Apple Clang uses different version numbering than "regular" clang, but
it also provides the "regular" version it's based on. Update the sed
pattern to pull out this "regular" version number instead of the Apple
version number.

Passing higher alignment values gives the optimization passes more freedom since it can copy in larger chunks. This change results in rustc outputting the same post-optimization IR as clang for swaps and most copies excluding the lack of information about padding.

Code snippet:

```rust
#[inline(never)]
fn swap<T>(x: &mut T, y: &mut T) {
    util::swap(x, y);
}
```

Original IR (for `int`):

```llvm
define internal fastcc void @_ZN9swap_283417_a71830ca3ed2d65d3_00E(i64*, i64*) #1 {
static_allocas:
  %2 = icmp eq i64* %0, %1
  br i1 %2, label %_ZN4util9swap_283717_a71830ca3ed2d65d3_00E.exit, label %3

; <label>:3                                       ; preds = %static_allocas
  %4 = load i64* %0, align 1
  %5 = load i64* %1, align 1
  store i64 %5, i64* %0, align 1
  store i64 %4, i64* %1, align 1
  br label %_ZN4util9swap_283717_a71830ca3ed2d65d3_00E.exit

_ZN4util9swap_283717_a71830ca3ed2d65d3_00E.exit:  ; preds = %3, %static_allocas
  ret void
}
```

After #6710:

```llvm
define internal fastcc void @_ZN9swap_283017_a71830ca3ed2d65d3_00E(i64* nocapture, i64* nocapture) #1 {
static_allocas:
  %2 = load i64* %0, align 1
  %3 = load i64* %1, align 1
  store i64 %3, i64* %0, align 1
  store i64 %2, i64* %1, align 1
  ret void
}
```

After this change:

```llvm
define internal fastcc void @_ZN9swap_283017_a71830ca3ed2d65d3_00E(i64* nocapture, i64* nocapture) #1 {
static_allocas:
  %2 = load i64* %0, align 8
  %3 = load i64* %1, align 8
  store i64 %3, i64* %0, align 8
  store i64 %2, i64* %1, align 8
  ret void
}
```

Another example:

```rust
#[inline(never)]
fn set<T>(x: &mut T, y: T) {
    *x = y;
}
```

Before, with `(int, int)` (align 1):

```llvm
define internal fastcc void @_ZN8set_282517_8fa972e3f9e451983_00E({ i64, i64 }* nocapture, { i64, i64 }* nocapture) #1 {
static_allocas:
  %2 = bitcast { i64, i64 }* %1 to i8*
  %3 = bitcast { i64, i64 }* %0 to i8*
  tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* %3, i8* %2, i64 16, i32 1, i1 false)
  ret void
}
```

After, with `(int, int)` (align 8):

```llvm
define internal fastcc void @_ZN8set_282617_8fa972e3f9e451983_00E({ i64, i64 }* nocapture, { i64, i64 }* nocapture) #1 {
static_allocas:
  %2 = bitcast { i64, i64 }* %1 to i8*
  %3 = bitcast { i64, i64 }* %0 to i8*
  tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* %3, i8* %2, i64 16, i32 8, i1 false)
  ret void
}
```

Preliminary implementation for: https://github.com/mozilla/rust/issues/6275

This is my first (non hello world) rust code, so it may not be idiomatic.

Fix #6145. In particular, handle operator overloading.

There were several old `#[doc(hidden)]` attributes in libstd and
libextra, left over from when rustdoc didn't hide private
definitions, tagged with `FIXME #3538`.

Since #3538 is now closed, I removed the `#[doc(hidden)]` attributes
as well as the FIXMEs, but I left `#[doc(hidden)]` in
libstd/task/spawn.rs and libstd/task/rt.rs since those two are
apparently `pub`, as well as in libextra/std.rc since std/extra is
`pub`.

There were several old `#[doc(hidden)]` attributes in libstd and
libextra, left over from when rustdoc didn't hide private
definitions, tagged with `FIXME #3538`.

Since #3538 is now closed, I removed the `#[doc(hidden)]` attributes
as well as the FIXMEs, but I left `#[doc(hidden)]` in
libstd/task/spawn.rs and libstd/task/rt.rs since those two are
apparently `pub`, as well as in libextra/std.rc since std/extra is
`pub`.

When I submitted #6748 yesterday, I used the old name.

r? @thestinger

When I submitted #6748 yesterday, I used the old name.

...mentioned in #2625.

This change makes the module more oriented around
Process values instead of having to deal with process ids
directly.

Apart from issues mentioned in #2625, other changes include:
- Changing the naming to be more consistent - Process/process
  is now used instead of a mixture of Program/program and
  Process/process.
- More docs/tests.

Some io/scheduler related issues remain (mentioned in #2625). I am not sure how best to address these.

@brson. Also fix a few documentation bugs.

directory to be the parent of the current-current directory,
instead of changing to the tmp directory, which was causing
issues with OS X and its /tmp => /private/tmp symlink.

mentioned in #2625.

This change makes the module more oriented around
Process values instead of having to deal with process ids
directly.

Apart from issues mentioned in #2625, other changes include:
- Changing the naming to be more consistent - Process/process
  is now used instead of a mixture of Program/program and
  Process/process.
- More docs/tests.

Some io/scheduler related issues remain (mentioned in #2625).

Whatever it was, it is no longer a problem here.

Fixes #6697

There was some before, but now we have a big header, as well as
lots of individual bits of documentation.

This indents all but the first line of multi-line annotations for individual enum variants with four spaces so that pandoc will recognize everything as belonging to the same list item.

Since that introduces `<p>` tags for some list items, I've gone ahead and inserted blank lines after each list item so that consistently get `<p>` tags for all `<li>`s documenting variants. It's a bit less compact now but still tolerable, I think.

There was some before, but now we have a big header, as well as
lots of individual bits of documentation.

This allows them to make use of the type's alignment, instead of being
pessimistic and assuming it is only 1.

This marks `&mut` function arguments with the `noalias` attribute. Since the borrow checker enforces this property, this is worth doing.

I'm not sure if the place I'm doing it in is ideal, but it generates the correct code.

Closes #6350

This indents all but the first line of multi-line annotations for
individual enum variants with four spaces so that pandoc will
recognize everything as belonging to the same list item.

Since that introduces `<p>` tags for some list items, I've gone ahead
and inserted blank lines after each list item so that consistently
get `<p>` tags for all `<li>`s documenting variants. It's a bit less
compact now but still tolerable, I think.

r? @brson (Yes, this did happen in real life...)

...and don't treat Path("/") like Path("").

(Yes, this did happen in real life...)

Apple Clang uses different version numbering than "regular" clang, but
it also provides the "regular" version it's based on. Update the sed
pattern to pull out this "regular" version number instead of the Apple
version number.

Changes the int/uint modules to all use macros instead of using the `merge` attribute. It would be nice to have #4375 resolved as well for this, but that can probably come at a later date.

Closes #4219.

This pull request is more of an RFC than a finished implementation.

It adds some basic atomic types, with an interface modelled off of C++11's atomic types.

It also adds free functions that provide a slightly nicer interface for atomic operations, though they are unsafe because there isn't a way to be generic over "word-sized" types.

See also #5042

Currently, keywords are stored in hashsets that are recreated for every
Parser instance, which is quite expensive since macro expansion creates
lots of them. Additionally, the parser functions that look for a keyword
currently accept a string and have a runtime check to validate that they
actually received a keyword.

By creating an enum for the keywords and inserting them into the
ident interner, we can avoid the creation of the hashsets and get static
checks for the keywords.

For libstd, this cuts the parse+expansion part from ~2.6s to ~1.6s.