Chalk: panicked at 'not implemented'

[cynecx]

use std::thread::spawn;
fn it_works() {
    let x = spawn(|x| {});
}

The thread closure doesn't take any arguments, so this is code that doesn't type check, however rust-analyzer/chalk shouldn't panic in this case.

thread '<unnamed>' panicked at 'not implemented', /home/cynecx/dev/chalk/chalk-solve/src/clauses.rs:60:40
stack backtrace:
   0: std::panicking::begin_panic
   1: chalk_solve::clauses::push_auto_trait_impls
   2: chalk_solve::clauses::builder::ClauseBuilder<I>::push_binders
   3: chalk_solve::clauses::program_clauses_for_goal
   4: chalk_recursive::recursive::Solver<I>::solve_new_subgoal
   5: <chalk_recursive::recursive::Solver<I> as chalk_recursive::solve::SolveDatabase<I>>::solve_goal
   6: <chalk_recursive::recursive::RecursiveSolver<I> as chalk_solve::solve::Solver<I>>::solve_limited
   7: hir_ty::traits::trait_solve_query
   8: salsa::runtime::Runtime::execute_query_implementation
   9: salsa::derived::slot::Slot<Q,MP>::read_upgrade
  10: salsa::derived::slot::Slot<Q,MP>::read
  11: <salsa::derived::DerivedStorage<Q,MP> as salsa::plumbing::QueryStorageOps<Q>>::try_fetch
  12: <DB as hir_ty::db::HirDatabase>::trait_solve::__shim
  13: <DB as hir_ty::db::HirDatabase>::trait_solve
  14: hir_ty::infer::InferenceContext::resolve_ty_as_possible
  15: hir_ty::infer::expr::<impl hir_ty::infer::InferenceContext>::infer_expr_inner
  16: hir_ty::infer::expr::<impl hir_ty::infer::InferenceContext>::infer_expr_coerce
  17: hir_ty::infer::expr::<impl hir_ty::infer::InferenceContext>::check_call_arguments
  18: hir_ty::infer::expr::<impl hir_ty::infer::InferenceContext>::infer_expr_inner
  19: hir_ty::infer::expr::<impl hir_ty::infer::InferenceContext>::infer_expr_coerce
  20: hir_ty::infer::expr::<impl hir_ty::infer::InferenceContext>::infer_expr_inner
  21: hir_ty::infer::expr::<impl hir_ty::infer::InferenceContext>::infer_expr_coerce
  22: hir_ty::infer::infer_query
  23: salsa::runtime::Runtime::execute_query_implementation
  24: salsa::derived::slot::Slot<Q,MP>::read_upgrade
  25: salsa::derived::slot::Slot<Q,MP>::read
  26: <salsa::derived::DerivedStorage<Q,MP> as salsa::plumbing::QueryStorageOps<Q>>::try_fetch
  27: <DB as hir_ty::db::HirDatabase>::infer_query::__shim
  28: hir_ty::db::infer_wait
  29: hir_ty::diagnostics::validate_body
  30: hir::code_model::Module::diagnostics
  31: ide::diagnostics::diagnostics
  32: std::panicking::try
  33: ide::Analysis::diagnostics
  34: rust_analyzer::handlers::publish_diagnostics
  35: core::ops::function::impls::<impl core::ops::function::FnMut<A> for &mut F>::call_mut
  36: <alloc::vec::Vec<T> as alloc::vec::SpecFromIter<T,I>>::from_iter
  37: <F as threadpool::FnBox>::call_box

rust-analyzer: e5f252a

@flodiebold?

[kjeremy]

Weirdly I can't seem to get this to panic in a test.

[cynecx]

@kjeremy I am also using rust nightly and I can reproduce this quite deterministically (I am just pasting that snipped in an empty vscode text editor with rust-analyzer being enabled).

[kjeremy]

To clarify: I can reproduce this in vscode. but when I put it in as a #[test] it doesn't panic.

[cynecx]

I can also reproduce this in a #[test] (also in vscode):

Screen recording

[kjeremy]

Right. I should have specified... I was expecting this to panic:

    #[test]
    fn blah() {
        check_diagnostics(
            r#"
            use std::thread::spawn;
            fn it_works() {
                let x = spawn(|x| {});
            }
            "#,
        );
    }

Anyway this is definitely a valid issue.

[bjorn3]

Tests in rust-analyzer don't load the sysroot by default to make them faster.

[cynecx]

A more standalone reproduction code:

pub fn spawn<F, T>(f: F) -> !
where
    F: FnOnce() -> T,
    F: Send + 'static,
    T: Send + 'static,
{
    todo!()
}

fn test() {
    let x = spawn(|x| {});
}

[Veetaha]

cc #6120

[cynecx]

cc rust-lang/chalk#621
fix: rust-lang/chalk#622

[Ameobea]

Now that the fix in Chalk has been merged, you can git apply this patch in order to get rust-analyzer to be usable again: https://gist.github.com/Ameobea/ed01812726b2a20f818194da4d1134dc

Hooray, I can write code again