ninjin

> Of course I get downvoted for saying this. HN isn't interested in reality any more.

I suspect that rather many of us are simply just tired of Claude and friends getting shoehorned into any conversation about programming at this point. It is about as fun as the Rust Brigade entering any discussion about C. It adds nothing new to the discussion and it is frankly tiring since we pretty much at any time have a handful of conversations on the front page already covering "AI" topics anyway (counting four at the time of writing this).

Hmm, probably beats my default: abuse@$COMPANY.com

Yes, that is common though. You have a variety of measurements, from a variety of technologies, from a variety of distances, and from a variety of sources. Given the damage a quake and tsunami can cause, especially the early measurements are estimates that later get corrected in light of new information. In Japan for example, it is very much not uncommon that early tsunami warnings are later cancelled. Yes, false alarms are bad, but the example I usually use in terms of how much time can matter is the 1983 Sea of Japan quake [1], where the tsunami hit in 12 minutes after the quake.

[1]: https://en.wikipedia.org/wiki/1983_Sea_of_Japan_earthquake

As a type-1 diabetic, in my experience it is perfectly possible to end up in a situation like this even if you are responsible. Remember, type-1 diabetes is a condition that requires your attention tens if not a hundred times per day. This is not just on good days, but also on your very worst days when you are juggling everything else that life throws at you and you may not have slept properly for days. So, maybe you had noticed that your emergency glucose was low, but in this instance out of a hundred you forgot to replace it when you came home because you were exhausted and then overslept and had to rush out in the morning. What makes type-1 diabetes challenging is not that you can not technically manage every single one of the decisions and actions that it demands out of you, it is that you must carry them out under all conditions and largely without fail for every day in your life or you will suffer both short-term and long-term consequences. Everyone I know has had a nightmare episode.

Just a final recent anecdote to illustrate some of the complexity. It may look like maths on the surface, but even something like "carbohydrate ratio" is a fluid concept. Yesterday for example I was out and about in the sun and had the very same bottle of sweet tea that I have had for months. Given that I knew I was physically active and it was hot, I reduced the usual dosage for this specific drink by 80% to be on the safe side. To my surprise, on this particular day for whatever reason this was still far too much and I pushed myself into hypoglycemia in little over an hour. We humans are highly dynamic and complex systems and this is me failing with the very best technology we have available: Insulin pump and continuous glucose monitor.

I have run NixOS for about eight years on server and desktop and been a nixpkgs maintainer. Yes, most of the time I would agree with you. The fact that you get warnings in the terminal for a lot of incompatibilities and changes when upgrading is a really nice touch and upgrades tend to be smooth. I do not use rollbacks much, but when you do need them they are really handy. Having every configuration in a single file makes you more bold to play around with configurations, which felt really empowering when I first got into NixOS, as I knew it could roll things back and I no longer had to keep notes on how each box was set up to refer to in the case of a reinstall or migration.

However, I have had one machine become unbootable as it could no longer mount its encrypted disks after an upgrade, forcing me to mount a rescue image remotely, mount the disks manually, lift the data out, and do a complete reinstall (migrated the box to OpenBSD at that time). Similarly, NixOS once messed up systemd (or vice versa) so badly that I could not even reboot without forcing a power cycle. Lastly, I have had a package break for my use cases by maintainers enabling so many custom flags by default for a package that they enabled one I have never seen enabled by any other packaging team and that then broke RTSP in "funny" ways. Ubuntu did tend to break things like graphics between releases at times back when I used it, but I have never had any other distribution or operating system throw curve balls like the three things I mentioned here.

My general impression of NixOS is that the core is solid, but that nixpkgs just has such a large number of things that it supports that the maintainers struggle to test them all and can not anticipate the interactions between all the packages and options. The default Julia package being so broken that it produced incorrect mathematics due to nixpkgs' insistence on allowing you to swap out the Blas library and also having turned off the unit tests for example springs to mind. This was shipped to end users for a long time before I noticed it by accident by enabling the unit tests and stepped in to clean it up. It all feels very "Gentoo", which was indeed an inspiration for NixOS by the way.

Now, return to that last sentence in the first paragraph that I wrote about feeling empowered to tinker, ultimately, I feel like you should try to resist that urge as it is what pushes you into the untested fractal of possible configurations that NixOS allows you to explore. My other main operating system is OpenBSD, where the mentality is "Stick to the defaults or suffer the consequences"; with NixOS, I feel like everyone's box is more or less a tailored suit, which comes with both its ups and downs.

Modern operating system booting on hardware that is closing in on 40 years old in just over three minutes, this is wild to see:

https://www.youtube.com/watch?v=btwiiZw3B2s

Kenji Aoyama truly is aligned with the best of the hacker spirit. As for getting your hands on a luna88k, I have no clue. The only thing I managed to find was a broken one that sold for ~USD 750 at an online auction.

In memoriam: https://www.tuhs.org/pipermail/tuhs/2026-May/033750.html

Well, it surely did not help that the government has been drip-feeding us computational resources. First we had about 16 GPU nodes to share between the whole country for over a decade. Then just before Isambard-AI came online they made one open call for about a week where you could get nearly enough to train a sizeable model, but the call was poorly advertised and in the middle of high vacation season. After this, the only big call explicitly cut out training large language models by its scope and the general calls have been peanuts and less than me and the UK-LLM team had access to during the Isambard-AI beta phase! When I gave a talk at White Hall recently my message was clear: We have the team, the knowledge, the data, etc. We just need an open call for enough compute to train the darn thing! Here is to hoping that they listened.

If I had the time and energy, I would make some sort of simple code language model and generate infinite junk and feed that to them in the hope that it ruins their future training runs. But, I lack the former and some of the latter. Alternatively, maybe I would actually read one of those "backdoor papers" and try to inject something like that.

I can report that Facebook does not respect robots.txt. Heck, I even mailed domain@fb.com with the specific IP ranges and log samples three times over a month and they of did not even respond. Keeps on wasting my CPU cycles to this day by crawling massive development forks (I hope they choke on the data...):

    $ (cat /var/www/logs/access.log; zcat /var/www/logs/access.log*.gz) | grep 2a03:2880: | wc -l
    626396

About three hits per second for months now.

Use whatever gets the job done. As for scaling and cropping, to the best of my knowledge, ffplay(1) supports all the options that ffmpeg(1) does. Here is what I for example use when capturing my Famicom: ffplay -sws_flags fast_bilinear -framerate 60 -video_size 1920x1080 -vf crop=iw-360:ih,scale=-1:1080 $DEV.

My preferred way is ffplay(1). Last time I checked I get lower latencies than OBS at that, at least when I use `-sws_flags fast_bilinear`, which is the same scaling OBS uses by default.

> Nvi 2 (nvi under openbsd) has Unicode support. It's like BSD Vi but with Iconv. Perfect for my needs.

My primary editor for text (not code) editing and e-mail. I just wish Unicode support made it into base vi(1) on OpenBSD, but for now I can at least use it for all other purposes other than the ones in the previous sentence.

Fair I suppose. I guess one can treat this either as a personal story (although frustratingly scattered across multiple places and incomplete) or as a description of a single instance of an arrest in Japan.

> It's frustrating to try to interpret these stories with a lot of writing and video describing everything except the crucial detail about what the charges were for.

Is it really a crucial detail though? As someone having lived in Japan for a long time, I see no reason why we can not discuss the fact that civil rights and detention treatment in Japan are lacking without resorting to "Do they deserve it in light of what they were suspected for?". I personally see no reason why suspects can not deserve decent sleep, meal, bedding, etc. even if they may be Shoko Asahara himself.

For the record, I have not watched any video or read anything else about this individual. Nor do I intend to.

Likely some sort of stimulant as you point out. It is hardly the first time either as there have been public cases like this numerous times over the last two decades. Some cases even ending with deportation. The one I remember most vividly was someone carrying an unlabeled bottle of ADHD medication that had been sent to them while they were in South Korea by their pharmacist mum in the US; that they then ran afoul of when entering Japan. Similarly, there was a case at the University of Tokyo in the 00s, where an overseas student got sent an (allegedly) unprompted package with cannabis (not a stimulant though) from friends abroad. Allegedly, they were expelled and we got university-wide, anti-drug campaigns with memorable slogans like: "Illicit drugs are illegal".

Due to their history, laws regarding stimulants are harsher in Japan than in many other places in the world [1] and this frequently takes people by surprise. Not that Japanese laws related to illegal drugs are lenient to begin with.

[1]: https://en.wikipedia.org/wiki/Illegal_drug_trade_in_Japan

The Japanese have some wonderful programming along these lines. For preschoolers there is PythagoraSwitch (ピタゴラスイッチ) [1] which features amazing Rube Goldberg machines, geometric reasoning, algorithmic thinking, etc. Sadly, NHK loves to keep their programmes under lock and key, so I could not find anything to share other than the name.

https://en.wikipedia.org/wiki/PythagoraSwitch

For those considering the US model which is listed as USD 549 [1], the Japanese one currently sells for ~USD 260 in Japan [2].

[1]: https://pomera.us/products/pomera-full-suite-typewriter-for-...

[2]: https://kakaku.com/item/K0001457601

The list price appears to be ~USD 380, which is still a hefty saving compared to what they charge for the US model. There are of course software and hardware differences [3], but if it is primarily the hardware you are interested in, I see no reason not to at least consider importing one.

[3]: https://pomera.us/pages/support-help-support-pomera-dm250us-...

I lack an answer, but for other interested, here is a link with more details regarding the event:

https://en.wikipedia.org/wiki/Tony_Shafrazi#Shafrazi_and_Gue...

> I think the record was 5 million tons of bombs over Vietnam (170 kg of bombs per capita), still the collapse didn't happen.

Unsure about the tonnage, but the parallels to current events [1] and the illusion that a bombing campaign will suffice to end a war in a matter of weeks is rather eerie to me.

[1]: https://en.wikipedia.org/wiki/2026_Iran_war

Always fun to read jcs@'s write ups. I am seriously tempted to look into getting one of these as light, fanless alternatives to run OpenBSB on the move are few and far between. My current main laptop is not big at 11", but it clocks in at ~1.7kg and that makes you reluctant to carry it around everywhere.

Main downsides with the DM250 seems to be that you need a USB-C dongle for audio (they are tiny though) and that you can not hook it up to an external screen for better ergonomics and presentations. Still, small, means compromises are necessary, and I do not think the external screen is a deal break if one would consider the DM250 as a daily commute driver.

It is an utterly amazing collection, although not the era I am the most familiar with. However, let me add a few as I love your initiative and went through them all and skimmed the ones below to get a sense of the quality.

Billy Bragg:

* https://archive.org/details/ajc02362_bbragg1986-12-04.ajcpro...

* https://archive.org/details/ajc02359_bbragg1988-05-06.ajcpro...

Not the cleanest vocals on either of the recordings, but the former is overall higher quality from what I could tell.

Boogie Down Productions:

* https://archive.org/details/ajc02338_bdp1988-05-18

Excellent recording and great live act, just a pity that it is rather short.

Mojo Nixon:

* https://archive.org/details/ajc02226_mojonixon-skidroper1987...

Another great live act and recording. I am sure Mojo would very much approve of this being shared as well.

Many thanks for all the code! The second hand market is currently flooded with Wii U hardware that is cheap enough to buy enough stock to last a life time for peanuts. Would be amazing fun for PowerPC development and if I had an alternative timeline where I went into low-level programming, I would love to push for OpenBSD support inspired by your work.

In addition, here is the Bright White Lightning discography:

https://dataairlines.bandcamp.com/album/bad-teeth-data024

https://dataairlines.bandcamp.com/album/dirty-nails-data038

Having not seen anything from them since 2014, so I am very happy for another track.

> I'm reminded of that scene in "Ghost in the Shell" where some guy ask the Major why he is on the team (full of cyborgs) and she responds something along the line of "Because you are basically un-enhanced (maybe without a ghost?) and are likely to respond differently then the rest of us; Overspecialization is death."

The scene you mentioned (amazing movie and holds up to this day) with the Major and Togusa:

https://youtube.com/watch?v=VQUBYaAgyKI

While I frequently use a similar argument, "We need someone 'untainted' to provide a different point of view", my honest opinion is somewhat more nuanced. These models tend to gravitate towards some sort of level of writing competence based on how good we are at filtering pre-training data and creating supervised data for fine-tuning. However, that level is still far below where my current professional writing is and I find it dreadful to read compared to good writing. Plenty of my students can not "see" this, as they are still below the level of current LLMs and I caution them to overly rely on LLMs for writing as they can then never learn good writing and "reach above" LLM-level writing. Instead, they must read widely, reflect, and also I always provide written feedback on their writing (rather than making edits myself) so that they must incorporate it manually into their own and when doing so they consider why I disagree with the current writing and hopefully learn to become better writers.

Indeed, Heathrow security is the rudest I have experienced. They get aggressive if you so much as ask a question. Furthermore, I have on numerous occasions had them argue with me to go against the medical advice from both doctors and medical advice manufacturers. Last time going as far as claiming that a scanner does not emit electromagnetic radiation.

Would have loved to see how it holds up with some load via FastCGI and CGI (via slowcgi(8)), since httpd(8) can be used with both of them.

> Although I guess the argument will be that email clients should use AI to summarise the HTML into a plain text summary.

Or you could pass it through ~5,000 lines of C [1] and you will have it done in milliseconds even on hardware that would be old enough to drink.

[1]: https://codemadness.org/webdump.html

I think our contexts are all different. But, to share a different experience, as an academic (with plenty of conversations involving people in industry as well each year) I have used interleaved and bottom-posting for decades and it causes confusion maybe once a year at most and mostly because Microsoft's online client is broken and at times does not even render anything below "Dear Foo," in the HTML view (got to give this small start up in Redmond some more time though, we can not expect them to implement standards that have only been around for over 40 years).

Exactly. The problem is that by their very nature some content has to be dynamically generated.

Just to add further emphasis as to how absurd the current situation is. I host my own repositories with gotd(8) and gotwebd(8) to share within a small circle of people. There is no link on the Internet to the HTTP site served by gotwebd(8), so they fished the subdomain out of the main TLS certificate. I am getting hit once every few seconds for the last six or so months by crawlers ignoring the robots.txt (of course) and wandering aimlessly around "high-value" pages like my OpenBSD repository forks calling blame, diff, etc.

Still managing just fine to serve things to real people, despite me at times having two to three cores running at full load to serve pointless requests. Maybe I will bother to address this at some point as this is melting the ice caps and wearing my disks out, but for now I hope they will choke on the data at some point and that it will make their models worse.