@burntsushi5 Hey, just curious: Why does your bstr crate implement its own UTF-8 validation, Utf8Error type, etc., instead of using the standard library's core::str::from_utf8? Is yours optimized for something else?
Hard to answer in a tweet. It does have some (ASCII/SIMD related) optimizations over std. However, the main reason was available APIs. Consider how bstr would implement its `Chars` iterator using std APIs. I think it's possible, but is contorted and I think has overhead.