{"id":32478305,"user":"matheusmoreira","time":1660614333,"time_ago":"3 years ago","type":"comment","content":"

> Now, do I think it would make total sense for syscall wrappers and NSS to be split into their own libs (or dbus interfaces maybe) with stable ABIs to enable other libc's, absolutely!

I worked on this a few years ago: liblinux.

https://github.com/matheusmoreira/liblinux

It's still a nice proof of concept but I abandoned it when I found out the Linux kernel itself has a superior nolibc library that they use for their own tools:

https://github.com/torvalds/linux/tree/master/tools/include/...

It used to be a single header but it looks like they've recently organized it into a proper project!

> This is something the BSD's got absolutely right.

BSDs and all the other operating systems force us to use their C libraries and the C ABI. I think Linux's approach is better. It has a language-agnostic system call binary interface: it's just a simple calling convention and the system call instruction.

The right place for system call support is the compiler. We should have a system_call keyword that makes it emit code in the aforementioned calling convention. With this single keyword, it's possible to do literally anything on Linux. Wrappers for every specific system call should be part of every language's standard library with language-specific types and semantics.

An example of one such language is Virgil by HN used titzer:

https://news.ycombinator.com/item?id=28283632","comments":[],"comments_count":0,"url":"item?id=32478305"}

Comment by matheusmoreira