Deduplicate `IntTy`/`UintTy`/`FloatTy`.
There are identical definitions in `rustc_type_ir` and `rustc_ast`. This commit removes them and places a single definition in `rustc_ast_ir`. This requires adding `rust_span` as a dependency of `rustc_ast_ir`, but means a bunch of silly conversion functions can be removed.
r? `@fmease`
uniquify root goals during HIR typeck
We need to rely on region identity to deal with hangs such as https://github.com/rust-lang/trait-system-refactor-initiative/issues/210 and to keep the current behavior of `fn try_merge_responses`.
This is a problem as borrowck starts by replacing each *occurrence* of a region with a unique inference variable. This frequently splits a single region during HIR typeck into multiple distinct regions. As we assume goals to always succeed during borrowck, relying on two occurances of a region being identical during HIR typeck causes ICE. See the now fixed examples in https://github.com/rust-lang/trait-system-refactor-initiative/issues/27 and rust-lang/rust#139409.
We've previously tried to avoid this issue by always *uniquifying* regions when canonicalizing goals. This prevents caching subtrees during canonicalization which resulted in hangs for very large types. People rely on such types in practice, which caused us to revert our attempt to reinstate `#[type_length_limit]` in https://github.com/rust-lang/rust/pull/127670. The complete list of changes here:
- rust-lang/rust#107981
- rust-lang/rust#110180
- rust-lang/rust#114117
- rust-lang/rust#130821
After more consideration, all occurrences of such large types need to happen outside of typeck/borrowck. We know this as we already walk over all types in the MIR body when replacing their regions with nll vars.
This PR therefore enables us to rely on region identity inside of the trait solver by exclusively **uniquifying root goals during HIR typeck**. These are the only goals we assume to hold during borrowck. This is insufficient as type inference variables may "hide" regions we later uniquify. Because of this, we now stash proven goals which depend on inference variables in HIR typeck and reprove them after writeback. This closes https://github.com/rust-lang/trait-system-refactor-initiative/issues/127.
This was originally part of rust-lang/rust#144258 but I've moved it into a separate PR. While I believe we need to rely on region identity to fix the performance issues in some way, I don't know whether rust-lang/rust#144258 is the best approach to actually do so. Regardless of how we deal with the hangs however, this change is necessary and desirable regardless.
r? `@compiler-errors` or `@BoxyUwU`
Skip walking into param-env component if it has no placeholder/re-var
Although it only provides a minor perf improvement, it seems like it could matter in more pathological cases.
Normalize before computing ConstArgHasType goal in new solver
This is a fix for rust-lang/rust#139905. See the description I left in the test.
I chose to fix this by normalizing the type before matching on its `.kind()` in `compute_const_arg_has_type_goal` (since it feels somewhat consistent with how we normalize types before assembling their candidates, for example); however, there are several other solutions that come to mind for fixing this ICE:
1. (this solution)
2. Giving `ConstKind::Error` a proper type, like `ConstKind::Value`, so that consts don't go from failing to passing `ConstArgHasType` goals after normalization (i.e. `UNEVALUATED` would normalize into a `ConstKind::Error(_, bool)` type rather than losing its type altogether).
3. Just suppressing the errors and accepting the fact that goals can go from fail->pass after normalization.
Thoughts? Happy to discuss this fix further.
r? `@BoxyUwU`
New const traits syntax
This PR only affects the AST and doesn't actually change anything semantically.
All occurrences of `~const` outside of libcore have been replaced by `[const]`. Within libcore we have to wait for rustfmt to be bumped in the bootstrap compiler. This will happen "automatically" (when rustfmt is run) during the bootstrap bump, as rustfmt converts `~const` into `[const]`. After this we can remove the `~const` support from the parser
Caveat discovered during impl: there is no legacy bare trait object recovery for `[const] Trait` as that snippet in type position goes down the slice /array parsing code and will error
r? ``@fee1-dead``
cc ``@nikomatsakis`` ``@traviscross`` ``@compiler-errors``
Fix some fixmes that were waiting for let chains
Was inspired by looking at rust-lang/rust#143066 and spotting two fixmes that were missed, so... r? `@compiler-errors` 😅
Yay, let chains!
Add note to `find_const_ty_from_env`
Add a note to `find_const_ty_from_env` to explain why it has an `unwrap` which "often" causes ICEs.
Also, uplift it into the new trait solver. This avoids needing to go through the interner to call this method which is otherwise an inherent method in the compiler. I can remove this part if desired.
r? `@boxyuwu`
improve search graph docs, reset `encountered_overflow` between reruns
I think this shouldn't really matter for now. It will be more relevant for my current rework as we otherwise cannot partially reevaluate the root goal in case there has been overflow during the prervious iteration.
r? ````@BoxyUwU````
This commit adds a lint to prevent the use of rustc_type_ir in random
compiler crates, except for type system internals traits, which are
explicitly allowed. Moreover, this fixes diagnostic_items() to include
the CRATE_OWNER_ID, otherwise rustc_diagnostic_item attribute is ignored
on the crate root.
As a performance optimization, skip elaborating the supertraits of
`Sized`, and if a `MetaSized` obligation is being checked, then look for
a `Sized` predicate in the parameter environment. This makes the
`ParamEnv` smaller which should improve compiler performance as it avoids
all the iteration over the larger `ParamEnv`.
Expand the automatic implementation of `MetaSized` and `PointeeSized` so
that it is also implemented on non-`Sized` types, just not `ty::Foreign`
(extern type).
Introduce the `MetaSized` and `PointeeSized` traits as supertraits of
`Sized` and initially implement it on everything that currently
implements `Sized` to isolate any changes that simply adding the
traits introduces.
Tweak fast path trait handling
(1.) Make it more sound by considering polarity (lol)
(2.) Make it more general, by considering higher-ranked size/copy/clone
(2.) Make it less observable, by only doing copy/clone fast path if there are no regions involved
r? lcnr
