Reattach all grandchildren when constructing specialization graph.
Specialization graphs are constructed by incrementally adding impls in the order of declaration. If the impl being added has its specializations in the graph already, they should be reattached under the impl. However, the current implementation only reattaches the one found first. Therefore, in the following specialization graph,
```
Tr1
|
I3
/ \
I1 I2
```
If `I1`, `I2`, and `I3` are declared in this order, the compiler mistakenly constructs the following graph:
```
Tr1
/ \
I3 I2
|
I1
```
This patch fixes the reattach procedure to include all specializing grandchildren-to-be.
Fixes#50452.
Implement by-value object safety
This PR implements **by-value object safety**, which is part of unsized rvalues #48055. That means, with `#![feature(unsized_locals)]`, you can call a method `fn foo(self, ...)` on trait objects. One aim of this is to enable `Box<FnOnce>` in the near future.
The difficulty here is this: when constructing a vtable for a trait `Foo`, we can't just put the function `<T as Foo>::foo` into the table. If `T` is no larger than `usize`, `self` is usually passed directly. However, as the caller of the vtable doesn't know the concrete `Self` type, we want a variant of `<T as Foo>::foo` where `self` is always passed by reference.
Therefore, when the compiler encounters such a method to be generated as a vtable entry, it produces a newly introduced instance called `InstanceDef::VtableShim(def_id)` (that wraps the original instance). the shim just derefs the receiver and calls the original method. We give different symbol names for the shims by appending `::{{vtable-shim}}` to the symbol path (and also adding vtable-shimness as an ingredient to the symbol hash).
r? @eddyb
resolve: Support custom attributes when macro modularization is enabled
Basically, if resolution of a single-segment attribute is a determined error, then we interpret it as a custom attribute.
Since custom attributes are integrated into general macro resolution, `feature(custom_attribute)` now requires and implicitly enables macro modularization (`feature(use_extern_macros)`).
Actually, a few other "advanced" macro features now implicitly enable macro modularization too (and one bug was found and fixed in process of enabling it).
The first two commits are preliminary cleanups/refactorings.
Add errors for unknown, stable and duplicate feature attributes
- Adds an error for unknown (lang and lib) features.
- Extends the lint for unnecessary feature attributes for stable features to libs features (this already exists for lang features).
- Adds an error for duplicate (lang and lib) features.
```rust
#![feature(fake_feature)] //~ ERROR unknown feature `fake_feature`
#![feature(i128_type)] //~ WARNING the feature `i128_type` has been stable since 1.26.0
#![feature(non_exhaustive)]
#![feature(non_exhaustive)] //~ ERROR duplicate `non_exhaustive` feature attribute
```
Fixes#52053, fixes#53032 and address some of the problems noted in #44232 (though not unused features).
There are a few outstanding problems, that I haven't narrowed down yet:
- [x] Stability attributes on macros do not seem to be taken into account.
- [x] Stability attributes behind `cfg` attributes are not taken into account.
- [x] There are failing incremental tests.
dead-code lint: say "constructed" for structs
Respectively.
This is a sequel to November 2017's #46103 / 1a9dc2e9. It had been
reported (more than once—at least #19140, #44083, and #44565) that the
"never used" language was confusing for enum variants that were "used"
as match patterns, so the wording was changed to say never "constructed"
specifically for enum variants. More recently, the same issue was raised
for structs (#52325). It seems consistent to say "constructed" here,
too, for the same reasons.
~~While we're here, we can also use more specific word "called" for unused
functions and methods. (We declined to do this in #46103, but the
rationale given in the commit message doesn't actually make sense.)~~
This resolves#52325.
Reintroduce `Undef` and properly check constant value sizes
r? @RalfJung
cc @eddyb
basically all kinds of silent failures that never occurred are assertions now
resolve: Implement prelude search for macro paths, implement tool attributes
When identifier is macro path is resolved in scopes (i.e. the first path segment - `foo` in `foo::mac!()` or `foo!()`), scopes are searched in the same order as for non-macro paths - items in modules, extern prelude, tool prelude (see later), standard library prelude, language prelude, but with some extra shadowing restrictions (names from globs and macro expansions cannot shadow names from outer scopes). See the comment in `fn resolve_lexical_macro_path_segment` for more details.
"Tool prelude" currently contains two "tool modules" `rustfmt` and `clippy`, and is searched immediately after extern prelude.
This makes the [possible long-term solution](https://github.com/rust-lang/rfcs/blob/master/text/2103-tool-attributes.md#long-term-solution) for tool attributes exactly equivalent to the existing extern prelude scheme, except that `--extern=my_crate` making crate names available in scope is replaced with something like `--tool=my_tool` making tool names available in scope.
The `tool_attributes` feature is still unstable and `#![feature(tool_attributes)]` now implicitly enables `#![feature(use_extern_macros)]`. `use_extern_macros` is a prerequisite for `tool_attributes`, so their stabilization will happen in the same order.
If `use_extern_macros` is not enabled, then tool attributes are treated as custom attributes (this is temporary, anyway).
Fixes https://github.com/rust-lang/rust/issues/52576
Fixes https://github.com/rust-lang/rust/issues/52512
Fixes https://github.com/rust-lang/rust/issues/51277
cc https://github.com/rust-lang/rust/issues/52269
rustc_resolve: record single-segment extern crate import resolutions.
Fixes#52489 by recording special-cased single-segment imports for later (e.g. stability) checks.
cc @alexcrichton @Mark-Simulacrum @petrochenkov
Does this need to be backported?
Previously linker diagnostic were being hidden when two modules were linked
together but failed to link. This commit fixes the situation by ensuring that we
have a diagnostic handler installed and also adds support for handling linker
diagnostics.
