Treating DST constructor as a function aborts compiler

[Stebalien]

Aborts (no backtrace):

struct Test([i32]);
fn main() {
  let x = Test;
}

With:

rustc: /buildslave/rust-buildbot/slave/nightly-dist-rustc-linux/build/src/llvm/lib/IR/Instructions.cpp:995: void llvm::StoreInst::AssertOK(): Assertion `getOperand(0)->getType() == cast<PointerType>(getOperand(1)->getType())->getElementType() && "Ptr must be a pointer to Val type!"' failed.

Forked from: #30240.

[Stebalien]

According to @Aatch,

The last error is the most complex as it's a weird case of struct Test([i32]) being a perfectly valid type, but fn([i32]) -> Test not being a valid function. We probably need to check whether you're forcing the creation of an invalid function and error somewhere earlier. That's where the error comes from, btw, the creation of the fn Test(_0: [i32]) -> Test { ... } function body.

[Aatch]

The minimum reproduction is this:

struct Test([i32]);
fn main() {
  let x = Test;
}

Just to show that it's not got anything to do with taking a reference.

[Stebalien]

@Aatch, updated.

[steveklabnik]

This no longer ICEs!

[eddyb]

@steveklabnik Not so fast:

struct Test([i32]);
fn main() {
  let x: fn(_) -> Test = Test;
}

Reification is now lazy, Test itself is a zero-sized function item.

[nikomatsakis]

cc me

[Stebalien]

This appears to have been fixed.

[eddyb]

This shouldn't have been "fixed", hiding the ICE only happened by accident (#45225 unhides it).

[bjorn3]

error[E0277]: the size for values of type `[i32]` cannot be known at compilation time
--> src/main.rs:3:36
  |
3 | fn main() { let x: fn(_) -> Test = Test; }
  |                                    ^^^^ doesn't have a size known at compile-time
  |
  = help: the trait `std::marker::Sized` is not implemented for `[i32]`
  = note: to learn more, visit <https://doc.rust-lang.org/book/ch19-04-advanced-types.html#dynamically-sized-types-and-the-sized-trait>
  = note: all function arguments must have a statically known size
  = help: unsized locals are gated as an unstable feature

error: aborting due to previous error

Rustc now errors on unsized locals being feature gated.

[workingjubilee]

No longer ICEs (again), but correctly refuses to compile on nightly even with #![feature(unsized_locals)]. Test was removed previously, should the test be brought back, then?

[nikomatsakis]

@workingjubilee seems like yes!