IANAL nor intelligent, but after skimming the text of the directive I felt like the definition of damage is very limited. In particular, if I understand correctly:

our business to lose this giant contract

would not be covered by this directive, this directive is only about a human being hurt in some way,

thousands of consumers left with bricked devices

would be covered in case of “your game installs a kernel-level anticheat and the anticheat breaks PCs”, but not in the case of “you uploaded an upgrade to a firmware of the washing machine you produced and it bricked the machines”; the directive is not about a product breaking, but about the product breaking your health, other property or data,

my washing machine to eat my dog

is basically the exact case this directive covers.

Same in Python, Rust, Haskell and probably many others.

But apparently JS does work that way, that is its filter always iterates over everything and returns a new array and not some iterator object.

Hasn’t Google already made advances through its Alpha Geometry AI?? Admittedly, that’s a geometry setting which may be easier to code than other parts of Math and there isn’t yet a clear indication AI will ever be able to reach a certain level of creativity that the human mind has, but at the same time it might get there by sheer volume of attempts.

Wanted to focus a bit on this. The thing with AlphaGeometry and AlphaProof is that they really treat doing math as a game, not unlike chess. For example, AlphaGeometry has a basic set of rules, it can apply them and it knows when it is done. And when it is done, you can be 100% sure that the solution is correct, because the rules of the game are known; the 28/42 score reported in the article is really four perfect scores and three zeros. Those systems do use LLMs, but they really are only there to suggest to the system what to try doing next. There is a very enlightening picture in the AlphaGeometry paper here: https://www.nature.com/articles/s41586-023-06747-5#Fig1

You can automatically verify correctness of code the same way. For example Lean, the language AlphaProof uses internally, can be used for general programming. In general, we call similar programming techniques formal methods. But most people don’t do this, since this is more time-consuming than normal programming, and in many cases we don’t even know how to define the goal of our code (how to define correct rendering in a game?). So this is only really done when the correctness of the program is critical, like famously they verified the code of the automatic metro in Paris this way. And so most people don’t try to make programming AI work this way.

I really need to try out Mercury one day. When we did a project in Prolog at uni, it felt cool, but also incredibly dynamic in a bad way. There were a few times when we misspelled some clause, which normally would be an error, but in our case it just meant falsehood. We then spent waaay to much time searching for these. I can’t help but think that Mercury would be as fun as Prolog, but less annoying.

I actually use from time to time the Bower email client, which is written in Mercury.

Interesting. I looked this up and I think that in Poland, the wait time in let’s say Warsaw peaked at like 2 months during pandemic, but is around 2 weeks now.

Many people living in big cities will have their exams in smaller WORDs anyway, as the pass rates tend to be higher there (not a surprise, less traffic means an easier exam). Apparently in some WORDs you can even get a new attempt the same day after failing one.

In Poland:

• driver’s permits are not a thing. In general, it’s illegal to drive without a professional instructor (with parents, for example) before getting a driving license, though a lot of people, especially in the countryside, will still do so,
• you can only drive after turning 18. You can start the course a few months earlier, but you can only take the final exam after you turn 18 (there exists a category that allows you to drive after turning 16, but it’s limited and IME extremely unpopular),
• you need to go to a paid course, which includes theory classes and at least 30 hours of driving with the instructor,
• most people drive in a car owned by the instructor or the driving school, as the car must have another pair of brakes for the instructor,
• you need to pass a theoretical and a practical exam in one of the centers (Wojewódzki ośrodek ruchu drogowego),
• the theoretical exam is just closed questions. You need to get 68 out of 74 points, but (AFAIK, this has changed over time) all the questions are known, so people will just cram them,
• the practical exam is first some maneuvers on the center grounds, and then a ride around the city. The exam is rather objective and is failed if you do any big mistake or fail any exercise twice,
• the exams are not easy. The data I found is for each WORD, but in general I feel like the pass rate is around 50% for the practical exam and 70% for theory. It’s not incommon for somebody to only pass their practical exam on like 5th attempt,
• there were supposed to be some restrictions for new drivers, but they had been discussed for a long time, even back when I passed my license before the pandemic, and I have no idea if they ever actually came into force,
• some people think that the system is super flawed. Here’s some discussion by the Supreme Audit Office in Polish: https://www.nik.gov.pl/aktualnosci/system-szkolenia-kandydatow-na-kierowcow.html,
• costwise, it’s apparently like 4000 zł for the course right now. Exams are paid per attempt, 50 zł for the theory and 200 zł for practice. 1 euro is 4.33 zł as of writing, but you need to take into account the difference in purchasing power and it’s probably not much cheaper than Germany even if you pass both exams the first time.

Phoenotopia: Awakening – an amazing metroidvania-related game. Relatively more popular than the other games I list, but is honestly one of my favorite games of all time.

Vision: Soft Reset – a metroidvania, but you can travel backwards and forwards in time and this really matters for gameplay.

Bombe – Minesweeper, but instead of solving the puzzles manually, you create rules (“if there is a cell with the number N and there are N empty cells around it, mark them all as mines”) which the game applies automatically.

SOLAS 128 – a puzzle game where you redirect signals in a huge machine, just a great experience if you like puzzle games.

Edit: Actually, I thought about it, and I don’t think clang’s behavior is wrong in the examples he cites. Basically, you’re using an uninitialized variable, and choosing to use compiler settings which make that legal, and the compiler is saying “Okay, you didn’t give me a value for this variable, so I’m just going to pick one that’s convenient for me and do my optimizations according to the value I picked.” Is that the best thing for it to do? Maybe not; it certainly violates the principle of least surprise. But, it’s hard for me to say it’s the compiler’s fault that you constructed a program that does something surprising when uninitialized variables you’re using happen to have certain values.

You got it correct in this edit. But the important part is that gcc will also do this, and they both are kinda expected to do so. The article cites some standard committee discussions: somebody suggested ensuring that signed integer overflow in C++20 will not UB, and the committee decided against it. Also, somebody suggested not allowing to optimize out the infinite loops like 13 years ago, and then the committee decided that it should be allowed. Therefore, these optimisations are clearly seen as features.

And these are not theoretical issues by any means, there has been this vulnerability in the kernel for instance: https://lwn.net/Articles/342330/ which happened because the compiler just removed a null pointer check.

Isn’t this the point though? Like, if you spot that (let’s concretize) the trash is starting to overflow, you can either take it out right now which will take you 2 minutes and (hopefully) barely interrupt your day, or you can add it to your list of things to do. And so you get that list of 59 things by ignoring the 2-minute rule, not by applying it.