We're not outright removing fmt! just yet, but this prevents it from leaking
into the compiler further (it's still turned on by default for all other code).

Closes #9446

r? @alexcrichton test output may contain color codes, so check for several small strings
instead of one big one.

test output may contain color codes, so check for several small strings
instead of one big one.

Closes #9621

Closes #7999

Closes #7999

This is mostly an incremental change, picking off some uses of
@- or @mut-pointers that can be replaced by references.

Almost all of the builder functions in trans::build are updated,
mostly using `&Block` arguments instead of `@mut Block`.
 

Use &mut Block and &Block references where possible in the builder
functions in trans::build.

@mut Block remains in a few functions where I could not (not yet at
least) track down the runtime borrowck failures.

Use borrowed references in a few locations that were encountered when
working on other changes.

Closes #9621

Closes #8815.

Cleanup in middle::lang_items

Use static string keys in the lang item hashmap, and convert
the traversal function to be an iterator.

Closes #8815.

This patch exposes actual ownership of an `ast::Crate` structure so it's not implicitly copied and reference counted via `@`.

The main purpose for this patch was to get rid of the massive spike in memory during the start of the compiler (this can be seen on isrustfastyet). The reason that this spike exists is that during `phase_2` we're creating many copies of the crate by folding. Because these are reference counted, all instances of the old crates aren't dropped until the end of the function, which is why so much memory is accumulated.

This patch exposes true ownership of the crate, meaning that it will be destroyed ASAP when requested. There are no code changes except for dealing with actual ownership of the crate. The large spike is then avoided: http://i.imgur.com/IO3NENy.png

This shouldn't help our overall memory usage (that still is the highest at the end), but if we ever manage to bring that down it should help us not have a 1GB spike at the beginning of compilation.

(This was to un-stuck bors (hopefully).)

Forgot to do this when I was refactoring logging :(

This help enable some later refactorings.

Previously, if tests failed, you'd only get stderr which isn't very
useful, especially if the failure didn't happen directly in a test
function (e.g None.unwrap()).

Previously, if tests failed, you'd only get stderr which isn't very
useful, especially if the failure didn't happen directly in a test
function (e.g None.unwrap()).

r? strcat 

(please double check that it fixes things on your chroot, if possible, before approval)

Fix #9531

They're getting smaller each time though!

The highlight of this round is source files in documentation. Still trying to figure out the best syntax-highlighting solution.

This removes the internal type representation of an `External` type and instead
relies on passing around DefId structures and interpreting them accordingly.

Progress on #9539, but there's still the problem of a crate => url mapping.

std::vec: Sane implementations for connect_vec and concat_vec

Avoid unnecessary copying of subvectors, and calculate the needed space
beforehand. These implementations are simple but better than the
previous.

Also only implement it once, for all `Vector<T>` using:

    impl<'self, T: Clone, V: Vector<T>> VectorVector<T> for &'self [V]

Closes #9581

change formula (other solution could be using abs()).

Small change that suppresses a warning because of an unused import.