String class is error prone and has usability issues about rfcs HOT 2 OPEN

A good reference to use for a safe string API would be Rust's std::string::String and std::ffi::OsString which allow users to interact with valid UTF-8 strings and platform-specific string representations respectively.

There are a few differences between Rust's string design and what is proposed above.

• Rust strings have a String::chars method which allows iterating over the char type instead of treating it as a sequence of strings itself.
• Rust additionally offers a String::char_indices method that allows you to see the byte indices of each character you iterate over
• Rust does allow users to do some indexing into strings as byte arrays, but doing so risks a panic (example with index) or returns an Option (example with get).

Users have a tendency to assume indexing is a constant time operation which is not the case for character indexing into UTF-8 and I believe is why that ability is only offered through an iterator (e.g. s.chars().nth(...)) in Rust.

from rfcs.

FYI there's a String RFC that is being worked on so anyone interested in this might want to check that out:

#179

from rfcs.