So you want an LLM that says they don’t train on “copyleft” code, even if they do, so you can virtue signal to yourself without facing any kind of real inconvenience?

How about you start by clarifying what you mean by “copyleft” as that’s not a standard term, and it means different things to different people. I’ve seen it used to refer to MIT style licenses, GPL style licenses, and public domain stuff… All of which are very different things.

Conan and vcpkg are the two big ones I’m familiar with, but some projects use CMake FetchContent (with github acting as the CDN/registry), and depending on your requirements you can also just use system package managers on Linux, maybe Homebrew or Nix/Guix, or more niche ones like Spack. I’ve seen projects packaged using self hosted Pacman registries, and that works surprisingly well.

So you’re saying PhotoCraft is okay because it counts as a clean room reimplementation of Photoshop, yet you’re not commenting on whether LLM-generated code is tainted by the training data, and thus can never be considered “clean” for the purposes of clean rooming?

I’m not a lawyer, but my understanding of clean rooming has always been that the engineer can’t have specific knowledge of the thing he’s reimplementing. For example, those lawyers in that room can’t describe a Microsoft product to an ex-Microsoft engineer, because that engineer has seen Microsoft’s code base and may use that knowledge to inform their design. Likewise, you can’t give the engineer a copy of, say, Microsoft Office’s source code to use as a reference.

Yet an LLM trained on Microsoft Office’s code base being used to reimplement Microsoft Office counts as a clean room reimplementation? It seems to me like the benefit of the doubt in that situation should go to Microsoft, and the engineer is the one responsible for proving they didn’t use the part of the LLM trained on Microsoft code…which is impossible.

…although tbf Photoshop isn’t open source, so its not obvious how much of its code exists in Claude or ChatGPT or whatever. Maybe the code leaked in the past, or maybe Adobe engineers who use Claude inadvertently shared company code, and that ended up in the training data.

(Also, for the record, fuck Microsoft and Adobe)

AI clean rooming

The only way “AI clean rooming” is possible is if the model was trained with verifiably clean room code. Claude, ChatGPT, Gemini, etc only exist because they took our work without paying or asking for permission. Anthropic has a profitable business because they’re reselling the thing they built with our work. They built X with Y, but they didn’t pay for Y. It’s as clear an example of copyright infringement as possible.

So is your article about “clean rooming” specifically, or is that just a term you’re misuing here informally to refer to LLM-powered reverse engineering in general?

I was never convinced by Meson’s value proposition. From what I remember, it was always just meant to reinvent CMake with a different DSL. Like, literally just CMake but slightly different.

I understand CMake’s DSL isn’t the prettiest, but as far as improving ergonomics of a build system, minimizing the amount of build scripts you write is far more effective than changing the syntax. If you find yourself writing too much CMake in a project, then switching to a build system with a different design (eg Bazel, Scons, Waf, …) is probably the better choice.

And of course, since Meson is just a CMake clone, it’s always going to be compared to it, and its always going to have to compete on features. On that front, CMake has always been ahead, and I don’t see that changing. So in the end, Meson is just another example of that one XKCD meme about standards.

It’s not so much an LLM thing as it is a data/curation thing. Idk the specifics, but I do remember back when I used it, chatgpt occasionally presented me with multiple responses, and asked me to pick the one that I liked best. It’s like how Facebook/tiktok/whatever curate your feed by examining how much time you spend eg looking at one post vs another. There’s nothing inherent about LLMs that cause them to prioritize emdashes, thats all the result of extra massaging the operator has applied to the model.

This is also why I feel it’s important to not focus on perceived “LLMisms” as a way to detect slop. It’s extremely trivial to hide that, and the ones that don’t are low hanging fruit.

There are already plenty of package manager options in the C++ world. Having an “official” one arguably only helps improve the onboarding experience, but overall it’s a chokepoint for the entire ecosystem. There are rarely “one size fits all” solutions to a problem, and there certainly isn’t one for a problem as complex as software builds and packaging.

Yes, CMake is crazy. And ugly. And inconvenient

How many years has it been since you last used CMake?

CMake suffers from some of the same perception issues as C++. A lot of the criticism usually boils down to the designers prioritizing backwards compatibility, a thing even critics would agree is a good thing. As an example, the OP shows a CMake and Meson snippet. The CMake one is 3 lines, the Meson one is 2. The extra line in the CMake sample is a version declaration for backwards compatibility, but the rest are identical.

If you look past your preconceived notions, you’ll find that CMake is a good build system and very easy to use. It’s also one of the few that fully supports C++20 modules today.

It’s by no means a silver bullet, and the generator-based design isn’t always the best choice, but its undoubtedly the best choice for the vast majority of C or C++ projects.

100% this. Why’s he bragging about “having permission” to post those screenshots, as if it’s immoral to be a whistleblower? The only people who think like that are the ones on the wrong side of the scandal.

you are only hurting the workers that you claim to want to protect

Also, dawg, It’s my personal data and emails that I want to protect from an irresponsible CEO with AI psychosis. I’ll pretend to care about these random people I’ll never meet when my privacy is secure.

I’m already planning to switch to either Tuta or Fastmail sometime next week.

So basically, this guy is wrong.

I’ve had this (controversial) theory for a while that I’ve yet to see proven wrong. Before AI, it’s basically been known that most programmers “age out” of the craft, because it’s really difficult to keep up with it forever, and most people in middle of the bell curve simply can’t do it. Big tech used to get sued for ageism all the time, because they would do shit like only promote or hire young people (which is illegal).

The aging millennials of today were the young rockstar ninja programmers of the past 2~3 decades, but now most of them are getting too old to keep up. That’s why you see AI psychosis hitting the older coders harder than anyone else, since it provides a convenient way to not face the realities of aging. It’s easier to think you can still be a rockstar by “pivoting” to AI, than it is to admit that you don’t have the energy or motivation to keep your skills sharp. Tim Sweeney (and to a certain extent John Carmack) come to mind as high profile examples of this.

A pre-AI version of this is pivoting to management. That’s what pretty much everyone does, and it’s common to find people like that who haven’t written a line of code themselves in years. But I guess becoming an AI cheerleader is easier to do, and the number of open management roles certainly isn’t big enough for everyone.

Don’t use AI. Don’t do it. People telling you “it’s great for learning” are full of shit and going off of gut feeling, not experience. These LLMs have only been around for a few years, and have only gotten semi competent at generating code for even less time. There is no data or evidence that it helps people learn programming, and plenty of evidence that they do the opposite.

So avoid it. Pick a simple project, and see how far you can get. You will probably fail, but that’s the point! You’ll most likely find something interesting or tough to focus on in the middle of it, and you’ll learn a whole lot about whatever that is, and it’ll stick with you long after you’ve abandoned that project.

My advice is to start very small, and avoid anything GUI related. Write like a bare bones terminal only blackjack game, then add a high score table, persist those high scores to disk, etc. If you fail even at step one, that’s fine! Just try something else, like tictactoe, Simon says, or a non-game like an app for converting between Fahrenheit and Celsius.

Ask for help on a programming forum somewhere with real people.

Of course I sound silly. You’re putting your words in my mouth.

Honestly fuck capitalism for taking us down this road.

Capitalism has nothing to do with this. It’s corruption, incompetent politicians, and a lazy populace who would rather bitch online than go out and vote (something I’m definitely guilty of, but am actively working to change)

The direction things are going are to concentrate wealth and power into the hands of a very small group of assholes. There is no economic or political system that can fix the problems those assholes have been exploiting to pull this off. The only thing that can prevent that is an informed and energized population of voters.

Like, we all want to see Trump, Musk, Zuckerberg, etc in prison. It feels like they’re untouchable, but literally all it would take to do that is to spend a few minutes filling out a ballot to get rid of the career politicians who have been in office too long (and thus are more likely to be corrupt). Even local elections matter.

I see you’re deep into AI psychosis, so you’re brain isn’t working at its full capacity.

Slop means “high volume, low effort garbage,”

and

Slop is a measure of quantity (and to a certain extent, effort), not of quality

Are saying the same thing. Your last paragraph is also incoherent, so I won’t even go into that.

Turn off the AI waifu. It isn’t real.

This is tiring. Any AI generated thing is slop. Slop is a measure of quantity (and to a certain extent, effort), not of quality. Why? Because quality is subjective, and thus a useless point of comparison. You might think those AI generated studio ghibli rip offs are ugly/disgusting, but my mom thinks they’re “cute”.

So stop with this tiring “NoT eVeYtHiNg Ai Is SlOp” cope. The question people like this really want to ask is “will you still think I’m cool if I use AI???”, and that’s also subjective, and thus pointless. I personally will never think anyone who uses generative AI is cool, especially when they reveal how pathetically desperate for approval they are by making posts like this.

That never seems to work with wine

Wine has nothing to do with double clicking. You can set up your desktop environment to do this if you want. Wine requires environments to run apps (aka wine prefixes), and the way I do it is I reuse one environment as much as I can.

Nowadays, I just reuse the wine prefixes Heroic creates, since it usually installs a good chunk of dependencies. Lazy, but it works and saves disk space.