rename str::slice to str::slice_bytes

[erickt]

str::slice works on byte indices, so it would make more sense if the function was called str::slice_bytes.

[Kimundi]

I think it should stay as is, a str is just a transmuted [u8] with the additional invariant that it contains utf8, so it's consistent. Furthermore it would make the functions that actually allow to slice as a &[u8] more confusing.

[metajack]

nominating backwards compat

[bblum]

Can the function be used to cut apart unicode characters? I doubt the usefulness of such a function.

Whatever the answer, str::slice should be the name of a function that takes char indices, not byte indices.

[bluss]

str::slice has asserts to check that you don't slice in the middle of encoded codepoints. With that context, the slice_bytes name is not that great, because you have to slice it out as a valid string, not an arbitrary byte slice.

For working with strings, you either have to use byte indices or use character iterators, otherwise you run into the issues that come with char-based indexing being O(N).

[graydon]

This is working as intended. Slicing in the middle of a char is the same as slicing past the end of the array: it's a failure. We name the APIs with _char if they're character-based.