Add Utf8Error::error_len, to help incremental and/or lossy decoding.
Without this, code outside of the standard library needs to reimplement most of the logic `from_utf8` to interpret the bytes after `valid_up_to()`.
Leftovers from #39594; From<Box> impls
These are a few more impls that follow the same reasoning as those from #39594.
What's included:
* `From<Box<str>> for String`
* `From<Box<[T]>> for Vec<T>`
* `From<Box<CStr>> for CString`
* `From<Box<OsStr>> for OsString`
* `From<Box<Path>> for PathBuf`
* `Into<Box<str>> for String`
* `Into<Box<[T]>> for Vec<T>`
* `Into<Box<CStr>> for CString`
* `Into<Box<OsStr>> for OsString`
* `Into<Box<Path>> for PathBuf`
* `<Box<CStr>>::into_c_string`
* `<Box<OsStr>>::into_os_string`
* `<Box<Path>>::into_path_buf`
* Tracking issue for latter three methods + three from previous PR.
Currently, the opposite direction isn't doable with `From` (only `Into`) because of the separation between `liballoc` and `libcollections`. I'm holding off on those for a later PR.
Whenever we parse a chain of binary operations, as long as the first
operation is `<` and the subsequent operations are either `>` or `<`,
present the following diagnostic help:
use `::<...>` instead of `<...>` if you meant to specify type arguments
This will lead to spurious recommendations on situations like
`2 < 3 < 4` but should be clear from context that the help doesn't apply
in that case.
@jongiddy noticed bad performance due to the lack of inlining on `then`
and `then_with`. I confirmed that inlining really is the culprit by
creating a custom `then` function and repeating his benchmark on my
machine with and without the `#[inline]` attribute.
The numbers were exactly the same on my machine without the attribute.
With `#[inline]` I got the same performance as I did with manually
inlined implementation.
Their relationship is:
* `resume_from = error_len.map(|l| l + valid_up_to)`
* error_len is always one of None, Some(1), Some(2), or Some(3).
When I started using resume_from I almost always ended up subtracting
valid_up_to to obtain error_len.
Therefore the latter is what should be provided in the first place.
We don't want a dynamic dependency in the library that we ship, so link it
statically by configuring curl-sys's build script to not pick up the system
version via pkg-config.
Primarily opened to address the concerns brought up in
https://github.com/rust-lang/rust/issues/40498.
* run rustfmt on code blocks
* use `DefaultHasher` instead of deprecated `SipHasher`
* rename `hash` to `calculate_hash` to prevent confusion with the `hash`
method
