I agree. We are not starting out with a blank slate. There are already billions of lines of C++ in mission-critical production use. Companies such a Microsoft, Apple, Google, and Facebook run on C++. In addition, it has vendor support on all the major operating systems. In contrast, not even Mozilla, has bet the company on Rust (Servo is experimental and Mozilla has made no promises of actually integrating it into Firefox.) Also, Rust is not vendor supported on any platform.
So if you are a CTO or someone who has the responsibility of deciding what language to use for your performance critical software, you already have a large bias toward C++. You may be swayed toward Rust by the fact that Rust is memory safe. However, when someone comes and tells you that with this open source tool for C++ backed by the big names in C++, and developed and supported by Microsoft, and just by using the GSL types you can get most of the benefit of Rust, you will decide that the delta of improvement of Rust over C++ is not worth it.
TLDR: In programming language use pragmatic considerations trump purity. Multibillion dollar companies run on C++. With GSL and static analysis, C++ is "good enough".
That is very '90s, pre-programming-language-Renaissance thinking. If that thinking actually dominated, then you would never have seen the rise of Python, Ruby, Go, Node, Perl, PHP, Scala, Clojure, and so forth during the 2000s and 2010s. During the '90s, people predicted that Java would be the last programming language that anyone would ever use, using the same kinds of reasoning. That turned out to be false.
> You may be swayed toward Rust by the fact that Rust is memory safe. However, when someone comes and tells you that with this open source tool for C++ backed by the big names in C++, and developed and supported by Microsoft, and just by using the GSL types you can get most of the benefit of Rust, you will decide that the delta of improvement of Rust over C++ is not worth it.
Replace "Rust" with "Ruby on Rails", "C++" with "Java", and "GSL types" with "Spring", and you could make the same statement in 2005. It might even sound plausible. It would also be an incredibly wrong prediction.
> There are already billions of lines of C++ in mission-critical production use.
How many lines of memory-safe C++ are there in production use? The amount of memory-safe Rust code far exceeds that.
Ergonomics and the ecosystem matter. The fact is that the proposed static analysis tool for C++ is very restrictive and complex, from everything I've seen so far, and coupled with the annotation burden equal to that of Rust effectively makes it into a completely different language.
I agree that this might be a reason that people delay adoption, but I think things will change when the people who adopt Rust do better than the people who don't. This is similar to how things changed when people who adopted scripting languages for their web apps did better than people who didn't.
So if you are a CTO or someone who has the responsibility of deciding what language to use for your performance critical software, you already have a large bias toward C++. You may be swayed toward Rust by the fact that Rust is memory safe. However, when someone comes and tells you that with this open source tool for C++ backed by the big names in C++, and developed and supported by Microsoft, and just by using the GSL types you can get most of the benefit of Rust, you will decide that the delta of improvement of Rust over C++ is not worth it.
TLDR: In programming language use pragmatic considerations trump purity. Multibillion dollar companies run on C++. With GSL and static analysis, C++ is "good enough".