Suggest struct or union to add generic that impls trait

[xizheyin]

Fixes #135759

cc @tdittr

[rustbot]

Failed to set assignee to tdittr: invalid assignee

Note: Only org members with at least the repository "read" role, users with write permissions, or people who have commented on the PR may be assigned.

[rustbot]

HIR ty lowering was modified

cc @fmease

[xizheyin]

By the way, how to add the test? I tried but there were some problems. For example, here is a source file(lib.rs):

trait Trait {}

struct Foo {
    a: Trait,
    b: u32,

}

struct Foo1 {
    a: i32,
    b: Trait,
}

when I run it with cargo +stage1 build, it reports:

error[E0782]: expected a type, found a trait
 --> src/lib.rs:4:8
  |
4 |     a: Trait,
  |        ^^^^^
  |
help: you might be missing a type parameter
  |
3 ~ struct Foo<T: Trait> {
4 ~     a: T,
  |

error[E0782]: expected a type, found a trait
  --> src/lib.rs:11:8
   |
11 |     b: Trait,
   |        ^^^^^
   |
help: you can add the `dyn` keyword if you want a trait object
   |
11 |     b: dyn Trait,
   |        +++

For more information about this error, try `rustc --explain E0782`.

But with rustc +stage1 lib.rs, it reports:

error[E0601]: `main` function not found in crate `lib`
  --> src/lib.rs:12:2
   |
12 | }
   |  ^ consider adding a `main` function to `src/lib.rs`

warning: trait objects without an explicit `dyn` are deprecated
 --> src/lib.rs:4:8
  |
4 |     a: Trait,
  |        ^^^^^
  |
  = warning: this is accepted in the current edition (Rust 2015) but is a hard error in Rust 2021!
  = note: for more information, see <https://doc.rust-lang.org/nightly/edition-guide/rust-2021/warnings-promoted-to-error.html>
  = note: `#[warn(bare_trait_objects)]` on by default
help: if this is a dyn-compatible trait, use `dyn`
  |
4 |     a: dyn Trait,
  |        +++

warning: trait objects without an explicit `dyn` are deprecated
  --> src/lib.rs:11:8
   |
11 |     b: Trait,
   |        ^^^^^
   |
   = warning: this is accepted in the current edition (Rust 2015) but is a hard error in Rust 2021!
   = note: for more information, see <https://doc.rust-lang.org/nightly/edition-guide/rust-2021/warnings-promoted-to-error.html>
help: if this is a dyn-compatible trait, use `dyn`
   |
11 |     b: dyn Trait,
   |        +++

error[E0277]: the size for values of type `(dyn Trait + 'static)` cannot be known at compilation time
 --> src/lib.rs:4:8
  |
4 |     a: Trait,
  |        ^^^^^ doesn't have a size known at compile-time
  |
  = help: the trait `Sized` is not implemented for `(dyn Trait + 'static)`
  = note: only the last field of a struct may have a dynamically sized type
  = help: change the field's type to have a statically known size
help: borrowed types always have a statically known size
  |
4 |     a: &Trait,
  |        +
help: the `Box` type always has a statically known size and allocates its contents in the heap
  |
4 |     a: Box<Trait>,
  |        ++++     +

error: aborting due to 2 previous errors; 2 warnings emitted

Some errors have detailed explanations: E0277, E0601.
For more information about an error, try `rustc --explain E0277`.

[xizheyin]

r? compiler

[nnethercote]

If you run rustc directly it defaults to the 2015 edition. Using cargo you will use whatever edition is specified in Cargo.toml, which will default to a later edition. That explains the difference in output you are seeing for those two commands.

You should add the test as a new UI test under tests/ui/. See https://rustc-dev-guide.rust-lang.org/tests/adding.html for details.

[xizheyin]

@rustbot label -S-waiting-on-author +S-waiting-on-review

[xizheyin]

Thank you very much. I added ui test. @nnethercote

[nnethercote]

Looks good, thanks.

A suggestion for next time: if you add the new test in the first commit, and then make the change in the second commit, the reviewer can see how the test output changes in the second commit. This makes it easier to review.

@bors r+

[bors]

📌 Commit 7e199b1 has been approved by nnethercote

It is now in the queue for this repository.

[xizheyin]

You mean that submitting the UI test in the first commit and then failing it in the PR will tell you the output of the old code, right? Makes sense, thanks!