Comment by Yoric
9 months ago
Well, as far as I can tell, in the big language shootout (which is still the only decent language benchmark I know, but if you have others, I'd be happy to read them), in all the individual benchmarks I've looked at Rust is either first or within 3% performance of being first. So, this suggests that Rust is actually pretty good for raw performance.
In addition, I know that I'm way more productive for prototyping and iteration in Rust than in C, C++ (and I used to write C++ code professionally for a while), JavaScript, Go, Python (and I've been writing professionally in these three languages for a while now) etc. I _might_ be more productive in Java or OCaml. Yes, this is entirely anecdotal and it depends heavily on the kind of problems you're dealing with. I tend to focus on problems that have complicated invariants, and for which reproducing/pinpointing the issue in a debugger is a big annoyance.
> Google, fb, amazon, etc. use C/C++ to squeeze the most performance out of anything I/O heavy, and security is not an issue that deep in the stack, that's not the exploitation layer.
Interestingly, all these companies are migrating some of their systems to Rust. This suggests that they find the language convincing enough.
> Interestingly, all these companies are migrating some of their systems to Rust. This suggests that they find the language convincing enough.
I feel most large companies and some smaller ones are interested in trying out Rust, it's trendy. But time will show for which parts the switch was advantageous; and I'm very interested in the findings. The premise of the language is indeed convincing a lot of people. People do however, choose the wrong tool for the wrong job all the time, OP's article in point.
Absolutely.