Loss of trait bound

[stphnt]

I tried this code (playground):

pub trait Max {
    const MAX: usize;
}

pub struct Something();

impl Max for Something {
    const MAX: usize = 10;
}

struct Data<M: Max> {
    storage: [u8; <M as Max>::MAX],
    // alternatively you could define `storage` as the following and a similar error will result.
    // storage: [u8; M::MAX],
    _marker: std::marker::PhantomData<M>,
}

fn main() {
    println!("{}", std::mem::size_of::<Data<Something>>());
}

I expected to see this happen: it compile and the size of the type to be 10.

Instead, this happened: it did not compile and it reported the following confusing/wrong error message:

error[E0277]: the trait bound `M: Max` is not satisfied
  --> src/main.rs:12:19
   |
2  |     const MAX: usize;
   |     ----------------- required by `Max::MAX`
...
12 |     storage: [u8; <M as Max>::MAX],
   |                   ^^^^^^^^^^^^^^^ the trait `Max` is not implemented for `M`
   |
help: consider further restricting this bound
   |
11 | struct Data<M: Max + Max> {
   |                    ^^^^^

error: aborting due to previous error

For more information about this error, try `rustc --explain E0277`.

It asks for the trait bound that is currently present to be added (and adding it produces the same error again).

Meta

I checked this on the playground and it happens on:

• stable (1.46.0)
• beta (1.47.0-beta.5 2020-09-29 9f0e6fa...)
• nightly (1.48.0-nightly 2020-09-28 fc2daaa...)

stable and nightly also produced the same error on my personal computer.

rustc --version --verbose:

rustc 1.46.0 (04488afe3 2020-08-24)
binary: rustc
commit-hash: 04488afe34512aa4c33566eb16d8c912a3ae04f9
commit-date: 2020-08-24
host: x86_64-unknown-linux-gnu
release: 1.46.0
LLVM version: 10.0

[jonas-schievink]

Thanks for the report! This looks like a duplicate of #43408, however, so closing in favor of that.

Please leave a comment on that issue if you have more information to contribute.

[stphnt]

Thanks for the quick response! I suspected there was already an issue for it, but wasn't sure where to look for it.