Hacker News Reader

Comment by toast0 | Hacker News Reader

> Do we not want the next generation to exclusively use safe languages?

Someone has to write (and maintain) the safe languages.

You can easily write the compilers/interpreters/other tooling in high level, memory safe languages. Not all, but quite a few of those safe languages have their tooling built in the same language

PyPy is written in Python, golang has a compiler made in go, etc. Many others do rely in part on LLVM, but in the "long future" there's no technical reason why a common component like LLVM has to be written in C/C++, who knows, people might switch to something else written in a memory-safe language.

Counterexample: Rust compiler is written in Rust.

Doesn't Rust compilation rely on LLVM for all the actual compilation from IR to machine code?

For now it does, but there's an upcoming experimental alternative backend written in Rust. Furthermore, there's no reason that LLVM has to be written in C++, it was just the best choice available when LLVM got started 15 years ago. Suffice to say, there are higher priorities for the Rust project than porting LLVM. :P

Yes, but there is a research project here that replaces LLVM for codegen: https://github.com/stoklund/cretonne

Not like it requires C to implement a new language.

This is a remarkably common belief and it bugs me to no end. I don't know why, but many people have developed this conviction that C is required for computers to work, that that's the baseline of computation. This comes from both novices and experts. I haven't figured it out.

The C Abstract Machine looks like a generic machine based on the Von Neumann architecture, so it "feels" like it should be fundamental. Add in some sunk-cost fallacy, and a lot of people end up with the belief that C is fundamental to computing, despite the existence of systems that do not use any C.

Reply on news.ycombinator.com