The explicit configure + generate step, touched on in the article, precludes most forms of dynamic dependency generation. Making this work with C++ modules has been very painful. ↑

Safe C++ is an extension of C++. It adds a robust safety model, but it’s not a new language.

Memory leaks - pretty much never. RAII takes care of those way too well for them to happen. The only time I may make them is if I have a resource that I simply don’t know I have to cleanup. Segfaults - I do sometimes make them, particularly during development. They are much more likely when I interface with very low level code (e.g. OS APIs or other low-level libraries), which tends to have many pointers and/or poor type safety. In high level code I make them much less commonly. Overall, I personally don’t find memory management in C++ to be too troublesome. ↑

Ah, he’s from Thales. They had a C++/Rust job listing open recently, for working with secure communication. Wouldn’t be surprised if Thales internationally were generally moving towards Rust. ↑

i’ve spent over three years each using C, then Rust, and then returning to C. Initially, i believed that borrow checking (and Rust as a whole) was a perfect solution. However, the more complex tasks i encountered, the more i realized that "memory safety" is an ideal that doesn’t fully exist in practice. <…​> Here’s the real challenge: proving that an unsafe block in Rust is both safe and sound is often more difficult than using raw pointers correctly in C or C++. ↑

The moment your program contains UB, all bets are off.

Even if it’s just one little UB.

Even if it’s never executed.

Even if you don’t know it’s there at all.

For instance, I was recently exploring Interprocess, only to find out I needed Boost.Date, which led me to implement it myself rather than dealing with the extra dependency.

And in your lack of experience, you committed a cardinal sin of programming. NEVER write your own datetime implementation unless you are prepared to dedicate a team to maintain it. Time zones are impossible, leap seconds will screw you, and numerous other pitfalls await. It’s almost always a terrible idea. Use a library that gets tons of real world use and debugging. ↑

Boost is now modular you don’t need to build the full Boost Set of libraries + with vcpkg it is straightforward : I don’t understand why people try to reimplement sub parts of Boost in worse way. Sometime you can find better libraries than the Boost ones and that perfect, but imo Boost is the first to be tried. ↑

We here use boost extensively, both server and embedded. Honestly I don’t know what all the fuss is about. It’s excellent code, well maintained by top folks in the C++ world. If you don’t like it, don’t use it. But don’t whine when other people do. ↑

I really think debugging should be taught in school. Not for any programming language. Kids should learn how to systematically approach a problem, gather diagnostic, follow cause and effect and how to communicate the problem to others. Regardless if this is computer stuff, plumbing or social sciences.

AI company: we trained this dog to talk. It doesn’t actually understand language, but it kinda sounds like it’s having a conversation by mimicking the sound of human speech.

CEO: awesome, I’ve fired my entire staff, how quickly can it start diagnosing medical disorders

Behind every remote worker is a cat that hasn’t signed an NDA and will sell all the secrets for a piece of sashimi.

One of the best programming skills you can have is knowing when to walk away for awhile.