Evaluate fixed-length array length expressions lazily.
This is in preparation for polymorphic array lengths (aka `[T; T::A]`) and const generics.
We need deferred const-evaluation to break cycles when array types show up in positions which require knowing the array type to typeck the array length, e.g. the array type is in a `where` clause.
The final step - actually passing bounds in scope to array length expressions from the parent - is not done because it still produces cycles when *normalizing* `ParamEnv`s, and @nikomatsakis' in-progress lazy normalization work is needed to deal with that uniformly.
However, the changes here are still useful to unlock work on const generics, which @EpicatSupercell manifested interest in, and I might be mentoring them for that, but we need this baseline first.
r? @nikomatsakis cc @oli-obk
This commit makes librustc_passes::consts::CheckCrateVisitor properly
mark expressions as promotable if they reference a static, as it's
perfectly fine for one static to reference another. It fixes a
regression that prevented a temporary rvalue from referencing a static
if it was itself declared within a static.
Prior to commit https://github.com/rust-lang/rust/commit/b8c05fe90bc,
`region::ScopeTree` would only register a 'terminating scope' for function
bodies. Thus, while rvalues in a static that referenced a static would be marked
unpromotable, the lack of enclosing scope would cause
mem_categorization::MemCategorizationContext::cat_rvalue_node
to compute a 'temporary scope' of `ReStatic`. Since this had the same
effect as explicitly selecting a scope of `ReStatic`
due to the rvalue being marked by CheckCrateVisitor as promotable,
no issue occurred.
However, commit https://github.com/rust-lang/rust/commit/b8c05fe90bc
made ScopeTree unconditionally register a 'terminating scope'
Since mem_categorization would now compute a non-static 'temporary scope', the
aforementioned rvalues would be erroneously marked as living for too
short a time.
By fixing the behavior of CheckCrateVisitor, this commit avoids changing
mem_categorization's behavior, while ensuring that temporary values in
statics are still allowed to reference other statics.
Fixes issue #44373
This commit moves the calculation of the `LanguageItems` structure into a
query rather than being calculated before the `TyCtxt` exists, with the eventual
end goal of removing some `CrateStore` methods.
Forward-compatibly deny drops in constants if they *could* actually run.
This is part of #40036, specifically the checks for user-defined destructor invocations on locals which *may not* have been moved away, the motivating example being:
```rust
const FOO: i32 = (HasDrop {...}, 0).1;
```
The evaluation of constant MIR will continue to create `'static` slots for more locals than is necessary (if `Storage{Live,Dead}` statements are ignored), but it shouldn't be misusable.
r? @nikomatsakis
Use hir::ItemLocalId as keys in TypeckTables.
This PR makes `TypeckTables` use `ItemLocalId` instead of `NodeId` as key. This is needed for incremental compilation -- for stable hashing and for being able to persist and reload these tables. The PR implements the most important part of https://github.com/rust-lang/rust/issues/40303.
Some notes on the implementation:
* The PR adds the `HirId` to HIR nodes where needed (`Expr`, `Local`, `Block`, `Pat`) which obviates the need to store a `NodeId -> HirId` mapping in crate metadata. Thanks @eddyb for the suggestion! In the future the `HirId` should completely replace the `NodeId` in HIR nodes.
* Before something is read or stored in one of the various `TypeckTables` subtables, the entry's key is validated via the new `TypeckTables::validate_hir_id()` method. This makes sure that we are not mixing information from different items in a single table.
That last part could be made a bit nicer by either (a) new-typing the table-key and making `validate_hir_id()` the only way to convert a `HirId` to the new-typed key, or (b) just encapsulate sub-table access a little better. This PR, however, contents itself with not making things significantly worse.
Also, there's quite a bit of switching around between `NodeId`, `HirId`, and `DefIndex`. These conversions are cheap except for `HirId -> NodeId`, so if the valued reviewer finds such an instance in a performance critical place, please let me know.
Ideally we convert more and more code from `NodeId` to `HirId` in the future so that there are no more `NodeId`s after HIR lowering anywhere. Then the amount of switching should be minimal again.
r? @eddyb, maybe?
ast_validation: forbid "nonstandard" literal patterns
Since #42886, macros can create "nonstandard" PatKind::Lit patterns,
that contain path expressions instead of the usual literal expr. These
can cause trouble, including ICEs.
We *could* map these nonstandard patterns to PatKind::Path patterns
during HIR lowering, but that would be much effort for little gain, and
I think is too risky for beta. So let's just forbid them during AST
validation.
Fixes#43250.
beta-nominating because regression.
r? @eddyb
Since #42886, macros can create "nonstandard" PatKind::Lit patterns,
that contain path expressions instead of the usual literal expr. These
can cause trouble, including ICEs.
We *could* map these nonstandard patterns to PatKind::Path patterns
during HIR lowering, but that would be much effort for little gain, and
I think is too risky for beta. So let's just forbid them during AST
validation.
Fixes#43250.
rustc: Rearchitect lints to be emitted more eagerly
In preparation for incremental compilation this commit refactors the lint
handling infrastructure in the compiler to be more "eager" and overall more
incremental-friendly. Many passes of the compiler can emit lints at various
points but before this commit all lints were buffered in a table to be emitted
at the very end of compilation. This commit changes these lints to be emitted
immediately during compilation using pre-calculated lint level-related data
structures.
Linting today is split into two phases, one set of "early" lints run on the
`syntax::ast` and a "late" set of lints run on the HIR. This commit moves the
"early" lints to running as late as possible in compilation, just before HIR
lowering. This notably means that we're catching resolve-related lints just
before HIR lowering. The early linting remains a pass very similar to how it was
before, maintaining context of the current lint level as it walks the tree.
Post-HIR, however, linting is structured as a method on the `TyCtxt` which
transitively executes a query to calculate lint levels. Each request to lint on
a `TyCtxt` will query the entire crate's 'lint level data structure' and then go
from there about whether the lint should be emitted or not.
The query depends on the entire HIR crate but should be very quick to calculate
(just a quick walk of the HIR) and the red-green system should notice that the
lint level data structure rarely changes, and should hopefully preserve
incrementality.
Overall this resulted in a pretty big change to the test suite now that lints
are emitted much earlier in compilation (on-demand vs only at the end). This in
turn necessitated the addition of many `#![allow(warnings)]` directives
throughout the compile-fail test suite and a number of updates to the UI test
suite.
Closes https://github.com/rust-lang/rust/issues/42511
In preparation for incremental compilation this commit refactors the lint
handling infrastructure in the compiler to be more "eager" and overall more
incremental-friendly. Many passes of the compiler can emit lints at various
points but before this commit all lints were buffered in a table to be emitted
at the very end of compilation. This commit changes these lints to be emitted
immediately during compilation using pre-calculated lint level-related data
structures.
Linting today is split into two phases, one set of "early" lints run on the
`syntax::ast` and a "late" set of lints run on the HIR. This commit moves the
"early" lints to running as late as possible in compilation, just before HIR
lowering. This notably means that we're catching resolve-related lints just
before HIR lowering. The early linting remains a pass very similar to how it was
before, maintaining context of the current lint level as it walks the tree.
Post-HIR, however, linting is structured as a method on the `TyCtxt` which
transitively executes a query to calculate lint levels. Each request to lint on
a `TyCtxt` will query the entire crate's 'lint level data structure' and then go
from there about whether the lint should be emitted or not.
The query depends on the entire HIR crate but should be very quick to calculate
(just a quick walk of the HIR) and the red-green system should notice that the
lint level data structure rarely changes, and should hopefully preserve
incrementality.
Overall this resulted in a pretty big change to the test suite now that lints
are emitted much earlier in compilation (on-demand vs only at the end). This in
turn necessitated the addition of many `#![allow(warnings)]` directives
throughout the compile-fail test suite and a number of updates to the UI test
suite.
After repatriating error explanations to the global registry, some lurking
doctest failures surfaced and needed to be chased down. Sadly, a few doctests
needed to be ignored due to a not-yet-understood regression in the doctest
`compile_fail` functionality (filed #43707).
Provide positional information when visiting ty, substs and closure_substs in MIR
This will enable the region renumbering portion of #43234 (non-lexical lifetimes). @nikomatsakis's current plan [here](https://gist.github.com/nikomatsakis/dfc27b28cd024eb25054b52bb11082f2) shows that we need spans of the original code to create new region variables, e.g. `self.infcx.next_region_var(infer::MiscVariable(span))`. The current visitor impls did not pass positional information (`Location` in some, `Span` and `SourceInfo` for others) for all types. I did not expand this to all visits, just the ones necessary for the above-mentioned plan.
