling its imaging business to Japan Industrial Partners, hopefully (their hope) by the end of the year.
What does that mean? According to the text of the press release,
By adding support from JIP, the NewCo, as the successor of reputable brands such as “OM-D” and “ZUIKO,” will utilize the innovative technology and unique product development capabilities which have been developed within Olympus, and will realize continuous growth of the business by bringing better products and services to the users and customers and by making itself a productive and rewarding work place for its employees.
Clearly NewCo is a variable for a name that hasn't been revealed yet. But if you take the statement at face value, things will continue. As before? They don't say that. But if they think they can continue as before, why sell?
Obviously there's a lot of gloom and despondency in the Olympus users' camp. How bad is it really? Between 2000 and 2010 camera technology developed enormously. Even between the Olympus E-1 (2003) and the Olympus E-3 (2007) there was great progress. Then they went mirrorless—that, too, was 11 years ago, with the Olympus E-P1. And the current mirrorless line, starting with the Olympus OM-D E-M1 Mark I, is 7 years old.
And since then? Yes, the E-M1 Mark II is a significant, if not great, improvement over the E-M1 Mark I. But the E-M1 Mark III isn't really that much of an improvement over the Mark II. And I'm still using cameras on a daily basis that are up to 12 years old. By the time my E-M1 Mark II is that old, it will be 2030. What will the industry look like then? A shadow of its former self, I suspect. I expect that a number of other manufacturers will throw in the towel, probably more definitively than Olympus. In the meantime, I have all the equipment that I need, so even if they stopped production immediately, I wouldn't have any immediate concerns.
And Tony Northrup? “I told you so”. No, Tony, you didn't. You said that they would stop production. No discussion of that yet. And your reasons are completely unrelated to the current issues. You were talking technology, badly. The real issues seem to be the erosion of the market by mobile phones and the lack of room for innovation in camera technology, not helped by the current COVID-19 situation.
One question is: to what extent did opinions like Tony's contribute to the poor performance of companies that he damned? I've already idly wondered whether he hasn't been paid to spout the nonsense that he does.
|
Frying eggs, revisited
|
Topic: food and drink, opinion | Link here |
I've had a bad relationship with frying eggs for decades. Over 40 years ago they cost me my job. So mainly Yvonne fries the eggs now.
But on some days, like today, I eat Singapore-style nasi lemak and need a fried egg.
It's been a while, but we have rings in which we can fry eggs, producing a nice circular result, just what I need on top of the rice of the nasi lemak:
|
|||||||||||||||||||||||
So I tried them out again. They're sticky! How did they get like that? Cleaned one and fried the egg:
|
|||||||||||||||||||||||
Problems:
So I think that they're still not the right solution.
|
Fixing the gate opener, finally
|
Topic: Stones Road house, general, opinion | Link here |
I've been meaning to adjust the gate opener for 3 weeks now, but the weather or something else always got in the way. Not today. Finally out there with the tools I bought two weeks ago. Disconnect the connector from the opener to the gate and move the bracket by the width between the screws. Try out. Yes, more than closes. OK, drill a third hole. Yvonne had picked up a high-power cordless drill at ALDI last week. Try the 10.5 mm drill bit with that.
After about a minute there was a mark on the metal. OK, sending a boy to do a man's job? Found a couple of extension cables and brought out my mains electric drill. What a difference! Yes, I'm sure there are plenty of things that I can do with the cordless drill, but this is clearly not one of them.
|
|||||||||||||||||||||||
Then a bit of adjustment of the end switch, and voilà!
|
|||||||||||||||||||||||
Now wouldn't that have been so much easier if the instructions had been correct?
|
Finish that tiwi!
|
Topic: technology, opinion | Link here |
I've been dragging my heels with tiwi due to the numerous issues I've had. But teevee made it clear once again that I need to finish it:
Jun 25 21:24:40 teevee kernel: Timeout initializing vt_vga
Jun 25 21:24:40 teevee kernel: panic: page fault
This time the backtrace was slightly different:
(kgdb) bt
#0 0xffffffff80bf068f in sched_switch ()
#1 0xffffffff80bcaba7 in mi_switch ()
#2 0xffffffff80c19b53 in sleepq_catch_signals ()
#3 0xffffffff80c19e44 in sleepq_timedwait_sig ()
#4 0xffffffff80bca5b2 in _sleep ()
#5 0xffffffff80bd5f22 in kern_clock_nanosleep ()
#6 0xffffffff80bd60cf in sys_nanosleep ()
#7 0xffffffff81091e87 in amd64_syscall ()
#8 <signal handler called>
And what was that first message? “Timeout initializing vt_vga”? Could that be a smoking gun? The log files from previous panics are gone, and it seems that the modern version of kgdb is a shadow of its former self, but putting the dump of last Wednesday through less showed me: yes, also the same message just before the panic. And so did the previous one.
So: smoking gun or normal message under these circumstances? Given that it always seems to happen when a browser is changing a window, I'm beginning to think that it isn't a coincidence. In any case, the next step is clear: get tiwi finished, even if it means going back to the same kernel as teevee, and see if things happen there too.
| Friday, 26 June 2020 | Dereel | Images for 26 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Another grid power outage
|
Topic: Stones Road house, general | Link here |
Another 1 second grid power failure at 0:44:30.
|
tiwi: progress
|
Topic: technology, opinion | Link here |
As planned, back to look at my issues with tiwi today. I had three main issues:
I had some ideas about how to fix each of them. And I made some progress.
With the key remapping, it occurred to me that I had had similar problems with teevee, probably due to my lack of understanding of how xmodmap works. There are two basic ways to approach it. Firstly, the way I have been doing it for decades:
keysym Alt_L = Meta_L Meta_L
remove Lock = Caps_Lock
remove Control = Control_L
remove Mod1 = Meta_L
keysym Control_L = Meta_L
keysym Caps_Lock = Control_L
add Mod1 = Meta_L
add Control = Control_L
keycode 64 = grave asciitilde
keycode 147 = Multi_key
That's the entire file. Apart from being short, it has the advantage that it's relatively easy to understand. The disadvantage is that it's not repeatable: it refers to current bindings, and they change. The alternative is the output of xmodmap -pk, which starts like this:
keycode 8 =
keycode 9 = Escape NoSymbol Escape
keycode 10 = 1 exclam 1 exclam
keycode 11 = 2 at 2 at
keycode 12 = 3 numbersign 3 numbersign
...
There's one entry for each key, and the reference is to the key number, not the old binding. But it seems that this doesn't work right. Tried it out today: yes, the old method works. One down, two to go.
But why doesn't the new method work? I think it's RTFM time. Well, not yet, but some time. My guess is that the Remove and Add directives are still necessary, but I need to understand how to adapt them.
Next, the mouse middle button emulation. Here, too, I had an idea: I was running moused, which had given me problems in the past. OK, stop X, stop moused, startx.
Oh. Still no middle button emulation.
Somehow there seems to be more than one problem here, possibly up to three. First, since the end of May the moused emulation doesn't feed through to X. It works fine elsewhere. Then there was the issue with the configuration file: does Emulate3Buttons still have any meaning? Last month it did, but since last week, probably as the result of upgrading to a newer version of X, that doesn't seem to make any difference either. Somehow I'm baffled. I'll have to get used to using a bigger mouse with a side button.
One down, one failed, one to go.
Finally, configure X correctly for my Radeon card. Here, too, I had ideas: I still had the nvidia driver installed, and that had bent some links, like the one to libglx, as I found a few days ago. OK, remove the driver; even if I go back to using an Nvidia card, there's a good chance that I'll be using a different driver.
Start again. Try to display a video:
=== grog@tiwi (/dev/pts/4) /spool/Docco 5 -> mpv AlJaz-15
/usr/local/bin/mpv --write-filename-in-watch-later-config AlJaz-15
(+) Video --vid=1 (*) (h264 1920x1080 30.000fps)
(+) Audio --aid=1 (*) (aac 2ch 48000Hz)
error: XDG_RUNTIME_DIR not set in the environment.
libEGL warning: DRI2: could not open /dev/dri/card0 (Permission denied)
[vo/gpu/opengl] Suspected software renderer or indirect context.
libGL error: failed to open drm device: Permission denied
libGL error: failed to load driver: r600
[vo/gpu/opengl] Suspected software renderer or indirect context.
error: XDG_RUNTIME_DIR not set in the environment.
[vo/gpu/vulkan/libplacebo] EnumeratePhysicalDevices(inst, &num, NULL): VK_ERROR_INITIALIZATION_FAILED
[vo/gpu/vulkan/libplacebo] Found no suitable device, giving up.
[vo/gpu/vulkan/libplacebo] Failed initializing vulkan device
Failed to open VDPAU backend Shared object "libvdpau_r600.so" not found, required by "mpv"
[vo/vdpau] Error when calling vdp_device_create_x11: 1
[vo/xv] Warning: this legacy VO has bad quality and performance, and will in particular result in blurry OSD and subtitles. You should fix your graphics drivers, or not force the xv VO.
AO: [oss] 48000Hz stereo 2ch s32
VO: [xv] 1920x1080 yuv420p
OK, time to look at these messages more carefully. What's XDG_RUNTIME_DIR? I didn't need that for the Nvidia version. And DRI2? /dev/dri? Yes, it's there:
=== grog@tiwi (/dev/pts/4) /spool/Docco 6 -> l /dev/dri
total 0
crw-rw---- 1 root 44 0x8c 26 Jun 12:51 card0
crw-rw---- 1 root 44 0x8b 26 Jun 12:51 controlD64
What's that? teevee doesn't have the devices. And why is it assigned to (non-existent) group 44? Looking at the distribution /etc/group, that's video, which makes sense. But why did my system upgrade not install it, or prompt me to merge it?
Then a couple of yellow messages:
[vo/gpu/opengl] Suspected software renderer or indirect context.
[vo/gpu/opengl] Suspected software renderer or indirect context.
Isn't that helpful, even if you can read it? I have no idea what that means.
Then a red vulkan, whatever that is, and
Failed to open VDPAU backend Shared object "libvdpau_r600.so" not found, required by "mpv"
OK, I can check for that. Yes, I have a number of libvdpau_r600.so:
=== grog@tiwi (/dev/pts/5) /spool/Docco 5 -> locate libvdpau_r600.so
/compat/linux/usr/lib/vdpau/libvdpau_r600.so.1
/compat/linux/usr/lib/vdpau/libvdpau_r600.so.1.0
/compat/linux/usr/lib/vdpau/libvdpau_r600.so.1.0.0
/compat/linux/usr/lib64/vdpau/libvdpau_r600.so.1
/compat/linux/usr/lib64/vdpau/libvdpau_r600.so.1.0
/compat/linux/usr/lib64/vdpau/libvdpau_r600.so.1.0.0
But they're for Linux. And what are they? If mpv wants it, why didn't the problem show up with the nvidia driver? But no, ldd mpv denies all interest. It must be required by a dependent library.
Then another yellow message:
[vo/xv] Warning: this legacy VO has bad quality and performance, and will in particular result in blurry OSD and subtitles. You should fix your graphics drivers, or not force the xv VO.
Legacy? I don't want no steenking legacy, and I certainly didn't ask for it.
The funny thing is that mpv ran just fine anyway. But clearly that's not a configuration you can live with.
But wait, you wouldn't put all your basket, would you? Why not put some of it in the system log instead?
Jun 26 14:41:02 tiwi kernel: info: [drm] Initialized drm 1.1.0 20060810
Jun 26 14:41:02 tiwi kernel: drmn0: =======================================================
Jun 26 14:41:02 tiwi kernel: drmn0: This code is obsolete abandonware. Install the graphics/drm-legacy-kmod pkg
Jun 26 14:41:02 tiwi kernel: drmn0: =======================================================
Jun 26 14:41:02 tiwi kernel: drmn0: Deprecated code (to be removed in FreeBSD 13): drm2 drivers
Jun 26 14:41:02 tiwi kernel: drmn0: =======================================================
Jun 26 14:41:02 tiwi kernel: drmn0: This code is obsolete abandonware. Install the graphics/drm-legacy-kmod pkg
Jun 26 14:41:02 tiwi kernel: drmn0: =======================================================
Jun 26 14:41:02 tiwi kernel: drmn0: Deprecated code (to be removed in FreeBSD 13): drm2 drivers
Jun 26 14:41:02 tiwi kernel: drmn0: <Cedar PRO [Radeon HD 5450/6350]> on vgapci0
Jun 26 14:41:02 tiwi kernel: info: [drm] RADEON_IS_PCIE
Jun 26 14:41:02 tiwi kernel: info: [drm] initializing kernel modesetting (CEDAR 0x1002:0x68F9 0x1462:0x2181).
Jun 26 14:41:02 tiwi kernel: info: [drm] register mmio base: 0xF7E20000
Jun 26 14:41:02 tiwi kernel: info: [drm] register mmio size: 131072
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon_atrm_get_bios: ===> Try ATRM...
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon_atrm_get_bios: pci_find_class() found: 0:1:0:0, vendor=1002, device=68f9
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon_atrm_get_bios: Get ACPI device handle
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon_atrm_get_bios: Get ACPI handle for "ATRM"
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon_atrm_get_bios: Failed to get "ATRM" handle: AE_NOT_FOUND
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon_acpi_vfct_bios: ===> Try VFCT...
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon_acpi_vfct_bios: Get "VFCT" ACPI table
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon_acpi_vfct_bios: Failed to get "VFCT" table: AE_NOT_FOUND
Jun 26 14:41:02 tiwi kernel: info: [drm] igp_read_bios_from_vram: ===> Try IGP's VRAM...
Jun 26 14:41:02 tiwi kernel: info: [drm] igp_read_bios_from_vram: VRAM base address: 0xe0000000
Jun 26 14:41:02 tiwi kernel: info: [drm] igp_read_bios_from_vram: Map address: 0xfffff800e0000000 (262144 bytes)
Jun 26 14:41:02 tiwi kernel: info: [drm] igp_read_bios_from_vram: Incorrect BIOS signature: 0xFFFF
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon_read_bios: ===> Try PCI Expansion ROM...
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon_read_bios: Map address: 0xfffff800000c0000 (131072 bytes)
Jun 26 14:41:02 tiwi kernel: info: [drm] ATOM BIOS: Cedar
Jun 26 14:41:02 tiwi kernel: drmn0: info: VRAM: 512M 0x0000000000000000 - 0x000000001FFFFFFF (512M used)
Jun 26 14:41:02 tiwi kernel: drmn0: info: GTT: 512M 0x0000000020000000 - 0x000000003FFFFFFF
Jun 26 14:41:02 tiwi kernel: info: [drm] Detected VRAM RAM=512M, BAR=256M
Jun 26 14:41:02 tiwi kernel: info: [drm] RAM width 64bits DDR
Jun 26 14:41:02 tiwi kernel: [TTM] Zone kernel: Available graphics memory: 4125100 kiB
Jun 26 14:41:02 tiwi kernel: [TTM] Zone dma32: Available graphics memory: 2097152 kiB
Jun 26 14:41:02 tiwi kernel: [TTM] Initializing pool allocator
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon: 512M of VRAM memory ready
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon: 512M of GTT memory ready.
Jun 26 14:41:02 tiwi kernel: info: [drm] Supports vblank timestamp caching Rev 1 (10.10.2010).
Jun 26 14:41:02 tiwi kernel: info: [drm] Driver supports precise vblank timestamp query.
Jun 26 14:41:02 tiwi kernel: info: [drm] MSI enabled 1 message(s)
Jun 26 14:41:02 tiwi kernel: drmn0: info: radeon: using MSI.
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon: irq initialized.
Jun 26 14:41:02 tiwi kernel: info: [drm] GART: num cpu pages 131072, num gpu pages 131072
Jun 26 14:41:02 tiwi kernel: info: [drm] probing gen 2 caps for device 8086:c01 = 3/e
Jun 26 14:41:02 tiwi kernel: info: [drm] enabling PCIE gen 2 link speeds, disable with radeon.pcie_gen2=0
Jun 26 14:41:02 tiwi kernel: info: [drm] Loading CEDAR Microcode
Jun 26 14:41:02 tiwi kernel: info: [drm] PCIE GART of 512M enabled (table at 0x0000000000040000).
Jun 26 14:41:02 tiwi kernel: drmn0: info: WB enabled
Jun 26 14:41:02 tiwi kernel: drmn0: info: fence driver on ring 0 use gpu addr 0x0000000020000c00 and cpu addr 0x0xfffff8006e169c00
Jun 26 14:41:02 tiwi kernel: drmn0: info: fence driver on ring 3 use gpu addr 0x0000000020000c0c and cpu addr 0x0xfffff8006e169c0c
Jun 26 14:41:02 tiwi kernel: info: [drm] ring test on 0 succeeded in 1 usecs
Jun 26 14:41:02 tiwi kernel: info: [drm] ring test on 3 succeeded in 1 usecs
Jun 26 14:41:02 tiwi kernel: info: [drm] ib test on ring 0 succeeded in 0 usecs
Jun 26 14:41:02 tiwi kernel: info: [drm] ib test on ring 3 succeeded in 0 usecs
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon_device_init: Taking over the fictitious range 0xe0000000-0xf0000000
Jun 26 14:41:02 tiwi kernel: radeon_iicbb0 on drmn0
Jun 26 14:41:02 tiwi kernel: iicbus0: <Philips I2C bus> on iicbb0 addr 0xff
Jun 26 14:41:02 tiwi kernel: iic0: <I2C generic I/O> on iicbus0
Jun 26 14:41:02 tiwi kernel: radeon_iicbb1 on drmn0
Jun 26 14:41:02 tiwi kernel: iicbus1: <Philips I2C bus> on iicbb1 addr 0xff
Jun 26 14:41:02 tiwi kernel: iic1: <I2C generic I/O> on iicbus1
Jun 26 14:41:02 tiwi kernel: radeon_iicbb2 on drmn0
Jun 26 14:41:02 tiwi kernel: iicbus2: <Philips I2C bus> on iicbb2 addr 0xff
Jun 26 14:41:02 tiwi kernel: iic2: <I2C generic I/O> on iicbus2
Jun 26 14:41:02 tiwi kernel: radeon_iicbb3 on drmn0
Jun 26 14:41:02 tiwi kernel: iicbus3: <Philips I2C bus> on iicbb3 addr 0xff
Jun 26 14:41:02 tiwi kernel: iic3: <I2C generic I/O> on iicbus3
Jun 26 14:41:02 tiwi kernel: radeon_iicbb4 on drmn0
Jun 26 14:41:02 tiwi kernel: iicbus4: <Philips I2C bus> on iicbb4 addr 0xff
Jun 26 14:41:02 tiwi kernel: iic4: <I2C generic I/O> on iicbus4
Jun 26 14:41:02 tiwi kernel: radeon_iicbb5 on drmn0
Jun 26 14:41:02 tiwi kernel: iicbus5: <Philips I2C bus> on iicbb5 addr 0xff
Jun 26 14:41:02 tiwi kernel: iic5: <I2C generic I/O> on iicbus5
Jun 26 14:41:02 tiwi kernel: radeon_iicbb6 on drmn0
Jun 26 14:41:02 tiwi kernel: iicbus6: <Philips I2C bus> on iicbb6 addr 0xff
Jun 26 14:41:02 tiwi kernel: iic6: <I2C generic I/O> on iicbus6
Jun 26 14:41:02 tiwi kernel: radeon_iicbb7 on drmn0
Jun 26 14:41:02 tiwi kernel: iicbus7: <Philips I2C bus> on iicbb7 addr 0xff
Jun 26 14:41:02 tiwi kernel: iic7: <I2C generic I/O> on iicbus7
Jun 26 14:41:02 tiwi kernel: drm_iic_dp_aux0 on drmn0
Jun 26 14:41:02 tiwi kernel: info: [drm] Radeon Display Connectors
Jun 26 14:41:02 tiwi kernel: info: [drm] Connector 0:
Jun 26 14:41:02 tiwi kernel: info: [drm] DP-1
Jun 26 14:41:02 tiwi kernel: info: [drm] HPD2
Jun 26 14:41:02 tiwi kernel: info: [drm] DDC: 0x6460 0x6460 0x6464 0x6464 0x6468 0x6468 0x646c 0x646c
Jun 26 14:41:02 tiwi kernel: info: [drm] Encoders:
Jun 26 14:41:02 tiwi kernel: info: [drm] DFP1: INTERNAL_UNIPHY1
Jun 26 14:41:02 tiwi kernel: info: [drm] Connector 1:
Jun 26 14:41:02 tiwi kernel: info: [drm] DVI-I-1
Jun 26 14:41:02 tiwi kernel: info: [drm] HPD4
Jun 26 14:41:02 tiwi kernel: info: [drm] DDC: 0x6450 0x6450 0x6454 0x6454 0x6458 0x6458 0x645c 0x645c
Jun 26 14:41:02 tiwi kernel: info: [drm] Encoders:
Jun 26 14:41:02 tiwi kernel: info: [drm] DFP2: INTERNAL_UNIPHY
Jun 26 14:41:02 tiwi kernel: info: [drm] CRT1: INTERNAL_KLDSCP_DAC1
Jun 26 14:41:02 tiwi kernel: info: [drm] Internal thermal controller with fan control
Jun 26 14:41:02 tiwi kernel: info: [drm] radeon: power management initialized
Jun 26 14:41:02 tiwi kernel: info: [drm] Connector DP-1: get mode from tunables:
Jun 26 14:41:02 tiwi kernel: info: [drm] - kern.vt.fb.modes.DP-1
Jun 26 14:41:02 tiwi kernel: info: [drm] - kern.vt.fb.default_mode
Jun 26 14:41:02 tiwi kernel: info: [drm] Connector DVI-I-1: get mode from tunables:
Jun 26 14:41:02 tiwi kernel: info: [drm] - kern.vt.fb.modes.DVI-I-1
Jun 26 14:41:02 tiwi kernel: info: [drm] - kern.vt.fb.default_mode
Jun 26 14:41:02 tiwi kernel: info: [drm] fb mappable at 0xE0142000
Jun 26 14:41:02 tiwi kernel: info: [drm] vram apper at 0xE0000000
Jun 26 14:41:02 tiwi kernel: info: [drm] size 8294400
Jun 26 14:41:02 tiwi kernel: info: [drm] fb depth is 24
Jun 26 14:41:02 tiwi kernel: info: [drm] pitch is 7680
Jun 26 14:41:02 tiwi kernel: fbd0 on drmn0
Jun 26 14:41:02 tiwi kernel: VT: Replacing driver "vga" with new "fb".
Jun 26 14:41:02 tiwi kernel: info: [drm] Initialized radeon 2.29.0 20080528 for drmn0 on minor 0
What diarrhoea! On checking, it appeared at boot time. I got as far as “abandonware”, considered the “legacy” xv, and wondered whether I'm not barking up the wrong tree. Years ago I went through enough effort installing the Nvidia drivers. Should I maybe give up the attempt to migrate to Radeon? It might work better if I installed from scratch, but clearly upgrades are still not what they should be.
One down, two failed.
|
Hibiscus in winter
|
Topic: gardening | Link here |
As suspected, my Hibiscus rosa-sinensis “Uncle Max” is still flowering:
|
|||||||||||||||||||||||
The bush is showing signs of stress, though:
|
|||||||||||||||||||||||
But the Alyogyne huegelii to its left is still flowering profusely:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Maybe I should plant more of them.
| Saturday, 27 June 2020 | Dereel | Images for 27 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
eureka hang
|
Topic: technology, opinion | Link here |
Into the office as usual this morning and pressed the Shift key on my keyboard. I like observing which monitor lights up first. The only thing I can be sure of is that monitor 3 (of 4) will be last, but for reasons I'm still trying to understand, the others don't always come up in the same sequence.
Today, though, they were all the same: nothing happened. Damn. Has the system crashed? Looked at the LEDs on the front panel. Both on, power and disk. That's a clue, I suppose.
Off to teevee to see if I could contact eureka. Yes, it wasn't down, but very slow.
I've seen this before in the morning. For reasons that I don't understand, but which seem to have something to do with the virtual memory system, the system is very unresponsive for about the first 5 to 10 minutes after I start using it. During this time the big memory users (browsers and MediathekView) swap enormous quantities of memory, probably over 1 GB.
One of the strange things was that iftop was using most CPU time.
But today there wasn't that much swapping. I found my nightly backup backing up a temporary directory that I had set up yesterday—about 400 GB of data—and forgotten to mark as “no backup”. OK, not a problem, I can shoot down that backup.
But it didn't help. Stop a firefox? OK. And look at top's summary:
last pid: 50141; load averages: 0.28, 0.64, 0.94 up 290+01:03:38 10:03:32
1355 processes:1 running, 1343 sleeping, 3 stopped, 8 zombie
CPU: % user, % nice, % system, % interrupt, % idle
Mem: 5542M Active, 22G Inact, 3158M Wired, 348M Cache, 1655M Buf, 1640K Free
Swap: 20G Total, 13G Used, 7566M Free, 63% Inuse
This is a machine with 32 GB of memory, which I would have thought was more than enough. Why is so much memory inactive? If only half as much (still too much, I think) were inactive, we wouldn't need to swap.
Somehow I suspected memory leaks. Could it be X? No, the xterms on teevee were also slow. OK, let's shoot down the MediathekView, which uses java and thus is a natural suspect for memory leaks:
=== root@eureka (/dev/pts/11) ~ 832 -> ps aux | grep java
USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
grog 76975 0.0 5.6 10628260 1887972 v0 I 27Oct19 2737:40.49 /usr/local/openjdk8/bin/java -DproxySet=true -DproxyHo
But it only had less than 2 GB of over 10 GB resident. And it took nearly a minute to stop, probably because it was first swapping in.
Next, the other 5 browsers. That didn't take as long (with killall), only about 5 seconds. Does rdesktop use a lot of memory? It shouldn't be running anyway (just keeps the remote machine from hibernating), so down it goes. And at the end I had:
last pid: 52049; load averages: 0.41, 0.52, 0.72 up 290+01:15:17 10:15:11
1328 processes:1 running, 1323 sleeping, 3 stopped, 1 zombie
CPU: 0.0% user, 0.0% nice, 0.8% system, 0.4% interrupt, 98.8% idle
Mem: 1168M Active, 27G Inact, 2956M Wired, 333M Cache, 1655M Buf, 4040K Free
Swap: 20G Total, 7090M Used, 13G Free, 34% Inuse
All but 5 GB of RAM inactive! Why is that? I wish I understood the VM system.
Nothing seemed to help. Even the backups went far more slowly. Yesterday's dump showed that it took 8 minutes to back up my home directory, and about another 7 minutes to back up the rest of the disk:
=== root@eureka (/dev/pts/11) /photobackup/spool 853 -> l /backups/eureka-FreeBSD/2/tars.0/home/
total 2307
-rw-r--r-- 1 root wheel 229 25 Jun 21:02 dump.tar.bz2
-rw-r--r-- 1 root wheel 1,726,786,852 25 Jun 21:08 grog.tar.bz2
...
-rw-r--r-- 1 root wheel 255,735 25 Jun 21:15 yvonne.tar.bz2
But today it took much longer. Watching during the backup I saw successive data:
-rw-r--r-- 1 root wheel 1,198,990,096 27 Jun 11:19 grog.tar.bz2
...
-rw-r--r-- 1 root wheel 1,750,097,802 27 Jun 11:27 grog.tar.bz2
And after that the dump ran for nearly another hour:
-rw-r--r-- 1 root wheel 255,718 27 Jun 12:21 yvonne.tar.bz2
What's causing that? Is the disk dying? About the only other thing going on was copying my temporary partition to a different external disk. Could it be that the USB subsystem is the bottleneck? After the backups were done, things seemed to return to normal.
|
Buddleja flower
|
Topic: gardening | Link here |
As expected, one of my indoor Buddleja × weyeriana cuttings is flowering:
|
|||||||||||||||||||||||
Now where do I plant it outside so that it doesn't die?
|
Freeze-dried veal
|
Topic: food and drink, opinion | Link here |
Scaloppine al limone for dinner tonight. And I had found a small quantity of veal in the deep freezer:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Not even 7 years. It should still be OK. But where did all that ice come from?
|
|||||||||||||||||||||||
From the meat, it seems. It weighed 85 g when it was frozen, but now it was only 30 g, light as a feather. And the ice weighed 48 g, so something like 7 g must have escaped to the environment. Put the meat in some water to rehydrate it, but Yvonne decided that it was better for the dogs, and I didn't even get to weigh it after rehydration.
| Sunday, 28 June 2020 | Dereel | Images for 28 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Cold weather
|
Topic: general | Link here |
Another cold night tonight, with considerable frost and a minimum temperature of -0.7°. After getting up, by which time the temperature had reached 6°, measured the ground temperature and found areas as cold as -3.7°. That's still far from the record of measured -3.0° on 28 August 2018.
|
Still more tiwi surprises
|
Topic: technology, opinion | Link here |
So, where am I with tiwi? The mouse doesn't work as I want it to, and mpv vomits over the screen when started.
But apart from that, mpv works, and I can work around the mouse issue. On the other hand, the reason I bought the new machine was to help corner the continual panics that I was getting from teevee. So why not just connect it up to the TV and see how it works?
How I hate messing around with furniture! I need to drill a hole in the back of the TV cabinet and feed a number of cables through it. First let's kludge it and hang them round the front. Put in a switch for the two machines, connect up power, HDMI, ...
HDMI? The plug doesn't fit:
|
|||||||||||||||||||||||
I was wrong the week before last when I claimed that the card has an HDMI output. It doesn't: it's a DisplayPort socket.
Oh. So much for that idea. Clearly there's not much point in continuing along that path. Where's another Nvidia display card? Found one and put it in, to discover:
vgapci0@pci0:4:0:0: class=0x030000 card=0x83541043 chip=0x0a6510de rev=0xa2 hdr=0x00
vendor = 'NVIDIA Corporation'
device = 'GT218 [GeForce 210]'
class = display
subclass = VGA
OK, what driver do I need for that? Spent yet more time trying to fight my way through the Nvidia documentation before finally discovering that it's the 340 series drivers—older than the card in teevee. Downloaded the driver, loaded it and... it worked. Yes, of course, it should, but nowadays that seems to have become a rarity.
OK, does it work nicely with mpv?
=== grog@tiwi (/dev/pts/7) /spool/Docco 2 -> mmp AlJaz-15
/usr/local/bin/mpv --write-filename-in-watch-later-config AlJaz-15
(+) Video --vid=1 (*) (h264 1920x1080 30.000fps)
(+) Audio --aid=1 (*) (aac 2ch 48000Hz)
error: XDG_RUNTIME_DIR not set in the environment.
[vo/gpu/x11] X11 error: BadValue (integer parameter out of range for operation)
[vo/gpu/x11] Type: 0, display: 0x80d401000, resourceid: 0, serial: 55
[vo/gpu/x11] Error code: 2, request code: 98, minor code: 3
[vo/gpu] Could not create GLX context!
error: XDG_RUNTIME_DIR not set in the environment.
[vo/gpu/vulkan/libplacebo] EnumeratePhysicalDevices(inst, &num, NULL): VK_ERROR_INITIALIZATION_FAILED
[vo/gpu/vulkan/libplacebo] Found no suitable device, giving up.
[vo/gpu/vulkan/libplacebo] Failed initializing vulkan device
[vo/vdpau/x11] X11 error: BadRequest (invalid request code or no such operation)
[vo/vdpau/x11] Type: 0, display: 0x80d573000, resourceid: 11b, serial: 32
[vo/vdpau/x11] Error code: 1, request code: 97, minor code: 1
[vo/vdpau] Error when calling vdp_device_create_x11: 1
[vo/xv] Warning: this legacy VO has bad quality and performance, and will in particular result in blurry OSD and subtitles. You should fix your graphics drivers, or not force the xv VO.
AO: [oss] 48000Hz stereo 2ch s32
VO: [xv] 1920x1080 yuv420p
[vo/xv] Shared memory not supported
[vo/xv] Reverting to normal Xv.
[vo/xv] Shared memory not supported
[vo/xv] Reverting to normal Xv.
AV: 00:00:12 / 00:32:04 (0%) A-V: -0.000 Dropped: 3 Cache: 467s/150MB
Oh. That looks horribly familiar, but in fact the messages are only partially the same. But clearly I have broken something by messing around in the X installation. What now? The best is to start again from the state of teevee, replace the nvidia driver, and try again. Mañana.
| Monday, 29 June 2020 | Dereel | Images for 29 June 2020 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Another cold night
|
Topic: Stones Road house, general | Link here |
Another night with frost.
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The lowest temperature measured by the weather station was +0.1°, 0.8° higher than yesterday, but I got up earlier and watched the air conditioner struggle. Despite being set to 19° in the lounge room, the temperature was only 17°. I get the impression that the efficiency of the heating drops rapidly with ambient temperatures below 1°. And of course the air conditioner housing was covered in ice:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Outside again with the infrared thermometer. This time I measured down to -5° in a couple of places.
|
More curry tree pain
|
Topic: gardening | Link here |
After the treatment over the last 6 months, my curry tree is finally free of mites.
Or is it? No!
|
|||||||||||||||||||||||
Where do the bloody things come from? I've already established that the recommended treatment with soapy water is inappropriate, so it's back to using pyrethrum. Also cut off the affected branches:
|
|||||||||||||||||||||||
After washing, I can use them for cooking.
|
The daily tiwi pain
|
Topic: technology, opinion | Link here |
I've been trying all sorts of different approaches to getting tiwi working correctly, to the point where I'm no longer completely sure what I did where. Time for a log. Today's entries in that log, slightly condensed:
Clearly the Radeon card was a dead end, so I installed Nvidia graphics card:
=== root@tiwi (/dev/pts/7) ~ 5 -> pciconf -l -v vgapci0
vgapci0@pci0:4:0:0: class=0x030000 card=0x83541043 chip=0x0a6510de rev=0xa2 hdr=0x00
vendor = 'NVIDIA Corporation'
device = 'GT218 [GeForce 210]'
class = display
subclass = VGA
Boot the teevee kernel of 2 April and install the correct driver for GeForce 210:
=== root@tiwi (/dev/pts/0) ~ 4 -> pkg delete nvidia-driver-390
Checking integrity... done (0 conflicting)
Deinstallation has been requested for the following 1 packages (of 0 packages in the universe):
Installed packages to be REMOVED:
nvidia-driver-390: 390.132
Number of packages to be removed: 1
Proceed with deinstalling packages? [Y/n]: y
[1/1] Deinstalling nvidia-driver-390-390.132...
[1/1] Deleting files for nvidia-driver-390-390.132: 100%=== root@tiwi (/dev/pts/0) ~ 5 -> pkg install nvidia-driver-340
The following 2 package(s) will be affected (of 0 checked):
New packages to be INSTALLED:
gnatcross-sysroot-aarch64: 1_1
nvidia-driver-340: 340.108
Number of packages to be installed: 2
After installation:
=== root@tiwi (/dev/pts/0) ~ 8 -> kldunload nvidia=== root@tiwi (/dev/pts/0) ~ 9 -> kldload nvidia
kldload: can't load nvidia: module already loaded or in kernel=== root@tiwi (/dev/pts/0) ~ 10 -> kldunload nvidia
kldunload: can't unload file: Device busy=== root@tiwi (/dev/pts/0) ~ 11 ->
The problem here was the second module, nvidia-modeset. Remove that and nvidia goes too:
=== root@tiwi (/dev/pts/0) ~ 12 -> kldunload nvidia-modeset=== root@tiwi (/dev/pts/0) ~ 13 -> kldunload nvidia
kldunload: can't find file nvidia=== root@tiwi (/dev/pts/0) ~ 14 -> kldstat
Id Refs Address Size Name
1 39 0xffffffff80200000 226f308 kernel
3 4 0xffffffff8391a000 95c0 linux_common.ko
4 1 0xffffffff83924000 b72b8 linux.ko
5 1 0xffffffff839dc000 26e00 fuse.ko
7 1 0xffffffff83d21000 496c linprocfs.ko
8 1 0xffffffff83d26000 a4b8 if_ed.ko
9 1 0xffffffff83d31000 2968 ums.ko
10 1 0xffffffff83d34000 1880 uhid.ko
11 1 0xffffffff83d36000 35490 linux64.ko
No nvidia-modeset any more. I have a vague recollection that it didn't appear until driver set 390. But after that, the card was recognized:
nvidia-modeset: Unloading
nvidia0: <GeForce 210> on vgapci0
vgapci0: child nvidia0 requested pci_enable_io
vgapci0: child nvidia0 requested pci_enable_io
And that worked. No error messages with mpv, and mouse emulation works correctly. So: this is not a hardware issue. Something in mouse support has broken in the past few months.
What is it? I almost think that it's a conspiracy of kernel and X upgrades. But now I have a basis where it works, at least sort of, so I can investigate from there.
None of this directly addresses the panics I've been having on teevee. But if the hardware works, why not just use it? Swapped disks between tiwi and teevee and started from the old teevee disk with the new hardware. If the issue is with the display card, nvidia driver or the motherboard, it should now be gone, since all have been changed. And I can carry on checking the software on the old hardware.
Current status:
| Tuesday, 30 June 2020 | Dereel | Images for 30 June 2020 |
| Top of page | ||
| previous day |
|
tiwi continued
|
Topic: technology, opinion | Link here |
Didn't do much work on tiwi today. To my surprise, on booting I was greeted with:
Jun 30 11:55:15 tiwi savecore[844]: reboot after panic: vm_fault: fault on nofault entry, addr: 0xfffffe004c6e5000
Jun 30 11:55:15 tiwi savecore[844]: writing core to /var/crash/vmcore.5
And how about that, I then found:
Jun 29 14:48:48 tiwi kernel: panic: vm_fault: fault on nofault entry, addr: 0xfffffe004c6e5000
Jun 29 14:48:48 tiwi kernel: cpuid = 2
Jun 29 14:48:48 tiwi kernel: time = 1593405878
Jun 29 14:48:48 tiwi kernel: KDB: stack backtrace:
Jun 29 14:48:48 tiwi kernel: #0 0xffffffff80c0b3f5 at kdb_backtrace+0x65
Jun 29 14:48:48 tiwi kernel: #1 0xffffffff80bbf4ae at vpanic+0x17e
Jun 29 14:48:48 tiwi kernel: #2 0xffffffff80bbf323 at panic+0x43
Jun 29 14:48:48 tiwi kernel: #3 0xffffffff80ef68bc at vm_fault+0x248c
Jun 29 14:48:48 tiwi kernel: #4 0xffffffff80ef4310 at vm_fault_trap+0x60
Jun 29 14:48:48 tiwi kernel: #5 0xffffffff810914af at trap_pfault+0x19f
Jun 29 14:48:48 tiwi kernel: #6 0xffffffff810909b8 at trap+0x288
Jun 29 14:48:48 tiwi kernel: #7 0xffffffff8106a91c at calltrap+0x8
It's worth noting the time of the messages. This was directly before I swapped systems, so tiwi was still the ThinkCentre M93p with the GeForce 210 board. But this was supposed to be the stable machine. Took a look at the dump, but it said little of use. It looks as if I'm going to have to relearn or repair kgdb.
Things didn't stop there. When I tried to start X, it failed. Further investigation showed:
Jun 30 11:55:38 tiwi kernel: NVRM: API mismatch: the client has the version 340.108, but
Jun 30 11:55:38 tiwi kernel: NVRM: this kernel module has the version 390.132. Please
Jun 30 11:55:38 tiwi kernel: NVRM: make sure that this kernel module and all NVIDIA driver
Jun 30 11:55:38 tiwi kernel: NVRM: components have the same version.
On reflection, I hadn't installed the correct nvidia driver (version 390). I still had the 340 driver installed. So what related to a 390? Looking, I found:
=== root@tiwi (/dev/pts/0) /var/crash 18 -> kldstat
Id Refs Address Size Name
1 45 0xffffffff80200000 226f308 kernel
...
5 1 0xffffffff82559000 14a8ff8 nvidia.ko
6 1 0xffffffff83a02000 165770 nvidia-modeset.ko
What's that nvidia-modeset doing there? That belongs to the 390 driver, and it should have been long gone. Looking at /etc/loader.conf, I found:
# Depending on the version, one or the other of these needs loading.
# This sequence should work.
nvidia-modeset_load="YES"
nvidia_load="YES"
But that should work. On the 390 driver, it loads nvidia-modeset, which itself loads nvidia, so the second attempt fails. On the 340 driver, there is no nvidia-modeset, so it loads nvidia. The problem is that the nvidia-modeset was still there.
OK, unload everything, remove the 340 driver, install the 390 driver, and start. All well and good. Shoot down X with Ctrl-C from an xterm. Panic! On reboot, noted that it was exactly the same panic. Was there some corruption left behind by the driver mismatch? To be on the safe side, removed the nvidia packages and started again. At least it's running now.
|
Heavy winds
|
Topic: general, photography, opinion | Link here |
The cold, sunny, windstill weather has come to an end, and today we had quite a bit of wind. But I still wasn't expecting it when Yvonne came in and showed me her camera:
|
|||||||||||||||||||||||
That's not dust: it's sand. She had had the camera out on the tripod with the PIXIO “Robot Cameraman”, and the wind blew it over! That's the first time I've experienced that, though of course the combination of the relatively light tripod and the wind resistance of the PIXIO would have helped. Fortunately no damage was done, and it's worth noting that the tripod has a hook for a sandbag.
That wasn't the only effect of the winds. They blew over a rubbish bin:
|
|||||||||||||||||||||||
They also slightly opened the gate, which I've been observing carefully. Here from last week (seen from the other side) and today:
|
|
|||||||||
|
|||||||||||||||||||||||
Will I need to adjust it again?
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 |
