As CloudABI is a capability-based runtime environment, file descriptors
are the mechanism that grants rights to a process. These file
descriptors may be passed into processes on startup using a utility
called cloudabi-run. Unlike the POSIX shell, cloudabi-run does not
follow the UNIX model where file descriptors 0, 1 and 2 represent stdin,
stdout and stderr. There can be arbitrary many (or few) file descriptors
that can be provided. For this reason, CloudABI's C library also doesn't
define STD*_FILENO. liblibc should also not declare these.

Disable the code in liballoc_system that tries to print error messages
over file descriptor 2. For now, let's keep this function quiet. We'll
see if we can think of some other way to log this in the future.

After rust-lang#47089 lands, CloudABI will no longer be considered UNIX. We need
to pick the right allocator flavour now.

CloudABI doesn't do dynamic linking. For this reason, there is no need
to handle any other TLS model than local-exec. CloudABI's C library
doesn't provide a __tls_get_addr() function to do Dynamic TLS.

By forcing local-exec to be used here, we ensure that we don't generate
function calls to __tls_get_addr().

Proc macro spans serve two mostly unrelated purposes: controlling name
resolution and controlling error messages. It can be useful to mix the
name resolution behavior of one span with the line/column error message
locations of a different span.

In particular, consider the case of a trait brought into scope within
the def_site of a custom derive. I want to invoke trait methods on the
fields of the user's struct. If the field type does not implement the
right trait, I want the error message to underline the corresponding
struct field.

Generating the method call with the def_site span is not ideal -- it
compiles and runs but error messages sadly always point to the derive
attribute like we saw with Macros 1.1.

```
  |
4 | #[derive(HeapSize)]
  |          ^^^^^^^^
```

Generating the method call with the same span as the struct field's
ident or type is not correct -- it shows the right underlines but fails
to resolve to the trait in scope at the def_site.

```
  |
7 |     bad: std::thread::Thread,
  |     ^^^^^^^^^^^^^^^^^^^^^^^^
```

The correct span for the method call is one that combines the def_site's
name resolution with the struct field's line/column.

```
field.span.resolved_at(Span::def_site())

// equivalently
Span::def_site().located_at(field.span)
```

Adding both because which one is more natural will depend on context.

For testing if the TrustedRandomAccess implementation works.

rust-lang#47113 renamed the private size
field to chunk_size for consistency.

The function was added as a wrapper to handle compatibility with older
LLVM versions that we no longer support, so it can be removed.

Refs rust-lang#46437

Refs rust-lang#46437

…,ChunksMut}

The example with the ? operator was missing file.write_all

Allow non-alphabetic underscores in camel case

Certain identifiers, such as `X86_64`, cannot currently be unambiguously represented in camel case (`X8664`, `X86_64`, `X8_664`, etc. are all transformed to the same identifier). This change relaxes the rules so that underscores are permitted between two non-alphabetic characters under `#[forbid(non_camel_case_types)]`. Fixes rust-lang#34633 and fixes rust-lang#41621.

Correct a few stability attributes

* The extra impls for `ManuallyDrop` were added in rust-lang#44310 which was only stabilised in 1.22.0.
* The impls for `SliceIndex` were stabilised in rust-lang#43373 but as `RangeInclusive` and `RangeToInclusive` are still unstable the impls should remain unstable.
* The `From` impls for atomic integers were added in rust-lang#45610 but most atomic integers are still unstable.
* The `shared_from_slice2` impls were added in rust-lang#45990 but they won't be stable until 1.24.0.
* The `Mutex` and `RwLock` impls were added in rust-lang#46082 but won't be stable until 1.24.0.

Disable printing of error message on file descriptor 2 on CloudABI.

As CloudABI is a capability-based runtime environment, file descriptors
are the mechanism that grants rights to a process. These file
descriptors may be passed into processes on startup using a utility
called cloudabi-run. Unlike the POSIX shell, cloudabi-run does not
follow the UNIX model where file descriptors 0, 1 and 2 represent stdin,
stdout and stderr. There can be arbitrary many (or few) file descriptors
that can be provided. For this reason, CloudABI's C library also doesn't
define STD*_FILENO. liblibc should also not declare these.

Disable the code in liballoc_system that tries to print error messages
over file descriptor 2. For now, let's keep this function quiet. We'll
see if we can think of some other way to log this in the future.

Use the right TLS model for CloudABI.

CloudABI doesn't do dynamic linking. For this reason, there is no need
to handle any other TLS model than local-exec. CloudABI's C library
doesn't provide a __tls_get_addr() function to do Dynamic TLS.

By forcing local-exec to be used here, we ensure that we don't generate
function calls to __tls_get_addr().

…s, r=kennytm

Implement TrustedRandomAccess for slice::{Chunks, ChunksMut, Windows}

As suggested by @bluss in rust-lang#47115 (comment)

Span::resolved_at and Span::located_at to combine behavior of two spans

Proc macro spans serve two mostly unrelated purposes: controlling name resolution and controlling error messages. It can be useful to mix the name resolution behavior of one span with the line/column error message locations of a different span.

In particular, consider the case of a trait brought into scope within the def_site of a custom derive. I want to invoke trait methods on the fields of the user's struct. If the field type does not implement the right trait, I want the error message to underline the corresponding struct field.

Generating the method call with the def_site span is not ideal -- it compiles and runs but error messages sadly always point to the derive attribute like we saw with Macros 1.1.

```
  |
4 | #[derive(HeapSize)]
  |          ^^^^^^^^
```

Generating the method call with the same span as the struct field's ident or type is not correct -- it shows the right underlines but fails to resolve to the trait in scope at the def_site.

```
  |
7 |     bad: std::thread::Thread,
  |     ^^^^^^^^^^^^^^^^^^^^^^^^
```

The correct span for the method call is one that combines the def_site's name resolution with the struct field's line/column.

```rust
field.span.resolved_at(Span::def_site())

// equivalently
Span::def_site().located_at(field.span)
```

Adding both because which one is more natural will depend on context.

Addresses rust-lang#38356 (comment). r? @jseyfried

Return None from Span::join if in different files

Fixes rust-lang#47148. r? @abonander

…lexcrichton

This isn't in Rust 1.23

r? @alexcrichton

Remove some outdated LLVM-related code

Ticks two boxes on rust-lang#46437

Equivalent example for ? operator

The example with the ? operator in the documentation for try! macro was missing file.write_all.
Now all three examples are consistent.