For the vast majority of software you want a managed runtime.
Some of the problems Rust “solves” are problems you shouldn’t be having in the first place because we mostly write software that doesn’t need direct control over memory. Borrow checking isn’t something you want to have to deal with - it is something you have to accept when you have chosen to manage memory. That choice has a high cost that cost never gets paid off in most projects that could work just as fine with managed memory.
I’m a Go programmer, but this article reminded me that I should have more experience with Rust. From my perspective Rust seems a bit less practical. The standard library lacks support for cryptography, for instance. The compiler is slow, which is a productivity killer. Overall concurrency seems like a bit of an afterthought. Again.
What makes me want to try Rust in production are things like option types. Those would be nice to have in any language. Any issues that can be caught by the compiler are a plus. Getting rid of nil would also be a plus, but to be quite frank, I don’t experience that many nil pointer errors.
The author does nod to the static analysis tools for Go. Yes, they are not part of the compiler (for good reason), but they do a pretty good job in practice. So you get more than the compiler can promise at a fraction of the cost (measured in build time). That’s a much bigger deal for actual developers than we generally give it credit for.
Then there’s the stuff that makes me less convinced in terms of arguments. For instance the fact that Go didn’t have generics early on and that the standard library doesn’t use them. Generics were not as important as people thought they were. In practical reality. The fact that the standard library doesn’t make wide use of them is not a weakness, it shows restraint. They didn’t go overboard and prematurely plaster generics all over the place as soon as the language supported it. This is the kind of restraint you want to see. Remember how horrible Java was after everyone started abusing generics? A brief generation of software that was significantly worse, and less maintainable resulted from this exuberance. For the ultimate example of what happens when you give people every feature they wish for: look at C++. It´s not a very good language because it is many languages. Just because there are standards and recommendations doesn’t mean that all code magically gets rewritten to a narrower definition of the language. It means that we accumulate intermediate forms. I expect people who are interested in languages to understand these dynamics.