Pack u128 in the compiler to mitigate new alignment
This is based on #116672, adding a new `#[repr(packed(8))]` wrapper on `u128` to avoid changing any of the compiler's size assertions. This is needed in two places:
* `SwitchTargets`, otherwise its `SmallVec<[u128; 1]>` gets padded up to 32 bytes.
* `LitKind::Int`, so that entire `enum` can stay 24 bytes.
* This change definitely has far-reaching effects though, since it's public.
Make `HirEqInterExpr::eq_block` take comments into account while checking if two blocks are equal
This PR:
- now makes `HirEqInterExpr::eq_block` take comments into account. Identical code with varying comments will no longer be considered equal.
- makes necessary adjustments to UI tests.
Closes#12044
**Lintcheck Changes**
- `match_same_arms` 53 => 52
- `if_same_then_else` 3 => 0
changelog: [`if_same_then_else`]: Blocks with different comments will no longer trigger this lint.
changelog: [`match_same_arms`]: Arms with different comments will no longer trigger this lint.
```
In #119606 I added them and used a `_mv` suffix, but that wasn't great.
A `with_` prefix has three different existing uses.
- Constructors, e.g. `Vec::with_capacity`.
- Wrappers that provide an environment to execute some code, e.g.
`with_session_globals`.
- Consuming chaining methods, e.g. `Span::with_{lo,hi,ctxt}`.
The third case is exactly what we want, so this commit changes
`DiagnosticBuilder::foo_mv` to `DiagnosticBuilder::with_foo`.
Thanks to @compiler-errors for the suggestion.
This works for most of its call sites. This is nice, because `emit` very
much makes sense as a consuming operation -- indeed,
`DiagnosticBuilderState` exists to ensure no diagnostic is emitted
twice, but it uses runtime checks.
For the small number of call sites where a consuming emit doesn't work,
the commit adds `DiagnosticBuilder::emit_without_consuming`. (This will
be removed in subsequent commits.)
Likewise, `emit_unless` becomes consuming. And `delay_as_bug` becomes
consuming, while `delay_as_bug_without_consuming` is added (which will
also be removed in subsequent commits.)
All this requires significant changes to `DiagnosticBuilder`'s chaining
methods. Currently `DiagnosticBuilder` method chaining uses a
non-consuming `&mut self -> &mut Self` style, which allows chaining to
be used when the chain ends in `emit()`, like so:
```
struct_err(msg).span(span).emit();
```
But it doesn't work when producing a `DiagnosticBuilder` value,
requiring this:
```
let mut err = self.struct_err(msg);
err.span(span);
err
```
This style of chaining won't work with consuming `emit` though. For
that, we need to use to a `self -> Self` style. That also would allow
`DiagnosticBuilder` production to be chained, e.g.:
```
self.struct_err(msg).span(span)
```
However, removing the `&mut self -> &mut Self` style would require that
individual modifications of a `DiagnosticBuilder` go from this:
```
err.span(span);
```
to this:
```
err = err.span(span);
```
There are *many* such places. I have a high tolerance for tedious
refactorings, but even I gave up after a long time trying to convert
them all.
Instead, this commit has it both ways: the existing `&mut self -> Self`
chaining methods are kept, and new `self -> Self` chaining methods are
added, all of which have a `_mv` suffix (short for "move"). Changes to
the existing `forward!` macro lets this happen with very little
additional boilerplate code. I chose to add the suffix to the new
chaining methods rather than the existing ones, because the number of
changes required is much smaller that way.
This doubled chainging is a bit clumsy, but I think it is worthwhile
because it allows a *lot* of good things to subsequently happen. In this
commit, there are many `mut` qualifiers removed in places where
diagnostics are emitted without being modified. In subsequent commits:
- chaining can be used more, making the code more concise;
- more use of chaining also permits the removal of redundant diagnostic
APIs like `struct_err_with_code`, which can be replaced easily with
`struct_err` + `code_mv`;
- `emit_without_diagnostic` can be removed, which simplifies a lot of
machinery, removing the need for `DiagnosticBuilderState`.
Extend `map_clone` lint to also work on non-explicit closures
I found it weird that this case was not handled by the current line so I added it. The only thing is that I don't see an obvious way to infer the current type to determine if it's copyable or not, so for now I always suggest `cloned` and I added a FIXME.
r? `@llogiq`
changelog: Extend `map_clone` lint to also work on non-explicit closures
This commit:
- now makes `HirEqInterExpr::eq_block` take comments into account. Identical code with varying comments will no longer be considered equal.
- makes necessary adjustments to UI tests.
Add support for `for await` loops
This adds support for `for await` loops. This includes parsing, desugaring in AST->HIR lowering, and adding some support functions to the library.
Given a loop like:
```rust
for await i in iter {
...
}
```
this is desugared to something like:
```rust
let mut iter = iter.into_async_iter();
while let Some(i) = loop {
match core::pin::Pin::new(&mut iter).poll_next(cx) {
Poll::Ready(i) => break i,
Poll::Pending => yield,
}
} {
...
}
```
This PR also adds a basic `IntoAsyncIterator` trait. This is partly for symmetry with the way `Iterator` and `IntoIterator` work. The other reason is that for async iterators it's helpful to have a place apart from the data structure being iterated over to store state. `IntoAsyncIterator` gives us a good place to do this.
I've gated this feature behind `async_for_loop` and opened #118898 as the feature tracking issue.
r? `@compiler-errors`
don't visit nested bodies in `is_const_evaluatable`
Fixes#11939
This ICE happened in `if_let_some_else_none`, but the root problem is in one of the utils that it uses.
It is (was) possible for `is_const_evalutable` to visit nested bodies which would lead to it trying to get the type of one of the expressions with the wrong typeck table, which won't have the type stored.
Notably, for the expression `Bytes::from_static(&[0; 256 * 1024]);` in the linked issue, the array length is an anonymous const in which type checking happens on its own, so we can't use the typeck table of the enclosing function in there.
Visiting nested bodies is also not needed for checking whether an expression can be const, so I think it's safe to ignore just ignore them altogether.
changelog: Fix ICE when checking for constness in nested bodies
Renamings:
- find -> opt_hir_node
- get -> hir_node
- find_by_def_id -> opt_hir_node_by_def_id
- get_by_def_id -> hir_node_by_def_id
Fix rebase changes using removed methods
Use `tcx.hir_node_by_def_id()` whenever possible in compiler
Fix clippy errors
Fix compiler
Apply suggestions from code review
Co-authored-by: Vadim Petrochenkov <vadim.petrochenkov@gmail.com>
Add FIXME for `tcx.hir()` returned type about its removal
Simplify with with `tcx.hir_node_by_def_id`
Introduce support for `async gen` blocks
I'm delighted to demonstrate that `async gen` block are not very difficult to support. They're simply coroutines that yield `Poll<Option<T>>` and return `()`.
**This PR is WIP and in draft mode for now** -- I'm mostly putting it up to show folks that it's possible. This PR needs a lang-team experiment associated with it or possible an RFC, since I don't think it falls under the jurisdiction of the `gen` RFC that was recently authored by oli (https://github.com/rust-lang/rfcs/pull/3513, https://github.com/rust-lang/rust/issues/117078).
### Technical note on the pre-generator-transform yield type:
The reason that the underlying coroutines yield `Poll<Option<T>>` and not `Poll<T>` (which would make more sense, IMO, for the pre-transformed coroutine), is because the `TransformVisitor` that is used to turn coroutines into built-in state machine functions would have to destructure and reconstruct the latter into the former, which requires at least inserting a new basic block (for a `switchInt` terminator, to match on the `Poll` discriminant).
This does mean that the desugaring (at the `rustc_ast_lowering` level) of `async gen` blocks is a bit more involved. However, since we already need to intercept both `.await` and `yield` operators, I don't consider it much of a technical burden.
r? `@ghost`
never_patterns: Parse match arms with no body
Never patterns are meant to signal unreachable cases, and thus don't take bodies:
```rust
let ptr: *const Option<!> = ...;
match *ptr {
None => { foo(); }
Some(!),
}
```
This PR makes rustc accept the above, and enforces that an arm has a body xor is a never pattern. This affects parsing of match arms even with the feature off, so this is delicate. (Plus this is my first non-trivial change to the parser).
~~The last commit is optional; it introduces a bit of churn to allow the new suggestions to be machine-applicable. There may be a better solution? I'm not sure.~~ EDIT: I removed that commit
r? `@compiler-errors`
Add `never_patterns` feature gate
This PR adds the feature gate and most basic parsing for the experimental `never_patterns` feature. See the tracking issue (https://github.com/rust-lang/rust/issues/118155) for details on the experiment.
`@scottmcm` has agreed to be my lang-team liaison for this experiment.
- Rename them both `as_str`, which is the typical name for a function
that returns a `&str`. (`to_string` is appropriate for functions
returning `String` or maybe `Cow<'a, str>`.)
- Change `UnOp::as_str` from an associated function (weird!) to a
method.
- Avoid needless `self` dereferences.
teach `eager_or_lazy` about panicky arithmetic operations
Fixes#9422Fixes#9814Fixes#11793
It's a bit sad that we have to do this because arithmetic operations seemed to me like the prime example where a closure would not be necessary, but this has "side effects" (changes behavior when going from lazy to eager) as some of these panic on overflow/underflow if compiled with `-Coverflow-checks` (which is the default in debug mode).
Given the number of backlinks in the mentioned issues, this seems to be a FP that is worth fixing, probably.
changelog: [`unnecessary_lazy_evaluations`]: don't lint if closure has panicky arithmetic operations
Implement new lint `iter_over_hash_type`
Implements and fixes https://github.com/rust-lang/rust-clippy/issues/11788
This PR adds a new *restriction* lint `iter_over_hash_type` which prevents `Hash`-types (that is, `HashSet` and `HashMap`) from being used as the iterator in `for` loops.
The justification for this is because in `Hash`-based types, the ordering of items is not guaranteed and may vary between executions of the same program on the same hardware. In addition, it reduces readability due to the unclear iteration order.
The implementation of this lint also ensures the following:
- Calls to `HashMap::keys`, `HashMap::values`, and `HashSet::iter` are also denied when used in `for` loops,
- When this expression is used in procedural macros, it is not linted/denied.
changelog: add new `iter_over_hash_type` lint to prevent unordered iterations through hashed data structures
Don't check for late-bound vars, check for escaping bound vars
Fixes an assertion that didn't make sense. Many valid and well-formed types *have* late-bound vars (e.g. `for<'a> fn(&'a ())`), they just must not have *escaping* late-bound vars in order to be normalized correctly.
Addresses rust-lang/rust-clippy#11230, cc `@jyn514` and `@matthiaskrgr`
changelog: don't check for late-bound vars, check for escaping bound vars. Addresses rust-lang/rust-clippy#11230
Fixes to `manual_let_else`'s divergence check
A few changes to the divergence check in `manual_let_else` and moves it the implementation to `clippy_utils` since it's generally useful:
* Handle internal `break` and `continue` expressions.
e.g. The first loop is divergent, but the second is not.
```rust
{
loop {
break 'outer;
};
}
{
loop {
break;
};
}
```
* Match rust's definition of divergence which is defined via the type system.
e.g. The following is not considered divergent by rustc as the inner block has a result type of `()`:
```rust
{
'a: {
panic!();
break 'a;
};
}
```
* Handle when adding a single semicolon would make the expression divergent.
e.g. The following would be a divergent if a semicolon were added after the `if` expression:
```rust
{ if panic!() { 0 } else { 1 } }
```
changelog: None
Lint `needless_borrow` and `explicit_auto_deref` on most union field accesses
Changes both lints to follow rustc's rules around auto-deref through `ManuallyDrop` union fields rather than just bailing on union fields.
changelog: [`needless_borrow`] & [`explicit_auto_deref`]: Lint on most union field accesses
[`map_identity`]: respect match ergonomics
Fixes#11764
Note: the original tests before this were slightly wrong themselves already and had to be changed. They were calling `map` on an iterator of `&(i32, i32)`s, so this PR would stop linting there, but they were meant to test something else unrelated to binding modes, so I just changed them to remove the references so that it iterates over owned values and they all bind by value. This way they continue to test what they checked for before: the identity function for tuple patterns.
changelog: [`map_identity`]: respect match ergonomics
Make SpanlessEq more consistent
1) Remove wildcard as requested in https://github.com/rust-lang/rust-clippy/issues/10267.
2) Implement `hir_utils::eq_expr` for `ExprKind::Closure`, `ExprKind::ConstBlock`, `ExprKind::InlineAsm` and `ExprKind::Yield`.
3) Reorder branches of `hir_utils::eq_expr` to be in alphabetical order.
---
changelog: none
Most notably, this commit changes the `pub use crate::*;` in that file
to `use crate::*;`. This requires a lot of `use` items in other crates
to be adjusted, because everything defined within `rustc_span::*` was
also available via `rustc_span::source_map::*`, which is bizarre.
The commit also removes `SourceMap::span_to_relative_line_string`, which
is unused.
