Perform lifetime elision (more) syntactically, before type-checking.
The *initial* goal of this patch was to remove the (contextual) `&RegionScope` argument passed around `rustc_typeck::astconv` and allow converting arbitrary (syntactic) `hir::Ty` to (semantic) `Ty`.
I've tried to closely match the existing behavior while moving the logic to the earlier `resolve_lifetime` pass, and [the crater report](https://gist.github.com/eddyb/4ac5b8516f87c1bfa2de528ed2b7779a) suggests none of the changes broke real code, but I will try to list everything:
There are few cases in lifetime elision that could trip users up due to "hidden knowledge":
```rust
type StaticStr = &'static str; // hides 'static
trait WithLifetime<'a> {
type Output; // can hide 'a
}
// This worked because the type of the first argument contains
// 'static, although StaticStr doesn't even have parameters.
fn foo(x: StaticStr) -> &str { x }
// This worked because the compiler resolved the argument type
// to <T as WithLifetime<'a>>::Output which has the hidden 'a.
fn bar<'a, T: WithLifetime<'a>>(_: T::Output) -> &str { "baz" }
```
In the two examples above, elision wasn't using lifetimes that were in the source, not even *needed* by paths in the source, but rather *happened* to be part of the semantic representation of the types.
To me, this suggests they should have never worked through elision (and they don't with this PR).
Next we have an actual rule with a strange result, that is, the return type here elides to `&'x str`:
```rust
impl<'a, 'b> Trait for Foo<'a, 'b> {
fn method<'x, 'y>(self: &'x Foo<'a, 'b>, _: Bar<'y>) -> &str {
&self.name
}
}
```
All 3 of `'a`, `'b` and `'y` are being ignored, because the `&self` elision rule only cares that the first argument is "`self` by reference". Due implementation considerations (elision running before typeck), I've limited it in this PR to a reference to a primitive/`struct`/`enum`/`union`, but not other types, but I am doing another crater run to assess the impact of limiting it to literally `&self` and `self: &Self` (they're identical in HIR).
It's probably ideal to keep an "implicit `Self` for `self`" type around and *only* apply the rule to `&self` itself, but that would result in more bikeshed, and #21400 suggests some people expect otherwise.
Another decent option is treating `self: X, ... -> Y` like `X -> Y` (one unique lifetime in `X` used for `Y`).
The remaining changes have to do with "object lifetime defaults" (see RFCs [599](https://github.com/rust-lang/rfcs/blob/master/text/0599-default-object-bound.md) and [1156](https://github.com/rust-lang/rfcs/blob/master/text/1156-adjust-default-object-bounds.md)):
```rust
trait Trait {}
struct Ref2<'a, 'b, T: 'a+'b>(&'a T, &'b T);
// These apply specifically within a (fn) body,
// which allows type and lifetime inference:
fn main() {
// Used to be &'a mut (Trait+'a) - where 'a is one
// inference variable - &'a mut (Trait+'b) in this PR.
let _: &mut Trait;
// Used to be an ambiguity error, but in this PR it's
// Ref2<'a, 'b, Trait+'c> (3 inference variables).
let _: Ref2<Trait>;
}
```
What's happening here is that inference variables are created on the fly by typeck whenever a lifetime has no resolution attached to it - while it would be possible to alter the implementation to reuse inference variables based on decisions made early by `resolve_lifetime`, not doing that is more flexible and works better - it can compile all testcases from #38624 by not ending up with `&'static mut (Trait+'static)`.
The ambiguity specifically cannot be an early error, because this is only the "default" (typeck can still pick something better based on the definition of `Trait` and whether it has any lifetime bounds), and having an error at all doesn't help anyone, as we can perfectly infer an appropriate lifetime inside the `fn` body.
**TODO**: write tests for the user-visible changes.
cc @nikomatsakis @arielb1
Use __SIZEOF_INT128__ to test __int128 presence
Previously we tested whether a handful of preprocessor variables indicating certain 64 bit
platforms, but this does not work for other 64 bit targets which have support for __int128 in C
compiler.
Use the `__SIZEOF__INT128__` preprocessor variable instead. This variable gets set to 16 by gcc and
clang for every target where __int128 is supported.
Fix another endianness issue in i128 trans
Apparently LLVMArbitraryPrecisionInteger demands integers to be in low-endian 64-bytes, rather than host-endian 64-bytes. This is weird, and obviously, not documented. And rustc now works a teeny bit more on big endians.
r? @eddyb
Rewrite the first sentence in slice::sort
For every method, the first sentence should consisely explain what it does,
not how. This sentence usually starts with a verb.
It's really weird for `sort` to be explained in terms of another function,
namely `sort_by`. There's no need for that because it's obvious how `sort`
sorts elements: there is `T: Ord`.
If `sort_by_key` does not have to explicitly state how it's implemented,
then `sort` doesn't either.
r? @steveklabnik
Avoid ICE when pretty-printing non-local MIR item.
This comes up when using `-Zunstable-options --unpretty=mir`. Previously, rustc would ICE due to an unwrap later in this function (after `as_local_node_id`). Instead, we should just ignore items from other crates when pretty-printing MIR.
This was reported in #rust: [this playground code](https://is.gd/PSMBZS) causes an ICE if you click the MIR button. The problem is the mention of the non-local item `std::usize::MAX`, so you can reduce the test case [a lot](https://is.gd/SaLjaa).
r? @eddyb
std: Stabilize APIs for the 1.16.0 release
This commit applies the stabilization/deprecations of the 1.16.0 release, as
tracked by the rust-lang/rust issue tracker and the final-comment-period tag.
The following APIs were stabilized:
* `VecDeque::truncate`
* `VecDeque::resize`
* `String::insert_str`
* `Duration::checked_{add,sub,div,mul}`
* `str::replacen`
* `SocketAddr::is_ipv{4,6}`
* `IpAddr::is_ipv{4,6}`
* `str::repeat`
* `Vec::dedup_by`
* `Vec::dedup_by_key`
* `Result::unwrap_or_default`
* `<*const T>::wrapping_offset`
* `<*mut T>::wrapping_offset`
* `CommandExt::creation_flags` (on Windows)
* `File::set_permissions`
* `String::split_off`
The following APIs were deprecated
* `EnumSet` - replaced with other ecosystem abstractions, long since unstable
Closes#27788Closes#35553Closes#35774Closes#36436Closes#36949Closes#37079Closes#37087Closes#37516Closes#37827Closes#37916Closes#37966Closes#38080
travis: Upload all artifacts in build/dist
Previously we only uploaded tarballs, but this modifies Travis/AppVeyor to
upload everything. We shouldn't have anything else in there to worry about and
otherwise we need to be sure to pick up pkg/msi/exe installers.
Hide uninhabitedness checks behind feature gate
This reverts the fix to match exhaustiveness checking so that it can be discussed. The new code is now hidden behind the `never_type` feature gate.
rustbuild: Add manifest generation in-tree
This commit adds a new tool, `build-manifest`, which is used to generate a
distribution manifest of all produced artifacts. This tool is intended to
replace the `build-rust-manifest.py` script that's currently located on the
buildmaster. The intention is that we'll have a builder which periodically:
* Downloads all artifacts for a commit
* Runs `./x.py dist hash-and-sign`. This will generate `sha256` and `asc` files
as well as TOML manifests.
* Upload all generated hashes and manifests to the directory the artifacts came
from.
* Upload *all* artifacts (tarballs and hashes and manifests) to an archived
location.
* If necessary, upload all artifacts to the main location.
This script is intended to just be the second step here where orchestrating
uploads and such will all happen externally from the build system itself.
cc #38531
Detect double reference when applying binary op
``` rust
let vr = v.iter().filter(|x| {
x % 2 == 0
});
```
will now yield the following compiler output:
``` bash
ERROR binary operation `%` cannot be applied to type `&&_`
NOTE this is a reference of a reference to a type that `%` can be applied to,
you need to dereference this variable once for this operation to work
NOTE an implementation of `std::ops::Rem` might be missing for `&&_`
```
The first NOTE is new.
Fix#33877
----
Thanks to @estebank for providing the original PR #34420 (of which this is a tweaked rebase).
rustc: Remove all "consider using an explicit lifetime parameter" suggestions
These give so many incorrect suggestions that having them is
detrimental to the user experience. The compiler should not be
suggesting changes to the code that are wrong - it is infuriating: not
only is the compiler telling you that _you don't understand_ borrowing,
_the compiler itself_ appears to not understand borrowing. It does not
inspire confidence.
r? @nikomatsakis
Previously we tested whether a handful of preprocessor variables indicating certain 64 bit
platforms, but this does not work for other 64 bit targets which have support for __int128 in C
compiler.
Use the __SIZEOF__INT128__ preprocessor variable instead. This variable gets set to 16 by gcc and
clang for every target where __int128 is supported.
travis: Turn off core dumps on OSX
I've seen these take up quite a bit of log space and I have the sneaking
suspicion that they're just making our test suite take longer (sometimes timing
out on 32-bit OSX now). In any case the backtraces haven't proven too useful,
unfortunately.
Hide a few more standard library symbols
These commits touch up some of the symbol visibility rules for some crates related to the standard library, notably:
* Symbols that are `pub extern` and `#[no_mangle]` which are internal-to-rust ABI things are no longer at the `C` export level, but the `Rust` export level. This includes allocators, panic runtimes, and compiler builtins.
* The libbacktrace library is now compiled with `-fvisibility=hidden` to ensure that we don't export those symbols.
We don't want these symbols exported from the standard library, this is
just an internal implementation detail of the standard library
currently.
Closes#34984
This hides symbols from various unstable and implementation-detail
crates of the standard library. Although typically transitive exported
`pub extern` functions are exported from cdylibs, these crates aren't
necessary as they're all implementation details.
Closes#34493
