Bloody NBN again!
A while back I was severely angered by the National Broadband Network performing maintenance work that effectively took me off the net for a whole day. Fortunately?I thought?those days are over. But today I received mail from Aussie Broadband, who report, in their inimitable markup: NBN are planning network maintenance between **Mon 23rd September 2024 06:00 AEST** and **Tue 1st October 2024 18:00 AEST**, for **540 min**., for **480 min**., for **600 min**. NBN are planning network maintenance between **Wed 2nd October 2024 06:00 AEST** and **Fri 11th October 2024 18:00 AEDT**, for **600 min**.
Continuous reinvention: A brief history of block storage at AWS
Marc Olson, a long-time Amazonian, discusses the evolution of EBS, highlighting hard-won lessons in queueing theory, the importance of comprehensive instrumentation, and the value of incrementalism versus radical changes. It's an insightful look at how one of AWS?s foundational services has evolved to meet the needs of our customers.
Developing Affinity
Serif have brought out a new offer: 6 month extended free trial for their Affinity photo processing software. How can I refuse? Well, it took me several weeks and a reminder before I finally installed it. After all, I did try it 7 years ago and came to the conclusion that it wasn't for me. But 7 years is a long time, so it's worth trying again. And once again I'm presented with a new, confusing interface and confusing terminology: no longer do I process a photo, it's a ?document?. At least that presumably means that I won't have to ?develop?
Microsoft: unexpected success!
Overnight distress.lemis.com had to reboot, because Microsoft said so. When it came up again, it seems to have been surprised: Of course, everything's relative: === grog@hydra (/dev/pts/10) /dump/distress-Microsoft 151 -> l DISTRESS/Backup\ Set\ 2024-07-22\ 104007/ drwxr-xr-x 3 grog wheel 8,192 22 Jul 11:57 Backup Files 2024-07-22 104007 drwxr-xr-x 3 grog wheel 1,536 30 Jul 09:29 Backup Files 2024-07-30 091357 drwxr-xr-x 3 grog wheel 1,024 5 Aug 10:12 Backup Files 2024-08-05 100306 drwxr-xr-x 3 grog wheel 1,024 12 Aug 10:53 Backup Files 2024-08-12 104300 drwxr-xr-x 3 grog wheel 1,536 18 Aug 19:15 Backup Files 2024-08-18 190024 The last backup is 3 days old.
Selling the lawn mower
Spring is on its way, time to sell my old lawn mower: After the pain I have had selling things on eBay, I put the advertisement in Gumtree, the first time I have sold anything there. They don't make it easy, and after I finally submitted the advertisement (fourth attempt), I discovered that they had offered it for free! It seems that they don't like the $ symbol in their ads, and they're too polite to complain, so they just set the value to 0.
2024 Pollscrolling
The 2024 US election has, in the last few weeks, become the most interesting one I can recall. I?m pretty old, so that?s a strong statement; I can recall a lot of US elections. The Internet makes it way too easy to obsess over a story that?s this big and has this many people sharing opinions. Here is my opinion, not on who?s winning, but on how, with only a very moderate expenditure of time and money, you can be as well-informed as anybody in the world as to how it?s going. Disclosures: I?m not American, got no vote on this one.
Warning from lemis.com
Seen in today's mail: 11 + 19-08-2024 To groggyhimself@lemis.c ( 174) lemis.com + Security alert Huh? From: address (after the length) lemis.com? Still, let's read... Looks just like one of those silly Google warning that I keep complaining about. When did it happen? What IP address? About the only thing that I can be reasonably sure of is that I didn't log in with a MacBook Pro. It took a little while to sink in: it claims to come from lemis.com, specifically https://lemis.com/activity/notifications. That's simple: the URL doesn't exist, and I currently (still) don't do HTTPS.
?Wise? scam?
Interesting message my mail today: 88 + 16-08-2024 To groggyhimself@lem ( 593) W?i?s?e? + Ticket (AU-K421004) : Enable 2FA and Verify Your Details by August 20 Bloody 2FA again! Somehow ?Wise?, the money transfer people who used to have the too-descriptive name ?Transferwise?, get on my nerves. How do I even know if it's kosher? Is there anything on their web site about it? No, but a week ago they sent me mail offering a service that really does look OK, but there's nothing about that on their web site either.
Guillotining Greg
Baking bread today, which involves lining the bread form with baking paper. And what better to cut it with than with the guillotine that I bought for photographic purposes in December 1967? That's the way I always do it, but today, for some reason, I raised my hand inappropriately after putting the paper in place. After treatment, it looks like this: The mess on my thumbnail is adhesive from the plaster.
More proof of life
A letter from Info retraite today, apparently a successor institution of CNAV, the French pension fund, claiming « Le site officiel qui simplifie la retraite » (?the official site that simplifies retirement?) . They want to know if Yvonne is still alive. In principle that's normal, but in the past we have had untold communication problems. And once again the letter was nearly a month old?dated 27 July?though the post mark showed that it had been posted in France (Paris only two days later, so they must have found the cheapest postage method. On the whole, things look a lot more professional, but one thing was interesting: an on-line access, reminiscent of the German POSTIDENT.
The pain of upgrading
I've had hydra for 10 months now, and I'm almost done. But there are still a number of issues that I need to clarify, most of which don't have that much to do with the hardware, now that I have my monitors installed the way I want. By now eureka should have been degraded to Internet server roles such as mail and web servers. But I'm still using it for my day-to-day work. I can't use MediathekView on hydra because the newer versions have some weird Java (programming language) incompatibility. I still don't have a complete set of X fonts. VirtualBox doesn't work reliably, so one of the reasons for my ridiculous amount of memory has gone away.
Configuration: so nice, so nice, we do it twice
I still had time, so how about finally migrating teevee to fvwm version 3? Brought it into the office for that purpose, which proved to be helpful. And with only a couple of stumbles managed to get it up and running. Only somehow the .Xdefaults don't seem to work as specified. Have I forgotten to update a configuration file? Somehow included the wrong one? Now that the main work is done, I have time to check things through at my leisure.
More system upgrades
This flashing on hydra monitor 2 still isn't going away, so it's finally time to reconnect the monitors: swap monitor 0 (DVI via DisplayPort) and monitor 2 (HDMI). They have different resolutions, and I was expecting a bit of fun, so I started X server 4 to do the configuration on. I got my fun. Things froze completely. All X servers were in state D (short-term disk wait), but they didn't use any CPU time. The mouse was dead, and there was no way to switch away from the server. I could access the system from eureka, but I couldn't stop the X server.
Musk's X server issues
Today was another big day for Elon Musk, interviewing Donald Trump on TwitterX. But things didn't go according to plan. It seems that his X servers got a DDoS. I suppose that's a different problem from the ones I have with my X servers.
Basic Infrastructure
Recently, I was looking at the infrastructure bills for our CoSocial co-op member-owned Mastodon instance, mostly Digital Ocean and a bit of AWS. They seemed too high for what we?re getting. Which makes me think about the kind of infrastructure that a decentralized social network needs, and how to get it. I worked at AWS for 5½ years and part of my job was explaining why public-cloud infrastructure is a good idea. I had no trouble doing that because, for the people who are using it, it was (and is) a good idea.
More video flashing
This strange flashing (or flickering) on video output continues. So far I had only seen it on ZDF content, but this evening I saw it on another German channel. What's causing it? So far I have established: I have only ever seen it on our TV. It only happens on certain content. In the form I'm seeing now, it only happens on teevee, not tiwi. So what should I investigate now? teevee and tiwi are almost identical, both ThinkCentres running FreeBSD.
DFXP deciphered
More investigation into the DFXP subtitle format. To my surprise, Wikipedia redirects to TTML, where I read: Timed Text Markup Language (TTML), previously referred to as Distribution Format Exchange Profile (DFXP), Oh. I can do TTML. Why am I having trouble with DFXP? A quick look at it with Emacs is surprising: That doesn't look like any form of XML.
Countrywomen
In the last couple of weeks I?ve been at shows by Molly Tuttle and Sierra Ferrell (I recommend clicking both those links just for the front-page portraits). Herewith thoughts on the genres, performances, and sound quality. Tuttle is post-bluegrass and Ferrell is, um, well, Wikipedia says ?folk, bluegrass, gypsy jazz, and Latin styles? which, OK, I guess, but it doesn?t mention pure old-fashioned country, her strongest flavor. These days, ?Americana? is used to describe both these artists. The notion that Americana implies ?by white people? is just wrong, check out Rhiannon Giddens?
Converting subtitles
We have a series of videos with a new subtitle format, DFXP. And for some reason mpv can't understand that. There must be some program that can handle them. A web search found a number of online services, presumably requiring you to upload and download every file manually. There must be an easier way. Yes! Netflix-to-srt, as they title it, claims to do exactly that. OK, download and try: === grog@teevee (/dev/pts/6) /spool/Series/The-Marvelous-Mrs-Maisel/01 7 -> python3 /usr/local/bin/to_srt.py Traceback (most recent call last): File "/usr/local/bin/to_srt.py", line 216, in <module> main() File "/usr/local/bin/to_srt.py", line 212, in main f.write(to_srt(text, fn[-4:])) ^^^^^^^^^^^^^^^^^^^^^ File "/usr/local/bin/to_srt.py", line 73, in to_srt return xml_to_srt(text) ^^^^^^^^^^^^^^^^ File "/usr/local/bin/to_srt.py", line 165, in xml_to_srt content = re.search(content_re, s).group(1) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: 'NoneType' object ...
Learning X
Fixing my X resources is not easy. Apart from the issues regarding the clipboard interface, still not solved, there's the question of the sheer number of resource definitions needed in .Xdefaults. Adding another 12 names for xterms will blow that list up by over 100 entries. Surely there must be some kind of inheritance. After much searching, found: yes. There are resources and classes, and a whole lot of other things that I didn't know about. Chapter 9 of the X Window System User's Guide for X11 R3 and R4 of the X Window System, one of the books that I recently sent to Cornelius Keck, describes them.
For Additional Confusion
The Wikipedia article on antipopes can be pretty confusing! If you’d like to be even more confused, it can help with that!
More git pain
A few days ago I reinstated my Makefile patch for editors/emacs-devel and ?committed? it (locally). That seemed to have satisfied git to the point that it was happy to update the tree anyway. Until today. Now it complains again. Please push or stash. But pushing didn't work: it returned an error 403. Presumably that's the HTTP ?forbidden? error. Why? OK, stash: === grog@hydra (/dev/pts/6) /usr/ports/editors/emacs-devel 59 -> git stash Saved working directory and index state WIP on main: 844c0c7d6849 x11/xclicker: new port: autoclicker for x11 What does that mean?
The daily X pain
I've been dragging my heels on updating my X configurations. What do I do first? I want minimum down time, particularly on hydra. And it's becoming clear that I have to do something to rearrange my resource names. Currently I give my xterms names that have ?just growed?: xterm-rs, xterm-r, xterm-rx, xterm-rsx, xterm-rsxa, xterm-rsxd, xterm-rsxp, xterm-rsxrad, xterm-rsxrap, xterm-rsxraz, xterm-lx and xterm-lsxraz. What do they mean? The -l and -r refer to left and right sides, the s probably to a root shell. And the rest is a case of UTSL. It seems that some aren't even used! Time for better names that reflect the class of system to which they refer.
Chasing the TV flickering
What I did manage was to compare my flickering video output between teevee and tiwi. Surprise, surprise. It doesn't happen on tiwi. Same video file. Same software (mpv), though a slightly older version. Same TV. About the only thing that comes to mind is the display card. That would also explain why the flickering extends to other windows, including the root windows. The other thing that I noticed was that the flickering went away when I repositioned the video stream and then returned to the same place.
If it ain't broke, don't fix it
It's time to reconfigure my X displays Yet Again, so that the LG 27UP850 monitors are both connected via DisplayPort. But the flickering has stopped, so I moved on to other things, notably trying to update my X configuration. And I didn't even manage much of that.
Reader Q&A: What does it mean to initialize an int?
Acknowledgments: Thanks to Davis Herring, Jens Maurer, Richard Smith, Krystian Stasiowski, and Ville Voutilainen, who are all ISO C++ committee core language experts, for helping make my answer below correct and precise. I recently got this question in email from Sam Johnson. Sam wrote, lightly edited: That’s a great question. Cppreference is correct, and for … Continue reading Reader Q&A: What does it mean to initialize an int? →
teevee progress?
Somehow I'm not making much progress with teevee. Some reasons are clear: messing around with cables, especially HDMI, is stress-inducing, potentially it could take more time than I am willing to expend, and ?if it works, don't fix it?. But it needs to be done, and today I was reminded forcefully that there's still an issue with fvwm 2: Aug 7 18:02:59 teevee kernel: pid 25117 (fvwm), jid 0, uid 1004: exited on signal 6 (core dumped) Aug 7 18:02:59 teevee kernel: pid 25118 (fvwm), jid 0, uid 1004: exited on signal 6 (core dumped) Aug 7 18:02:59 teevee kernel: pid 25119 (fvwm), jid 0, uid 1004: exited on signal 6 (core dumped) For reasons I don't understand, fvwm(2) just didn't want to talk to the X server.
Flickering video display, again
More playing around with the flickering/flashing display that I had investigated yesterday, with little progress. Tried recoding it like this: === grog@teevee (/dev/pts/12) /spool/Series/Freunde-fuers-Leben/01 7 -> ffmpeg -i 01-04-Pillengluck.mp4 -c copy foo.mp4 The result? Still flashes. Even the root screen flashes! An obvious further investigation occurred to me later: try displaying it with tiwi. Both computers are connected to the TV, so it's as simple as changing the input. Mañana.
More flashes on TV
Watching an old programme on ZDF this evening, ?Freunde für's Leben?. And once again I got these strange flashes that appeared to be due to the video source confusing the TV. Even when I pressed ?Pause?, the flashing continued. Even describing it is difficult. Dark sections of the image flash white. How do I track it down? An obvious start would be to recode the file and see if the recoded version also causes the effect. If not, at least we have a workaround.
Understanding cable problems
My new 3 m DisplayPort cable is here already. Installing it isn't as easy as it seems. Yes, physically attaching it is straightforward enough, but using it properly means braving nvidia-settings and of course restarting all my X sessions. Still, I should check that it works, so I connected monitor 3 via the new cable to the output for monitor 2, and connected the identical monitor 2 with the existing HDMI cable to the output for monitor 3. All worked. By rights the flashes and dropouts that I had had before on monitor 3 should now have occurred on monitor 2.
How (and how not) to present related work
A key part in scientific writing is a description of related work. This section establishes what is known in the given area and the new publication?s contribution. It also provides a signal to reviewers and readers regarding the study?s innovativeness, credibility, and thoroughness. A paper with a shallow related work section may well have overlooked important relevant work that would have supported its theory building, methods, or conclusions. A mistake often made in related work sections is to present them as a laundry list ( A did X , B did Y ), often in chronological order.
Apple mail fail?
One of the many things that really get on my nerves is email written in HTML only. Apart from the fact that Mutt handles it badly, it makes it very difficult to reply to without resorting to ad-hoc commercial solutions. It also makes it difficult to follow mail threads. One of the relatively frequent participants in the Hugin mailing list has been sending out such mail, using iPad mail and iPhone mail. I thought I might have some input, so I sent him a message asking him to at least set multipart-alternative. He wasn't upset, but he didn't know how to set it up.
AI?s productivity theater
This week on my podcast, I read a recent post from my Pluralistic.net blog/newsletter: “AI’s productivity theater,” about the severe mismatch between the bosses who buy AI to increase their workers’ efficiency, and the utter bafflement of the workers who are expected to use the AI…somehow. A new research report from the Upwork Research Institute... more
I hate cables!
Much of my computer rearrangement recently has been impeded by cable issues: wrong length, inadequate quality (especially these horrible HDMI cables, which have already cost me a TV, and which seem inadequate for my 4K monitors), hard to trace the other end. Today I had a case: turn on the monitors for eureka. No display. After a while, discovered that the left-hand one had the wrong HDMI cable connected. Connect the right one and it works. But I couldn't get the right-hand one to run. It was connected to one of two DVI cables, but it wouldn't display with either of them.
Mail corruption
Seen this morning on my mail screen (Mutt): 17 + 02-08-2024 To Greg Lehey ( 22) Yvonne Lehey + GateKit 18 + 01-08-2024 To grog@freebsd ( 23) mailman-owner@ + ??>mailing list memberships reminder The second message, typical at the start of the month, looks to be a follow-up to a private message from Yvonne. Huh? Take a look: From MAILER-DAEMON Thu Aug 1 22:02:31 2024 Date: Thu, 01 Aug 2024 05:00:02 -0700 From: mailman-owner@ To: groggyhimself@freebsd.org Subject: mailing list memberships reminder Message-ID: <mailman.630.1722513602.810421.mailman@> In-Reply-To: <20240802002858.GB15490@lagoon.lemis.com> But that's impossible.
Tidying up teevee
Slow day today. Spent some time playing around with teevee, though there wasn't much to be done. I'm still puzzled by the difference in font size that mpv demonstrates between tiwi and teevee. Could it be in the configuration files? They're in ~/.config/mpv/. But no, they were the same. But somewhere I recall reducing the size years ago. How did I do it? It's not in my wrapper script. Somehow I'm missing something here. OK, RTFM. ?No manual entry for "mpv"?. Oh. Only recently did I discover that ports man pages have been moved from /usr/local/man to /usr/local/share/man, and I had had to update my MANPATH accordingly.
Waking distress
Lately I've been having slight trouble accessing distress, one of my Microsoft boxes. I access it by rdesktop, but first it needs to be woken. That's what wake(8) is for: it sends a Wake-on-LAN packet to the machine, which is listed in /etc/ethers. Problem: hydra has two network interfaces. Under those circumstances, wake wants to know which interface to use, so I run wake re1 distress. But lately that hasn't worked: === grog@hydra (/dev/pts/18) ~ 42 -> wake re1 distress wake: write(): Network is down wake: Cannot send Wake on LAN frame over `re1' to `distress': Network is down === grog@hydra (/dev/pts/18) ~ 43 -> wake distress === grog@hydra (/dev/pts/18) ~ 44 -> And yes, the second invocation works, the one that shouldn't work.
Completing teevee
Now that hydra is running relatively smoothly, I can turn my attention to teevee, the replacement for tiwi that I have been working on for nearly two months. It was happily running in my office, but moving it to its correct place was complicated by the cabling and my requirement to keep things running throughout. My concerns with the cabling were justified, but after that things came up smoothly, and it Just Ran. Well, of course there were a few details. I still need to migrate to fvwm3 and modify my .Xdefaults, but that's the same issue as with hydra, and for the time being things work well.
Fixing firefox
One of the problems with a 3840×2160 resolution monitor is that web browsers like firefox display everything too small. Spent some time today changing the default font sizes, with less than stellar results. Surely there's a better solution? A web search pointed me at the configuration parameter layout.css.devPixelsPerPx in about:config. By default it's set to -1.0, but it can be set to between 1.0 and 2.0 to scale all output.
hydra progress
More work on hydra today. What do I need to do with the configuration for server 1? Fired it up, and it came up, just with the monitors in the wrong sequence. And nvidia-settings didn't want to know about the fourth monitor until I insisted. OK, moving the monitors is almost straightforward, except now there are no X display 1 and friends. I have to tell nvidia-settings the (sometimes ambiguous) name of the monitor. Save the config file, and we're done. Start the server with the new config file. No devices detected. Fatal server error: no screens found(EE) Huh?