rustc: Allow `#[no_mangle]` anywhere in a crate
This commit updates the compiler to allow the `#[no_mangle]` (and
`#[export_name]` attributes) to be located anywhere within a crate.
These attributes are unconditionally processed, causing the compiler to
always generate an exported symbol with the appropriate name.
After some discussion on #54135 it was found that not a great reason
this hasn't been allowed already, and it seems to match the behavior
that many expect! Previously the compiler would only export a
`#[no_mangle]` symbol if it were *publicly reachable*, meaning that it
itself is `pub` and it's otherwise publicly reachable from the root of
the crate. This new definition is that `#[no_mangle]` *is always
reachable*, no matter where it is in a crate or whether it has `pub` or
not.
This should make it much easier to declare an exported symbol with a
known and unique name, even when it's an internal implementation detail
of the crate itself. Note that these symbols will persist beyond LTO as
well, always making their way to the linker.
Along the way this commit removes the `private_no_mangle_functions` lint
(also for statics) as there's no longer any need to lint these
situations. Furthermore a good number of tests were updated now that
symbol visibility has been changed.
Closes#54135
NLL: temps in block tail expression diagnostic
This change adds a diagnostic that explains when temporaries in a block tail expression live longer than block local variables that they borrow, and attempts to suggest turning the tail expresion into a statement (either by adding a semicolon at the end, when its result value is clearly unused, or by introducing a `let`-binding for the result value and then returning that).
Fix#54556
This commit updates the compiler to allow the `#[no_mangle]` (and
`#[export_name]` attributes) to be located anywhere within a crate.
These attributes are unconditionally processed, causing the compiler to
always generate an exported symbol with the appropriate name.
After some discussion on #54135 it was found that not a great reason
this hasn't been allowed already, and it seems to match the behavior
that many expect! Previously the compiler would only export a
`#[no_mangle]` symbol if it were *publicly reachable*, meaning that it
itself is `pub` and it's otherwise publicly reachable from the root of
the crate. This new definition is that `#[no_mangle]` *is always
reachable*, no matter where it is in a crate or whether it has `pub` or
not.
This should make it much easier to declare an exported symbol with a
known and unique name, even when it's an internal implementation detail
of the crate itself. Note that these symbols will persist beyond LTO as
well, always making their way to the linker.
Along the way this commit removes the `private_no_mangle_functions` lint
(also for statics) as there's no longer any need to lint these
situations. Furthermore a good number of tests were updated now that
symbol visibility has been changed.
Closes#54135
wasm: Explicitly export all symbols with LLD
This commit fixes an oddity on the wasm target where LTO can produce
working executables but plain old optimizations doesn't. The compiler
already knows what set of symbols it would like to export, but LLD only
discovers this list transitively through symbol visibilities. LLD may
not, however, always find all the symbols that we'd like to export.
For example if you depend on an rlib with a `#[no_mangle]` symbol, then
if you don't actually use anything from the rlib then the symbol won't
appear in the final artifact! It will appear, however, with LTO. This
commit attempts to rectify this situation by ensuring that all symbols
rustc would otherwise preserve through LTO are also preserved through
the linking process with LLD by default.
Remove unneccessary error from test, revealing NLL error.
Part of #52663.
Removes unnecessary type mismatch error from test that was hiding
borrow check error from NLL stderr.
r? @nikomatsakis
make `Parser::parse_foreign_item()` return a foreign item or error
Fixes `Parser::parse_foreign_item()` to follow the convention of `parse_trait_item()` and `parse_impl_item()` in that it *must* parse an item or return an error, and then the caller is responsible for detecting the closing delimiter.
This prevents it from looping endlessly on an unexpected token in `ext/expand.rs` where it was also leaking memory by continually pushing to `Parser::expected_tokens` via `Parser::check_keyword()`.
closes#54441
r? @petrochenkov
cc @dtolnay
resolve: Some refactorings in preparation for uniform paths 2.0
The main result is that in-scope resolution performed during macro expansion / import resolution is now consolidated in a single function (`fn early_resolve_ident_in_lexical_scope`), which can now be used for resolving first import segments as well when uniform paths are enabled.
r? @ghost
Merge `proc_macro_` expansion feature gates as `proc_macro_hygiene`
Merges `proc_macro_mod`, `proc_macro_expr`, `proc_macro_non_items`, and `proc_macro_gen` into a single feature: `proc_macro_hygiene`. These features are not all blocked on implementing macro hygiene *per se*, but rather on interactions with hygiene that have not been entirely resolved.
The restrictions were introduced in https://github.com/rust-lang/rust/pull/54277 and no longer necessary now because legacy plugins are now expanded in usual left-to-right order
adopt "placeholders" to represent universally quantified regions
This does a few preliminary refactorings that lay some groundwork for moving towards universe integration. Two things, primarily:
- Rename from "skolemized" to "placeholder"
- When instantiating `for<'a, 'b, 'c>`, just create one universe for all 3 regions, and distinguish them from one another using the `BoundRegion`.
- This is more accurate, and I think that in general we'll be moving towards a model of separating "binder" (universe, debruijn index) from "index within binder" in a number of places.
- In principle, it feels the current setup of making lots of universes could lead to us doing the wrong thing, but I've actually not been able to come up with an example where this is so.
r? @scalexm
cc @arielb1
[NLL] Improve "borrow later used here" messages
* In the case of two conflicting borrows, the later used message says which borrow it's referring to
* If the later use is a function call (from the users point of view) say that the later use is for the call. Point just to the function.
r? @pnkfelix
Closes#48643
Remove duplicate predicates in `explicit_predicates_of`
I took a more brutal approach than described in #52187. I could have used the `linked_hash_map` crate but this seems overkill, especially as we need a vec storage in the end.
r? @nikomatsakis
abolish ICE when pretty-printing async block
@jnetterf reported an ICE when the unused-parentheses lint triggered around an async block (#54752). In order to compose an autofixable suggestion, the lint invokes the pretty-printer on the unnecessarily-parenthesized expression. (One wonders why the lint doesn't just use `SourceMap::span_to_snippet` instead, to preserve the formatting of the original source?—but to answer that, you'd have to ask the author of 5c9f806d.)
But then the pretty-printer panics when trying to call `<pprust::State as PrintState>::end` when `State.boxes` is empty. Empirically, the problem would seem to be solved if we start some "boxes" beforehand in the `ast::ExprKind::Async` arm of the big match in `print_expr_outer_attr_style`, exactly like we do in the immediately-preceding match arm for `ast::ExprKind::Block`—it would seem pretty ("pretty") reasonable for the pretty-printing of async blocks to work a lot like the pretty-printing of ordinary non-async blocks, right??
Of course, it would be shamefully cargo-culty to commit code on the basis of this kind of mere reasoning-by-analogy (in contrast to understanding the design of the pretty-printer in such detail that the correctness of the patch is comprehended with all the lucid certainty of mathematical proof, rather than being merely surmised by intuition). But maybe we care more about fixing the bug with high probability today, than with certainty in some indefinite hypothetical future? Maybe the effort is worth [a fifth of a shirt](https://hacktoberfest.digitalocean.com/stats/zackmdavis)??
Humbly resolves#54752.
r? @cramertj
Limit the promotion of const fns to the libstd and the `rustc_promotable` attribute
There are so many questions around promoting const fn calls... it seems saner to try to limit automatic promotion to const fns which were explicitly opted in for promotion.
I added the attribute to all public stable const fns that were already promotable (e.g. not Cell::new) in order to not cause any breakage
r? @eddyb
cc @nikomatsakis
handle outlives predicates in trait evaluation
This handles higher-ranked outlives predicates in trait evaluation the same way they are handled in projection.
Fixes#54302. I think this is a more correct fix than #54401 because it fixes the root case in evaluation instead of making evaluation used in less cases. However, we might want to go to a direction closer to @nikomatsakis's solution with Chalk.
r? @nikomatsakis
suggest `crate::...` for "local" paths in 2018
Fixes#54230.
This commit adds suggestions for unresolved imports in the cases where
there could be a missing `crate::`, `super::`, `self::` or a missing
external crate name before an import.
r? @nikomatsakis
Give a special message when the later use is from a call. Use the span
of the callee instead of the whole expression. For conflicting borrow
messages say that the later use is of the first borrow.
resolve: Disambiguate a subset of conflicts "macro_rules" vs "macro name in module"
Currently if macro name may refer to both a `macro_rules` macro definition and a macro defined/imported into module we conservatively report an ambiguity error.
Unfortunately, these errors became a source of regressions when macro modularization was enabled - see issue https://github.com/rust-lang/rust/issues/54472.
This PR disambiguates such conflicts in favor of `macro_rules` if both the `macro_rules` item and in-module macro name are defined in the same normal (named) module and `macro_rules` is closer in scope to the point of use (see the tests for examples).
This is a subset of more general approach described in https://github.com/rust-lang/rust/issues/54472#issuecomment-424666659.
The subset is enough to fix all the regressions from https://github.com/rust-lang/rust/issues/54472, but it can be extended to apply to all "macro_rules" vs "macro name in module" conflicts in the future.
To give an analogy, this is equivalent to scoping rules for `let` variables and items defined in blocks (`macro_rules` behaves like "`let` at module level" in general).
```rust
{ // beginning of the block
use xxx::m; // (1)
// Starting from the beginning of the block and until here m!() refers to (1)
macro_rules! m { ... } // (2)
// Starting from here and until the end of the block m!() refers to (2)
} // end of the block
```
More complex examples with `use` and `macro_rules` from different modules still report ambiguity errors, even if equivalent examples with `let` are legal.
Fixes https://github.com/rust-lang/rust/issues/54472 (stable-to-beta regression)
This commit adds suggestions for unresolved imports in the cases where
there could be a missing `crate::`, `super::`, `self::` or a missing
external crate name before an import.
NLL fails to suggest "try removing `&mut` here"
Fixes#51191.
This PR adds ``try removing `&mut` here`` suggestions to functions where a mutable borrow is being taken of a `&mut self` or a `self: &mut Self`. This PR also enables the suggestion for adding a `mut` pattern to by-value implicit self arguments without `mut` patterns already.
r? @nikomatsakis
