Glad to see promising new languages pushing progress forward, though. I'm head high in C/C++/asm for decades so many of the complaints about those languages fall flat with me since I'm over the hurdles via sheer years of experience.
I think people are pretty straightforward about Rust's challenges. The borrow checker can be a hurdle and probably needs more work (although it is getting a big boost soon with non-lexical lifetimes). Compile times are not great. The macro system is powerful but can be confusing. People from some mainstream languages may have trouble wrapping their heads around the type system.
Regarding its features, these attributes:
* it can do low level, fast code;
* it's not garbage collected, but is memory safe;
* it has zero cost abstractions (and, conversely, it doesn't have ridiculous, unnecessary, slow abstractions);
* it doesn't ignore the last thirty years of innovation in programming languages (by which I mean it has a modern type system, closures, variant types, and a few other things);
* it doesn't have 30-40 years of cruft;
* it has a standard and very high quality build/packaging system;
* its developers are thoughtful and have good taste;
* it has a reasonably big community and has a good chance of catching on even more widely;
are huge. You say you think its features are exaggerated, but to my mind, what's to exaggerate? That list looks pretty great to me.
It's hard to refute or agree with you when I don't know what you're talking about. :)
The challenges are not exaggerated, but they do have solutions, so they're not deal-breakers in practice.
The learning curve is steep. It takes effort, but once you get it (and unlearn some C-isms), Rust is a quite productive language.
Some patters and data structures are not understood by borrow checker, so you have to fall back to coding as-safely-as-any-C program. However, thanks to the practice of wrapping risky bits in safe abstractions, they remain only a small fraction of any program.
It might feel that writing any "unsafe" code defeats the promise of Rust, but it doesn't. The concept of safe abstraction is quite powerful, and makes sense even for wrapping 100% C libraries, where you can encode informal constraints form the manual as code, and automatically prevent library API from being misused.
If you have experienc with large C++ code bases you probably know how impossible it is to write safe code. Also after you use only unique/linear types, single inharitance (which is advised when working with a big code base), you are writing something that's easier to express in Rust than in C/C++.
Inlined asm will stay of course, Rust doesn't try compete with it.
The fact that large C and C++ code bases exist and the products they drive are successful proves that it is possible to write code in those languages that offers a good enough risk/reward balance.
This is what many Rust fans are missing: you want to sell 100% memory safety, but you're assuming that everyone wants 100% instead of 80 or 90% and that they're willing to pay the price of switching.
* I can work in C and C++, which I can get to X% safe using the techniques and tools A, B, C at the cost of M.
OR
* I can invest in learning Rust and integrating it with my environment to get to 100% at the cost of N.
Is the difference from X% to 100% big enough to pay N instead of M? How big of an X do I need? Is it even possible for me to use A, B and C?
This is how an informed decision would look like, but Rust proponents just keep repeating that X < 100.
This cost-benefit analysis also is substantially different between switching an existing codebase and starting a new one. The fact that Mozilla considers it worthwhile to switch an extremely large C++ codebase to Rust is worth noting.
Not by the Rust development team, which is what matters. They are keenly aware of the challenges, and have done a huge amount of work (particularly over the last six to twelve months) working on ergonomics and developer experience.
It's a primary focus for the Rust community, now and into the future.
Many (perhaps most?) of the specifications lined up for the current "impl period" (a time for building rather than designing) are related to developer ergonomics in some form.