@lyse@lyse.isobeef.org These tables get shuffled around every time your OS switches to another process. It’s crazy that so much is going on behind the scenes.

@kiwu@twtxt.net Assembly is usually the most low-level programming language that you can get. Typical programming languages like Python or Go are a thick layer of abstraction over what the CPU actually does, but with Assembler you get to see it all and you get full control. (With lots of caveats and footnotes. 😅)

I’m interested in the boot process, i.e. what exactly happens when you turn on your computer. In that area, using Assembler is a must, because you really need that fine-grained control here.

I rewrote all my solutions in Rust (except for day 10 part 2) and these are the runtimes on my i7-3770 from 2013 (this measures CLOCK_PROCESS_CPUTIME_ID, not wallclock):

day01/1 [      00.000501311] Result: 1066
day01/2 [      00.000400298] Result: 6223
day02/1 [      00.000358848] Result: 12586854255
day02/2 [      00.000750711] Result: 17298174201
day03/1 [      00.000106537] Result: 17405
day03/2 [      00.000404632] Result: 171990312704598
day04/1 [      00.000257517] Result: 1626
day04/2 [      00.007495342] Result: 9173
day05/1 [      00.000237212] Result: 505
day05/2 [      00.000142731] Result: 344423158480189
day06/1 [      00.000229629] Result: 4076006202939
day06/2 [      00.000279552] Result: 7903168391557
day07/1 [      00.000204422] Result: 1622
day07/2 [      00.000283816] Result: 10357305916520
day08/1 [      00.029427421] Result: 84968
day08/2 [      00.028089859] Result: 8663467782
day09/1 [      00.000310304] Result: 4764078684
day09/2 [      00.015512554] Result: 1652344888
day10/1 [      00.000796663] Result: 375
day10/2 [      --.---------] Result: 15377 (Z3)
day11/1 [      00.000416804] Result: 753
day11/2 [      00.000660528] Result: 450854305019580
day12/1 [      00.000336081] Result: 577
day12/2 [      00.000000695] Result: no part 2

A little under 90 ms total.

On my Samsung NC10 netbook from 2011 with its Intel Atom N455 at 1.6 GHz:

day01/1 [      00.003771326] Result: 1066
day01/2 [      00.003267317] Result: 6223
day02/1 [      00.003902698] Result: 12586854255
day02/2 [      00.006659479] Result: 17298174201
day03/1 [      00.000747544] Result: 17405
day03/2 [      00.002737587] Result: 171990312704598
day04/1 [      00.001263892] Result: 1626
day04/2 [      00.044985301] Result: 9173
day05/1 [      00.001696761] Result: 505
day05/2 [      00.000978962] Result: 344423158480189
day06/1 [      00.001387660] Result: 4076006202939
day06/2 [      00.001734248] Result: 7903168391557
day07/1 [      00.001295528] Result: 1622
day07/2 [      00.001809659] Result: 10357305916520
day08/1 [      00.277251443] Result: 84968
day08/2 [      00.284359332] Result: 8663467782
day09/1 [      00.003152407] Result: 4764078684
day09/2 [      00.071123459] Result: 1652344888
day10/1 [      00.005279527] Result: 375
day10/2 [      --.---------] Result: 15377 (Z3)
day11/1 [      00.003273342] Result: 753
day11/2 [      00.005139719] Result: 450854305019580
day12/1 [      00.002857552] Result: 577
day12/2 [      00.000004421] Result: no part 2

A little over 700 ms total.

I like this. You get performance that’s more or less in the ballpark of C, but without the footguns.

@prologic@twtxt.net I couldn’t find the exact blog post from before, one that used redirection directives in its nginx config. but I found [this one ](https://melkat.blog/p/unsafe-pricing#:~:text=Something%20else%20I’ve%20been%20doing%20this%20year,%20fine.) mentioning a similar process but done differently.

cat /etc/mokou/yarnd.conf
exec=/usr/pkg/sbin/daemonize -c/var/db/yarnd -u www -p /var/run/yarnd.pid /usr/pkg/sbin/chpst -e /usr/local/etc/yarnd /usr/local/sbin/yarnd -b 127.0.0.1:[classified information]

I know this might seem a bit overengineered, but the previous command until now had the secrets exposed on the process list

@prologic@twtxt.net The periodic blacklists updates will be done automatically in the background, as for the different processing mechanisms (rules, collections of rules, remediation …etc) you just install/add the pre-made ones from the hub and call it a day, they’ll get periodic updates when needed. But you could easily create and add your own in case you want to block or white-list a specific behavior

@prologic@twtxt.net The main thing that I tought of is that whomever is abusing your services must be a well known actor (by range/set of IPs) that got reported by other Crowdsec users. So to my simpleton’s understanding, your reverse-proxy/web server passes the requests by crowdsec for processing, they get banned for $N hours if the source has already been blacklisted by the community or violates any of a set of behavior base rules (and even more hours for repeat offenders); otherwise the requests/responses go as per usual. Not sure if I got things right but this might help paint a better picture of the process.

Be it Java with Swing or PyQt6, it takes ~300 ms until a basic window with a treeview and a listbox appears. That is a very noticeable delay.

Is it unrealistic to expect faster startup times these days? 🤔

Once the program is running, a new second window (in the same process) appears very quickly. So it’s all just the initialization stuff that takes so long. I could, of course, do what “fat” programs have done for ages: Pre-launch the process during boot, windowless. But I was hoping that this wasn’t needed. 😞 (And it’s a bad model anyway. When the main process crashes, all windows crash with it.)

I used to run office hours at Google and the number of people who came into my office absolutely convinced that there was no way to search a dataset without having the entire thing in memory for every process was too damn high.

@bender@twtxt.net The only problem with uploading is the procesing. Do you expect any server-side processing of the WebP or just store and host?

@prologic@twtxt.net Let’s go through it one by one. Here’s a wall of text that took me over 1.5 hours to write.

This section says AI should not be treated as an authority. This is actually just what I said, except the AI phrased/framed it like it was a counter-argument.

The AI also said that users must develop “AI literacy”, again phrasing/framing it like a counter-argument. Well, that is also just what I said. I said you should treat AI output like a random blog and you should verify the sources, yadda yadda. That is “AI literacy”, isn’t it?

My text went one step further, though: I said that when you take this requirement of “AI literacy” into account, you basically end up with a fancy search engine, with extra overhead that costs time. The AI missed/ignored this in its reply.

Okay, so, the AI also said that you should use AI tools just for drafting and brainstorming. Granted, a very rough draft of something will probably be doable. But then you have to diligently verify every little detail of this draft – okay, fine, a draft is a draft, it’s fine if it contains errors. The thing is, though, that you really must do this verification. And I claim that many people will not do it, because AI outputs look sooooo convincing, they don’t feel like a draft that needs editing.

Can you, as an expert, still use an AI draft as a basis/foundation? Yeah, probably. But here’s the kicker: You did not create that draft. You were not involved in the “thought process” behind it. When you, a human being, make a draft, you often think something like: “Okay, I want to draw a picture of a landscape and there’s going to be a little house, but for now, I’ll just put in a rough sketch of the house and add the details later.” You are aware of what you left out. When the AI did the draft, you are not aware of what’s missing – even more so when every AI output already looks like a final product. For me, personally, this makes it much harder and slower to verify such a draft, and I mentioned this in my text.

You, @prologic@twtxt.net, also mentioned this in your car tyre example.

In my text, I gave two analogies: The gym analogy and the Google Translate analogy. Your car tyre example falls in the same category, but Gemini’s calculator example is different (and, again, gaslight-y, see below).

What I meant in my text: A person wants to be a programmer. To me, a programmer is a person who writes code, understands code, maintains code, writes documentation, and so on. In your example, a person who changes a car tyre would be a mechanic. Now, if you use AI to write the code and documentation for you, are you still a programmer? If you have no understanding of said code, are you a programmer? A person who does not know how to change a car tyre, is that still a mechanic?

No, you’re something else. You should not be hired as a programmer or a mechanic.

Yes, that is “skill evolution” – which is pretty much my point! But the AI framed it like a counter-argument. It didn’t understand my text.

(But what if that’s our future? What if all programming will look like that in some years? I claim: It’s not possible. If you don’t know how to program, then you don’t know how to read/understand code written by an AI. You are something else, but you’re not a programmer. It might be valid to be something else – but that wasn’t my point, my point was that you’re not a bloody programmer.)

Gemini’s calculator example is garbage, I think. Crunching numbers and doing mathematics (i.e., “complex problem-solving”) are two different things. Just because you now have a calculator, doesn’t mean it’ll free you up to do mathematical proofs or whatever.

What would have worked is this: Let’s say you’re an accountant and you sum up spendings. Without a calculator, this takes a lot of time and is error prone. But when you have one, you can work faster. But once again, there’s a little gaslight-y detail: A calculator is correct. Yes, it could have “bugs” (hello Intel FDIV), but its design actually properly calculates numbers. AI, on the other hand, does not understand a thing (our current AI, that is), it’s just a statistical model. So, this modified example (“accountant with a calculator”) would actually have to be phrased like this: Suppose there’s an accountant and you give her a magic box that spits out the correct result in, what, I don’t know, 70-90% of the time. The accountant couldn’t rely on this box now, could she? She’d either have to double-check everything or accept possibly wrong results. And that is how I feel like when I work with AI tools.

Gemini has no idea that its calculator example doesn’t make sense. It just spits out some generic “argument” that it picked up on some website.

The AI makes two points here. The first one, I might actually agree with (“bad bot behavior is not the fault of AI itself”).

The second point is, once again, gaslighting, because it is phrased/framed like a counter-argument. It implies that I said something which I didn’t. Like the AI, I said that you would have to adjust the copyright law! At the same time, the AI answer didn’t even question whether it’s okay to break the current law or not. It just said “lol yeah, change the laws”. (I wonder in what way the laws would have to be changed in the AI’s “opinion”, because some of these changes could kill some business opportunities – or the laws would have to have special AI clauses that only benefit the AI techbros. But I digress, that wasn’t part of Gemini’s answer.)

Except for one point, I don’t accept any of Gemini’s “criticism”. It didn’t pick up on lots of details, ignored arguments, and I can just instinctively tell that this thing does not understand anything it wrote (which is correct, it’s just a statistical model).

And it framed everything like a counter-argument, while actually repeating what I said. That’s gaslighting: When Alice says “the sky is blue” and Bob replies with “why do you say the sky is purple?!”

But it sure looks convincing, doesn’t it?

This took so much of my time. I won’t do this again. 😂

@movq@www.uninformativ.de Yeah, give it a shot. At worst you know that you have to continue your quest. :-)

Fun fact, during a semester break I was actually a little bored, so I just started reading the Qt documentation. I didn’t plan on using Qt for anything, though. I only looked at the docs because they were on my bucket list for some reason. Qt was probably recommended to me and coming from KDE myself, that was motivation enough to look at the docs just for fun.

The more I read, the more hooked I got. The documentation was extremely well written, something I’ve never seen before. The structure was very well thought out and I got the impression that I understood what the people thought when they actually designed Qt.

A few days in I decided to actually give it a real try. Having never done anything in C++ before, I quickly realized that this endeavor won’t succeed. I simply couldn’t get it going. But I found the Qt bindings for Python, so that was a new boost. And quickly after, I discovered that there were even KDE bindings for Python in my package manager, so I immediately switched to them as that integrated into my KDE desktop even nicer.

I used the Python KDE bindings for one larger project, a planning software for a summer camp that we used several years. It’s main feature was to see who is available to do an activity. In the past, that was done on a large sheet of paper, but people got assigned two activities at the same time or weren’t assigned at all. So, by showing people in yellow (free), green (one activity assigned) and red (overbooked), this sped up and improved the planning process.

Another core feature was to generate personalized time tables (just like back in school) and a dedicated view for the morning meeting on site.

It was extended over the years with all sorts of stuff. E.g. I then implemented a warning if all the custodians of an activitiy with kids were underage to satisfy new the guidelines that there should be somebody of age.

Just before the pandemic I started to even add support for personalized live views on phones or tablets during the planning process (with web sockets, though). This way, people could see their own schedule or independently check at which day an activity takes place etc. For these side quests, they don’t have to check the large matrix on the projector. But the project died there.

Here’s a screenshot from one of the main views: https://lyse.isobeef.org/tmp/k3man.png

This Python+Qt rewrite replaced and improved the Java+Swing predecessor.

There are only two hard problems in distributed systems: 2. Exactly-once delivery 1. Guaranteed order of processing 2. Exactly-once delivery

@prologic@twtxt.net Ah, I see. Yeah, you might be right. (Still a fragile process due to the general AI wonkiness, but it can help to some degree, yes.)

@movq@www.uninformativ.de I submitted it via the form on their website (https://digital-markets-act.ec.europa.eu/contact-dma-team_en) and got the following response:

Dear citizen,

Thank you for contacting us and sharing your concerns regarding the impact of Google’s plans to introduce a developer verification process on Android. We appreciate that you have chosen to contact us, as we welcome feedback from interested parties.

As you may be aware, the Digital Markets Act (‘DMA’) obliges gatekeepers like Google to effectively allow the distribution of apps on their operating system through third party app stores or the web. At the same time, the DMA also permits Google to introduce strictly necessary and proportionate measures to ensure that third-party software apps or app stores do not endanger the integrity of the hardware or operating system or to enable end users to effectively protect security.

We have taken note of your concerns and, while we cannot comment on ongoing dialogue with gatekeepers, these considerations will form part of our assessment of the justifications for the verification process provided by Google.

Kind regards,
The DMA Team

@lyse@lyse.isobeef.org Cool! 😎 You might be interested in my own learnings and toying around with building my own container engine / tooling (whatever you wanna call it) box. I had to learn a bunch of this stuff too 😅 Control Groups, Namespaces, Process Isolation, etc.

@zvava@twtxt.net oh duh! Sorry, I promised I read, my brain just didn’t process it right. I shall follow your progress, and offer bits and pieces of unrequested trivialities. :-)

I hear you, @movq@www.uninformativ.de! :‘-(

At work, too. For a few weeks now when I try to log into this horrible Outlook web intershit (Because why would they fix the Evolution integration?! It’s cactus for well over a year now. Probably more like two.), it forwards me to the corporate weblogin, I enter my credentials, even do the bloody MFA crap and get redirected back to Outlook. “Loading mailbox…” “Please wait for us to log you out, do not close this window while this process is underway.” Fuck you! I have to delete the cookies for this damn domain each and every fucking time. Otherwise, this goes in circles forever. I tried the game for 15 minutes, no joke.

But wait, there’s more! Why just fuck it up only a little bit? This week I get logged out at the middle of the day. Every. Single. Day. Not even close to eight hours since I started, no. What the hell!? I reckon I just don’t even bother reauthenticating anymore in the arvo. No more e-mails for Lyse after lunch. Fuck it. It’s just distraction, anyway, right?!

@movq@www.uninformativ.de I had to look it up! “Is decaf coffee real coffee?”

“Yes, decaf coffee is real coffee. It’s made from the same coffee beans as regular coffee, but the caffeine content is significantly reduced through a decaffeination process. This process involves removing 97% or more of the caffeine, leaving behind the coffee’s flavors and aromas.”

OK then! 😅

Fuck 🤣 Building and learning about machine learning and evolutionary processes is hard™ 🤣

One of the nicest things about Go is the language itself, comparing Go to other popular languages in terms of the complexity to learn to be proficient in:

• Go: 25 keywords (Stack Overflow); CSP-style concurrency (goroutines & channels)
  
• Python 2: 30 keywords (TutorialsPoint); GIL-bound threads & multiprocessing (Wikipedia)
  
• Python 3: 35 keywords (Initial Commit); GIL-bound threads, asyncio & multiprocessing (Wikipedia, DEV Community)
  
• Java: 50 keywords (Stack Overflow); threads + java.util.concurrent (Wikipedia)
  
• C++: 82 keywords (Stack Overflow); std::thread, atomics & futures (en.cppreference.com)
  
• JavaScript: 38 keywords (Stack Overflow); single-threaded event loop & async/await, Web Workers (Wikipedia)
  
• Ruby: 42 keywords (Stack Overflow); GIL-bound threads (MRI), fibers & processes (Wikipedia)
  

@bender@twtxt.net @prologic@twtxt.net Jokes aside, I don’t think that’s the right approach either. We had spell checkers, since I can remember, as well as other tools, like the smart image select, used mostly to remove backgrounds. These are tools, that just simplify the process of either opening up a dictionary and looking up a word, you can’t remember the spelling of, or the process of placing a billion little dots around the part of an image you want to select - none of these are creative or enjoyable tasks, we already had tools for them, decades before AI. I don’t think we need to go back to cave paintings, to be free of AIs influence on our creative work.

@prologic@twtxt.net @movq@www.uninformativ.de I don’t even think the premise of this makes much sense. If an artist is convinced they cannot compete, with the “AI” learning models, we already have today, they must have some self esteem issues, strange opinion on what the purpose of art is, or just be someone mindlessly redrawing already established things and not be all that good at it.

It might be connected to some typically non-artists assumption, that the more time and effort the artwork took to accomplish, the more artistic it is - this can be further twisted in these peoples minds, into the “more pointless detail = more artistic art” meme. AI often ads pointless and illogical details everywhere, “so it’s obviously better, than the human artist, who drew the original”.

Some people just enjoy having the picture they wanted or having the status of an artist to brag about and don’t actually enjoy the artistic process of discovery and small decisions, made while drawing, that shape the outcome into something, only you could have created.

@prologic@twtxt.net That’s an interesting premise in that article:

The fun has been sucked out of the process of creation because nothing I make organically can compete with what AI already produces—or soon will.

This is like saying it’s pointless to make music yourself because some professional player/audio engineer does a better job. Really, there’s always someone or something that’s better than you at a particular job.

If we focus too much on “competition”, then yes, you can just stop doing anything. I don’t know how common this mindset is, especially among artists or creative people. 🤔 I would have assumed that many writers, for example, simply enjoy the process of writing. Am I being too naive once more? 🤣

@kat@yarn.girlonthemoon.xyz I only listened to you while going through my photos, so I did not pay very close attention. :-)

Since you have a proper server – haha, not just one – and hence are not limited, I suggest you learn a real programming language and don’t waste your time with this PHP mess. It might have improved a wee bit since I was a kid, but it felt like some hacked together shit. The defaults also were questionable at best, it was easier to hold it wrong than right. This stands testament to bad design and is especially terrible from a security point of view.

You’re right, programming is like any other craft. You only truly learn by actually doing it. And this just takes time. Very long time to master it. Or as close to as it gets. The more you know, the more you realize what else you don’t know (yet). It’s a never ending process. So, take it easy, don’t get discouraged, happy hacking and enjoy the endeavor! :-)

@lyse@lyse.isobeef.org oooh that’s a good point! woodworking is scary and i don’t have much room for it but i do have SOME room in mind that could work for it… i feel like i’d just hurt myself in the process though LOL

@@twtxt.net The fact that it has an SDK and process management is quite amazing g! 🤯

ArkFlow: High-performance Rust stream processing engine
Comments ⌘ Read more

Extend (YC W23) is hiring engineers to build LLM document processing
Comments ⌘ Read more

Today I added support for Let’s Encrypt to eris via DNS-01 challenge. Updated the gcore libdns package I wrote for Caddy, Maddy and now Eris. Add support for yarn’s cache to support # type = bot and optionally # retention = N so that feeds like @tiktok@feeds.twtxt.net work like they did before, and… Updated some internal metrics in yarnd to be IMO “better”, with queue depth, queue time and last processing time for feeds.

Synology confirms that higher-end NAS products will require its branded drives
“Synology-branded drives will be needed for use in the newly announced Plus series, with plans to update the Product Compatibility List as additional drives can be thoroughly vetted in Synology systems,” a Synology representative told Ars by email. “Extensive internal testing has shown that drives that follow a rigorous validation process when paired with Synology systems are … ⌘ Read more

The White House has begun process of looking for new secretary of defense
Comments ⌘ Read more

@kat@yarn.girlonthemoon.xyz pandoc is a joy! I haven’t used any Microsoft word processing tools since forever. They want a Word document? Pandoc to the rescue!

That’s an interesting research article about Wallbleed, a memory disclosure vulnerability in the Great Firewall of China. They reverse-engineered the buggy DNS query processing code that injects a response if the hostname should be censored: https://gfw.report/publications/ndss25/data/paper/wallbleed.pdf

Just keeping this shell alive, and sending an useful twtxt in the process. :-P

@javivf@adn.org.es Generally speaking if it has been reviewed, discussed and merged, then we accept it as a standard to the set of specs we support. However we might want to document this process and set some guidelines about this to be clear 🤣 We’ve been fairly lax/lose here and I think that’s okay given teh size of our community 👌

AI problems, top to bottom:

1: Open AI nerds, believe fine tuning a language model algorithm, will eventually produce an AGI god.

2: Subpar artists and techbros who can’t code, convinced AI image bashing and vibe coding, will help convince the dumber parts of Internet, they are a real deal.

3: Parasites, using AI to scam people, because they just want passive income, selling crap, made by an automated process.

Side: Adobe&co, killing Flash/old web, pricing new artists and developers out, to face learning curves of free tools, or use AI, peddled as solution.

Erlang’s not about lightweight processes and message passing
Comments ⌘ Read more

This month in Redox, March 2025
Another month, another month of Redox improvements and bug fixes. This month saw a ton of work on process management as part of the NLnet grant, massive improvements to the USB stack, including a USB hub driver, as well as the usual kernel and driver improvements. On top of all this work, there’s the usual long list of bugfixes and smaller improvements. ⌘ Read more

@aelaraji@aelaraji.com Yeah I’m in the process of rewriting (incrementally) the cache storage backend. It’s now been live for at least a week now and pagination and peering are the last things left to do 🤞

Windows Hotpatch comes to client version of Windows
Good news for Windows users, and for once there’s not a hint of sarcasm here: Microsoft has started rolling out Windows Hotpatch to the client versions of Windows. This feature, which comes from the server versions of Windows, allows the operating system to install patches to in-memory processes, removing the need for a number of restarts. Obviously, this is hugely beneficial for users, as they won’t have to deal with constant r … ⌘ Read more

Show HN: Zig Topological Sort Library for Parallel Processing
Comments ⌘ Read more

Extend (YC W23) is hiring engineers to build LLM document processing
Comments ⌘ Read more

Microsoft releases Windows 11 roadmap tool to help make sense of Windows 11’s development
I’ve complained about the utter inscrutability of the Windows release process for a long time, with Microsoft seemingly using channels, build numbers, code names, date-based version numbers, and so on interchangeably, making it incredibly hard to keep track of what is being released when. It turns out even Microsoft itself started losing track, because it … ⌘ Read more

California bill aims to phase out harmful ultra-processed foods in schools
Comments ⌘ Read more

The Mastodon admins say that it’s probably because of the size of my account (~600 MB), so the export process times out. And I understand that. Here on twtxt, I always use auto-expiring links when I post images or videos. It just gets too much data otherwise. I think I’ll just set my Mastodon account to auto-delete posts after ~180 days or something like that. Nobody cares about old posts anyway.

Memory safety for web fonts in Chrome: Google replaces FreeType with Rust-based alternative
There’s no escaping Rust, and the language is leaving its mark everywhere. This time around, Chrome has replaced its use of FreeType with Skrifa, a Rust-based replacement. Skrifa is written in Rust, and created as a replacement for FreeType to make font processing in Chrome secure for all our users. Skifra takes advantage of Rust’s memory safety, and … ⌘ Read more

I saw 100% I/O wait in htop today but couldn’t find a process which actually does I/O. Turns out, I/O wait isn’t what it used to be anymore:

https://lwn.net/Articles/989272/

In my case, it was mpd which triggered this:

https://github.com/MusicPlayerDaemon/MPD/issues/2241

mpd doesn’t actually do anything, it just sits there and waits for events. To my understanding, this is similar to something blocking on read(). I’m not quite sure yet if displaying this as I/O wait (or “PSI some io”) is intentional or not – but it sure is confusing.

@andros@twtxt.andros.dev If something fits in a CSV file, it typically doesn’t require a database. I agree with that. Depending on the application, more complicated queries might benefit from a database, though. I don’t know awk very well, but I could imagine that grep, sed and cut reach their CSV processing limits rather quickly when you have to deal with escaped (multiline) fields.

I only very rarely have to deal with CSV files or databases in my day to day life. Maybe, these classic Unix tools offer some tricks I’m not aware of. When I have some more complicated CSV input, I generally reach for Python.

Porting the curl command-line tool and library with Goa
For more than a decade, we have a port of the curl library for Genode available. With the use of Sculpt OS as a daily driver as well as the plan to run Goa natively on Sculpt OS by the end of the year, the itch to also port the curl command-line tool became irresistible. Of course this is a perfect territory for using Goa. In this article, I will share the process of porting the curl command-line tool and shared library … ⌘ Read more