Create a new TyKind::DynStar variant

[eholk]

Previously dyn* was implemented by adding a DynKind field on the ty::Dynamic variant. This works well for cases where we can treat the two types mostly the same, which is most of the front parts of the compiler (type checker, etc.). It doesn't work so well for the backends like Miri, since the representation of a dyn and dyn* are quite different. This PR splits the dyn* type into its own variant to make it more explicit how it is handled.

r? compiler-errors

[rustbot]

Some changes occurred to the CTFE / Miri engine

cc @rust-lang/miri

Some changes occurred to the core trait solver

cc @rust-lang/initiative-trait-system-refactor

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

[eholk]

This is almost certainly going to break in CI... I fixed what I could test locally, but unfortunately I haven't found a great way to test things like cranelift on my own machine.

[rustbot]

Some changes occurred in compiler/rustc_codegen_cranelift

cc @bjorn3

[rustbot]

Some changes occurred in src/tools/clippy

cc @rust-lang/clippy

[rust-log-analyzer]

The job mingw-check failed! Check out the build log: (web) (plain)

Click to see the possible cause of the failure (guessed by this bot)

    Checking cargo_metadata v0.15.3
    Checking rustfix v0.6.1
    Checking clippy_lints v0.1.69 (/checkout/src/tools/clippy/clippy_lints)
    Checking rustc-workspace-hack v1.0.0 (/checkout/src/tools/rustc-workspace-hack)
error[E0004]: non-exhaustive patterns: `rustc_type_ir::sty::TyKind::DynStar(_, _)` not covered
     |
     |
1387 |         break match *ty.kind() {
     |                     ^^^^^^^^^^ pattern `rustc_type_ir::sty::TyKind::DynStar(_, _)` not covered
     |
note: `rustc_type_ir::sty::TyKind<TyCtxt<'_>>` defined here
    --> /checkout/compiler/rustc_type_ir/src/sty.rs:132:5
     |
50   | pub enum TyKind<I: Interner> {
...
...
132  |     DynStar(I::ListBinderExistentialPredicate, I::Region),
     |     ^^^^^^^ not covered
     = note: the matched value is of type `rustc_type_ir::sty::TyKind<TyCtxt<'_>>`
help: ensure that all possible cases are being handled by adding a match arm with a wildcard pattern or an explicit pattern as shown
1428 ~             }
1428 ~             }
1429 ~             rustc_type_ir::sty::TyKind::DynStar(_, _) => todo!(),

For more information about this error, try `rustc --explain E0004`.
error: could not compile `clippy_lints` due to previous error
Build completed unsuccessfully in 0:02:50

[RalfJung]

This will also cause tons of conflicts with #107728 :/

[compiler-errors]

Needs fixing, and probably should wait until Ralf's PR is landed.

@rustbot author

[bors]

☔ The latest upstream changes (presumably #108052) made this pull request unmergeable. Please resolve the merge conflicts.

[JohnCSimon]

@eholk
ping from triage - can you post your status on this PR?

This will also cause tons of conflicts with #107728 :/

this has merged

FYI: when a PR is ready for review, send a message containing
@rustbot ready to switch to S-waiting-on-review so the PR is in the reviewer's backlog.

[flip1995]

This file looks as if it was just re-formatted. Please don't commit formatting changes to Clippy in the Rust repo.

[compiler-errors]

Gonna close this since it's pretty stale, and probably just needs to be done from scratch.

[RalfJung]

it's pretty stale, and probably just needs to be done from scratch.

Where is that tracked?

[compiler-errors]

@RalfJung, what do you mean?

[RalfJung]

Do we have an issue tracking this cleanup and the reasons for it? Chances are it will just get forgotten otherwise. I guess the tracking issue for dynstar would be the right place?

[compiler-errors]

Ah, no, we currently do not have one. I'll leave a comment on the dyn* tracking issue.