4 ms·

>Rust found a ready audience in people intimidated by C++, frustrated with the limitations of C, Java, and Go, and ready to try something wholly new. What abou

>Rust found a ready audience in people intimidated by C++, frustrated with the limitations of C, Java, and Go, and ready to try something wholly new. What about people not intimidated by C++ and want to switch to Rust anyway? Why have you discounted them? Do such people exist in your world view? Do you think there might be a good reason for those people to choose Rust?

lordlic 6y ago
I read that the same way until I realized that it would be fair to say that even I'm "intimidated" by C++ despite having quite a bit of experience with it. It's not a matter of cleverness or experience. You can know something inside and out, and still admit that it's out of control and worry about the costs of its complexity.
ncmncm 6y ago
If you are not intimidated by C++, you will be using those features that intimidate others to construct better libraries and programs than they can. Switching to Rust would mean a huge step down, where vast gaps in the feature set disable expressing what has been so easy. There are kludgy workarounds for everything, more or less, but why switch to a limited language with immature tooling and practically nonexistent immediate prospects, when C++ demand is going up, up, and up? Memory safety is just not an issue at the level where coders of modern C++ work. I have had exactly one lifetime error in five years, caught in initial testing. My bugs are overwhelmingly specification omissions and logic mistakes that Rust suffers from exactly as much. My day-to-day experience is to write 2000 lines and, when it compiles, it works correctly, modulo a "!" short or extra. Then, I spend a few days refining the design, which often means deleting code. Things Rust promises to help with consume only trivial attention. Library features not expressible in Rust prevent many more and bigger problems. Rust has some nice qualities I would rather C++ had--pattern match support, simple iterators, default const and default pass-by-move, sane conversions, sane initialization--but they are all low-level conveniences, not anything I am willing to trade away architectura-level capabilities for.
- rational_indian 6y ago
  What architectural level capabilities?
  - ncmncm 6y ago
    Template partial specialization, function template overloading, variadic template argument lists, template operator overloading, generic lambdas, the list goes on and on and on. Look up any list of C++20 language features; most are not in Rust. Look up any list of Standard Library components; most are not expressible in Rust. Rust is going interesting places, is taking full advantage of starting from scratch, and is miles ahead of C, Java, and Go on every axis. But if Rust could get to where C++ was in 2011 by 2025, that would be amazing. Meanwhile, C++ has already shipped C++14, C++17, and C++20, will have done C++23, and will be firming up C++26. Assume Rust could match C++20, just five years later, in 2030, which would be more amazing. C++29 will be out.

lordlic 6y ago

I read that the same way until I realized that it would be fair to say that even I'm "intimidated" by C++ despite having quite a bit of experience with it. It's not a matter of cleverness or experience. You can know something inside and out, and still admit that it's out of control and worry about the costs of its complexity.

ncmncm 6y ago

If you are not intimidated by C++, you will be using those features that intimidate others to construct better libraries and programs than they can. Switching to Rust would mean a huge step down, where vast gaps in the feature set disable expressing what has been so easy. There are kludgy workarounds for everything, more or less, but why switch to a limited language with immature tooling and practically nonexistent immediate prospects, when C++ demand is going up, up, and up? Memory safety is just not an issue at the level where coders of modern C++ work. I have had exactly one lifetime error in five years, caught in initial testing. My bugs are overwhelmingly specification omissions and logic mistakes that Rust suffers from exactly as much. My day-to-day experience is to write 2000 lines and, when it compiles, it works correctly, modulo a "!" short or extra. Then, I spend a few days refining the design, which often means deleting code. Things Rust promises to help with consume only trivial attention. Library features not expressible in Rust prevent many more and bigger problems. Rust has some nice qualities I would rather C++ had--pattern match support, simple iterators, default const and default pass-by-move, sane conversions, sane initialization--but they are all low-level conveniences, not anything I am willing to trade away architectura-level capabilities for.

rational_indian 6y ago
What architectural level capabilities?
- ncmncm 6y ago
  Template partial specialization, function template overloading, variadic template argument lists, template operator overloading, generic lambdas, the list goes on and on and on. Look up any list of C++20 language features; most are not in Rust. Look up any list of Standard Library components; most are not expressible in Rust. Rust is going interesting places, is taking full advantage of starting from scratch, and is miles ahead of C, Java, and Go on every axis. But if Rust could get to where C++ was in 2011 by 2025, that would be amazing. Meanwhile, C++ has already shipped C++14, C++17, and C++20, will have done C++23, and will be firming up C++26. Assume Rust could match C++20, just five years later, in 2030, which would be more amazing. C++29 will be out.