str convert-from-bytes API cleanup

[erickt]

Good evening. This patch series cleans up the string API. str::from_bytes* was a bit of a mess. We had ways of taking a &[u8] and getting a ~str or &str, but no way to cheaply move a ~[u8] into a ~str. This patch series adds or changes to str:

fn from_bytes(v: ~[u8]) -> ~str;
fn from_bytes_with_null(v: ~[u8]) -> ~str;

fn from_bytes_slice<'a>(v: &'a [u8]) -> &'a str;
fn from_bytes_slice_with_null<'a>(v: &'a [u8]) -> &'a str;

and adds these to str::raw:

fn buf_len(buf: *u8) -> uint;

fn from_buf(buf: *u8) -> ~str;
fn from_buf_len(buf: *u8, len: uint) -> ~str;

fn buf_as_slice<T>(buf: *u8, f: &fn(&str) -> T) -> T;
fn buf_len_as_slice<T>(buf: *u8, len: uint, f: &fn(&str) -> T) -> T;

fn c_str_as_slice<T>(buf: *libc::c_char, f: &fn(&str) -> T) -> T;

And removes str_byte_slice and str_byte_slice_no_callback as those two functions are redundant. This fixes #7019.

[pcwalton]

I'm not a big fan of the callback-based functions, personally. I'd rather just have buf_as_slice return &str, the rationale being that as it's an unsafe function, you could just as easily transmute the &str out of the closure.

[pcwalton]

Oh, and what about having from_buf return &str? If the caller wants an ~str, he or she can just use .to_str() on the result…

[graydon]

see also 967c7d8 and #3843

[huonw]

@erickt Sorry for stepping on your toes a bit with #7060, by the way.