Python still has no standard and open antagonistic towards refactoring their standard library so that the language and the library aren't shipped as a monolith.
The most difficult part for alternative Python implementations is supporting the same stdlib.
Why does a language in which you write operating systems need a spec or a second compiler?
You have a compiler, that’s MIT licensed. It will be MIT licensed in perpetuity. It will continue to compile your code forever, with absolutely no changes needed to your code.
How precisely, does your life change if there is a second compiler or not? Or if there’s an spec? And more interestingly, why does it affect only operating systems development and not web server development?
That’s possible in theory, but do you think that’s what’s needed to improve the Rust compiler in practice?
The pace of development is pretty good, to the point where one of the main criticisms of Rust is that it moves too quickly. The stability is pretty good, I can only remember one time when code was broken. Compile times are mediocre but three ambitious projects hope to tackle that - parallel frontend, cranelift backend, changing the default linker to a better one(landing next month).
This argument would carry more weight if you could point out something specific that the Rust project isn’t addressing but might address if an alternate implementation existed. Remembering of course, that the second implementation would have to be compatible with the OG implementation.
If you don’t know in what way Rust needs to be improved, you don’t know enough to be making confident pronouncements of what it needs.
And like I’ve pointed out before on this thread, a second compiler does exist. It could help with bootstrapping because it’s written in C++ but that’s about it?
Plenty of successful languages have just the one implementation. Python is a great example. Slowing down python development by making everyone proposing a new feature have to convince two separate teams to adopt it wouldn’t help either the Python ecosystem or the Rust one.
Frankly, this thread has gone like every other thread on this subject. Advocates for a spec and second compiler are unable to come up with technical arguments in favour of it. Instead they simply say “well I’ve seen it done by this other language, so it should be done here as well.” That’s just cargo culting.
The only thing different about this thread is hearing from people that work in progress spec and second compiler aren’t enough. No, it needs to be ready right now!!! Yeah alright.
The most difficult part for alternative Python implementations is supporting the same stdlib.