The fact that JSON is UTF-8 doesn't contradict the fact that it's bytes!

That's a feature, not a bug.

i.e. "exterior designs are layered" - https://www.oilshell.org/blog/2023/06/ysh-design.html#exteri...

This is not a trivial point -- there are plenty of specs which are sloppy about text, try to abstract over code points, and result in non-working software.

A main example is the shitshow with libc and Unicode - https://thephd.dev/cuneicode-and-the-future-of-text-in-c#sta...

It suffers from what I call "interior fictions", like wchar_t.

Of course it’s all bytes. It’s all bytes. That doesn’t change the fact that you need to have some awareness of encoding before those bytes are fully sensible