rustc: Remove &str indexing from the language.

[brson]

Being able to index into the bytes of a string encourages
poor UTF-8 hygiene. To get a view of &[u8] from either
a String or &str slice, use the as_bytes() method.

Closes #12710.

[breaking-change]

If the diffstat is any indication this shouldn't have a huge impact but it will have some. Most changes in the str and path module. A lot of the existing usages were in tests where ascii is expected. There are a number of other legit uses where the characters are known to be ascii.

[emberian]

It might be valuable to have an as_ascii method that returns &[Ascii]

[alexcrichton]

As a language change, this has no RFC corresponding with it, but it's small enough that it may be ok to discuss here instead.

I personally would like to keep indexing. It is a fact that all our strings are utf-8 encoded, and that fact will never change. The type system will very quickly tell you you're getting a byte and not a char.

[aturon]

@alexcrichton As @brson points out, you can get the utf-8 view from as_bytes, so that's not a very compelling argument by itself.

On the other hand, a bunch of the other methods on &str use byte indices, e.g. slicing, char_at, find. So we're not really hiding the fact the utf-8-ness of strings. It seems like we need a consistent rationale about where and how to permit working with string data as utf-8-encoded.

@brson Do you see these other methods as also encouraging poor hygiene? Or is there something that makes indexing particularly pernicious?

[alexcrichton]

It seems that we have a bit of a push-back from any byte-based methods on strings, but there are so many that I think it's ergonomically infeasible to remove all of them. It seems a little odd to remove half but leave the other half, leading to my opinion that this is a nice sugar that should stick around.

[brson]

@aturon I don't have a strong opinion about whether this is the correct thing to do or not. It has been previously-accepted as a milestone issue, so I implemented it.

[brson]

Per https://github.com/rust-lang/meeting-minutes/blob/master/weekly-meetings/2014-07-01.md, we're going to do this.

[alexcrichton]

r=me with a rebase

[SimonSapin]

@cmr, the to_ascii and into_ascii methods already exist, but you have to use the traits from std::ascii.