Implement a feature for a sound specialization subset
This implements a new feature (`min_specialization`) that restricts specialization to a subset that is reasonable for the standard library to use.
The plan is to then:
* Update `libcore` and `liballoc` to compile with `min_specialization`.
* Add a lint to forbid use of `feature(specialization)` (and other unsound, type system extending features) in the standard library.
* Fix the soundness issues around `specialization`.
* Remove `min_specialization`
The rest of this is an overview from a comment in this PR
## Basic approach
To enforce this requirement on specializations we take the following approach:
1. Match up the substs for `impl2` so that the implemented trait and self-type match those for `impl1`.
2. Check for any direct use of `'static` in the substs of `impl2`.
3. Check that all of the generic parameters of `impl1` occur at most once in the *unconstrained* substs for `impl2`. A parameter is constrained if its value is completely determined by an associated type projection predicate.
4. Check that all predicates on `impl1` also exist on `impl2` (after matching substs).
## Example
Suppose we have the following always applicable impl:
```rust
impl<T> SpecExtend<T> for std::vec::IntoIter<T> { /* specialized impl */ }
impl<T, I: Iterator<Item=T>> SpecExtend<T> for I { /* default impl */ }
```
We get that the subst for `impl2` are `[T, std::vec::IntoIter<T>]`. `T` is constrained to be `<I as Iterator>::Item`, so we check only `std::vec::IntoIter<T>` for repeated parameters, which it doesn't have. The predicates of `impl1` are only `T: Sized`, which is also a predicate of impl2`. So this specialization is sound.
## Extensions
Unfortunately not all specializations in the standard library are allowed by this. So there are two extensions to these rules that allow specializing on some traits.
### rustc_specialization_trait
If a trait is always applicable, then it's sound to specialize on it. We check trait is always applicable in the same way as impls, except that step 4 is now "all predicates on `impl1` are always applicable". We require that `specialization` or `min_specialization` is enabled to implement these traits.
### rustc_specialization_marker
There are also some specialization on traits with no methods, including the `FusedIterator` trait which is advertised as allowing optimizations. We allow marking marker traits with an unstable attribute that means we ignore them in point 3 of the checks above. This is unsound but we allow it in the short term because it can't cause use after frees with purely safe code in the same way as specializing on traits methods can.
r? @nikomatsakis
cc #31844#67194
ast: `Mac`/`Macro` -> `MacCall`
It's now obvious that these refer to macro calls rather than to macro definitions.
It's also a single name instead of two different names in different places.
`rustc_expand` usually calls macro calls in a wide sense (including attributes and derives) "macro invocations", but structures and variants renamed in this PR are only relevant to fn-like macros, so it's simpler and clearer to just call them calls.
cc https://github.com/rust-lang/rust/pull/63586#discussion_r314232513
r? @eddyb
Remove spotlight
I had a few comments saying that this feature was at best misunderstood or not even used so I decided to organize a poll about on [twitter](https://twitter.com/imperioworld_/status/1232769353503956994). After 87 votes, the result is very clear: it's not useful. Considering the amount of code we have just to run it, I think it's definitely worth it to remove it.
r? @kinnison
cc @ollie27
parse: fuse associated and extern items up to defaultness
Language changes:
- The grammar of extern `type` aliases is unified with associated ones, and becomes:
```rust
TypeItem = "type" ident generics {":" bounds}? where_clause {"=" type}? ";" ;
```
Semantic restrictions (`ast_validation`) are added to forbid any parameters in `generics`, any bounds in `bounds`, and any predicates in `where_clause`, as well as the presence of a type expression (`= u8`).
(Work still remains to fuse this with free `type` aliases, but this can be done later.)
- The grammar of constants and static items (free, associated, and extern) now permits the absence of an expression, and becomes:
```rust
GlobalItem = {"const" {ident | "_"} | "static" "mut"? ident} {"=" expr}? ";" ;
```
- A semantic restriction is added to enforce the presence of the expression (the body).
- A semantic restriction is added to reject `const _` in associated contexts.
Together, these changes allow us to fuse the grammar of associated items and extern items up to `default`ness which is the main goal of the PR.
-----------------------
We are now very close to fully fusing the entirely of item parsing and their ASTs. To progress further, we must make a decision: should we parse e.g. `default use foo::bar;` and whatnot? Accepting that is likely easiest from a parsing perspective, as it does not require using look-ahead, but it is perhaps not too onerous to only accept it for `fn`s (and all their various qualifiers), `const`s, `static`s, and `type`s.
r? @petrochenkov
Step stage0 to bootstrap from 1.42
This also includes a commit which fixes the rustfmt downloading logic to redownload when the rustfmt channel changes, and bumps rustfmt to a more recent version.
parser: syntactically allow `self` in all `fn` contexts
Part of https://github.com/rust-lang/rust/pull/68728.
`self` parameters are now *syntactically* allowed as the first parameter irrespective of item context (and in function pointers). Instead, semantic validation (`ast_validation`) is used.
r? @petrochenkov
Implement `?const` opt-out for trait bounds
For now, such bounds are treated exactly the same as unprefixed ones in all contexts. [RFC 2632](https://github.com/rust-lang/rfcs/pull/2632) does not specify whether such bounds are forbidden outside of `const` contexts, so they are allowed at the moment.
Prior to this PR, the constness of a trait bound/impl was stored in `TraitRef`. Now, the constness of an `impl` is stored in `ast::ItemKind::Impl` and the constness of a bound in `ast::TraitBoundModifer`. Additionally, constness of trait bounds is now stored in an additional field of `ty::Predicate::Trait`, and the combination of the constness of the item along with any `TraitBoundModifier` determines the constness of the bound in accordance with the RFC. Encoding the constness of impls at the `ty` level is left for a later PR.
After a discussion in \#wg-grammar on Discord, it was decided that the grammar should not encode the mutual exclusivity of trait bound modifiers. The grammar for trait bound modifiers remains `[?const] [?]`. To encode this, I add a dummy variant to `ast::TraitBoundModifier` that is used when the syntax `?const ?` appears. This variant causes an error in AST validation and disappears during HIR lowering.
cc #67794
r? @oli-obk