@Alex Crichton Is there a reason why libpanic_abort uses
#[rustc_std_internal_symbol] while libpanic_unwind uses
I think it's just a mistake
@Amanieu fwiw, I'm planning on updating my PR shortly which should make our implementation as good as C++ examples I can test
@simulacrum Oh dear, I'm almost done with my PR which also does exactly that :D
well maybe good to compare
https://github.com/rust-lang/rust/pull/67502 -- I ended up updating my previous branch
Looks almost exactly like mine modulo additional fixes to emcc
@Amanieu do you want to make the PR then? I can close mine out
@simulacrum We can use yours as a base. I will add some feedback in a bit (somewhat busy atm).
Okay, sounds good -- feel free to push to my branch as well (you should have rights I think).
Cherry picked a few of your commits
@Amanieu back from break now and can follow up! @simulacrum is right here in that it's probably just a mistake, the main difference is symbol visibility in final cdylib/dylib artifacts, where we want these internal symbols hidden whereas normal no_mangle symbols are exported