> Being in the position of "I need to learn a first systems language, should I pick C++ or Rust" is not very common, IMO.
What makes you think that. Everyone who learns systems programming is in this position at some point. Perhaps not if they learn at university and the language is chosen for them, but that just pushes the choice onto the university itself. And universities are increasingly offering Rust courses.
Of course, plenty of programmers aren't systems programmers and never will be. But that doesn't negate the need for systems programming languages, as demonstrated by the continued use of C and C++. If you do need a systems languages then Rust can make a good case for being the simplest. And could well become the de facto standard over the next few years.
C++ uses far exceed systems programming.
Rust to me is starting to feel like how Haskell was a decade ago, where it seemed like everybody was talking about it but serious adoption was always just on the horizon.
But maybe my expectations are just too high. It seemed quite exciting at first to see everything being rewritten in Rust but I guess that’s not how the software world really works. Old projects are entrenched because they take many years of dev time to mature
I think you might just not be aware of how widely adopted Rust already is. Google, Microsoft, Amazon, Dropbox, Cloudflare to name just a few companies already have important projects in Rust. Curl is adding a Rust backend. The Linux Kernel hasn't committed yet, but groundwork has begun and it's looking like it's going to happen. Serious adoption has already happened.
You're right that we won't see mass rewrites (although there have been some). That's just too big a project. But new projects are increasingly choosing Rust. It'll be a decade before Rust is the de facto standard if it happens. But personally I consider that quite likely.
That's some very thoughtful and noble goal. Maybe people who use C++ (and bunch of other languages) are not the ones whining.
1) is as safe and as fast as Rust, or at most 1.5x slower
2) offers "fearless concurrency" and easy, safe, and efficient parallelism
3) has sum types and a good static type system
4) has great tooling and package management (see cargo)
5) has a vibrant and growing ecosystem
So far, that language doesn't seem to exist. Most fail at points 1+2.
1) reasoning about performance is just too difficult (due to lazy evaluation)
2) the community seems too focused on providing ever more abstract abstractions instead of practical solutions to actual problems
3) purity and freedom from side-effects cause a lot of friction and make many simple problems really hard; the libraries and patterns to solve that problem are too complicated for me to comprehend, so maybe I'm too dumb for Haskell (though I don't have any problems with Rust, so...)