|
> can't help bashing it. Perhaps this may disappoint you, but I -- like many and perhaps most developers -- don't get such emotional responses, positive or negative to any programming language [1], which might appear as resentment to the emotionally attached. I am very impressed with some aspects of Rust, less impressed with others, and overall, my feelings toward it overall are shaped just like yours: by personal aesthetics. I don't find Rust's aesthetics very appealing and so Rust isn't my cup of tea (although I would't resent working in it because I'm not emotional toward languages and I currently program mostly in a language whose aesthetics I like even less than Rust's [2]), while you find them appealing and so you do like Rust. It's all just a matter of taste, and I fully accept that not everyone shares mine. I think your approach is too coloured by emotion, and is therefore unconstructive. You're a zealot, and you project that attitude on others, so "unconvinced" appears to you as a personal attack, and scepticism or dislike seems to you like bashing. > Rust is even less complex than C, because of the “most vexing parse” issue. If you wanted a rigorous analysis of the syntactic complexity, you could attempt to measure how difficult it is to write a lexer and a parser for every popular languages, and see how Rust performs No. The complexity of a formal language, like that of any set, is the computational complexity of deciding whether a string is in the language (so, including type-checking), not as the complexity of the parsing phase (https://en.wikipedia.org/wiki/Computational_complexity_theor...). I'm not saying this is the most useful way to talk about language complexity in this context (and caveats are needed, anyway, to make a more fine distinction between languages), but that is certainly one well-known way to talk about the intrinsic complexity of a language. > Zig is a cool motorbike, Rust is a SUV. Arguing that your bike can indeed be safer than a SUV because you have more visibility and agility to avoid the danger is beyond childish. It is beyond childish to make such inane statements about software correctness when you're clearly not very familiar with the subject, and are drawn to arguments like "more correctness => more soundness". The effect of language design on correctness is a complex subject with mostly unsatisfying answers, and even in software verification research, the debate over the value of soundness is far from settled (and not currently leaning toward more soundness). An equally inane statement would be, "Zig is like a modern aeroplane, relying on multiple levels of safety, some mechanical and some human, while Rust is like an old train that breaks down and kills everyone once there's a problem with the tracks." If we've learned anything about software correctness in the past decade it's that there is not much we can assume in advance, and that we don't really know one best way to improve it. It is true that some researchers think that the best answer to any correctness issue is more soundness in the language, but not only is this not a consensus opinion, I doubt it's even a majority opinion. [1]: I would say I'm a "language sceptic." I'm generally sceptical toward any empirical claim about the bottom-line effectiveness of linguistic features without empirical support, and overall think that whatever empirical studies we do have show little impact overall to language design (comparing "reasonable" alternatives, at least), certainly compared to what all language fans claim. I would never, say, make a definitive claim like, "Zig yields more correct programs than Rust", or "Rust yields more correct programs than Zig," without clear empirical support (and my guess based on prior results would be that they're about the same). [2]: C++ |
Your little “I'm a rational agent, you are too emotional” is pretty cute. But it would work better if your whole attitude in this thread didn't contradicted it, don't you think? “Rust is among the five most complex language” is not a rational argument, it's a personal feeling. Why you feel the need to spread your feelings over the internet while pretending you're not an “emotional” person is quite intriguing. If you want to look more like a rational person (no human really is), try to keep as much personal and unsubstantiated judgement out of your writings.
“Rust marketing makes safety claims that we should not take at face values” is alright, “Rust is one of the five most complex language ever” doesn't pass this test.
> is the computational complexity of deciding whether a string is in the language (so, including type-checking)
But for both Rust, C, Zig, and many others, it's undecidable, so by this definition of complexity, these language are definitely too complex (and equally so).[1] In fact, your desperate attempt to save your initial argument about complexity, by narrowing it to a tiny technical corner makes me cringe a bit.
> I would never, say, make a definitive claim like, "Zig yields more correct programs than Rust", or "Rust yields more correct programs than Zig," without clear empirical support (and my guess based on prior results would be that they're about the same).
The technicality of what constitute a “definitive claim” in a human to human conversation is an interesting question, but in practice truisms like “we can't conclude whether Rust is safer than C” or “we can't conclude that Rust doesn't bring more bugs than Zig does” aren't neutral: what such claim attempts to do is insinuate the opposite. And when combined with gratuitous judgements like “Rust is one of the 5 most complex language ever”, it looks a lot like an attempt to deter people from using this language you don't like. (And now I have a clue on what the root cause of your bad feelings can be).
[1] And as I said earlier, in the case of Zig and unsafe Rust and it's not just about type-checking: because of UB, even after compilation whether the compiled binary is the binary of a valid program or not is also undecidable.