No need to write a const_assert RFC (at most it would be syntactic sugar)

[rodrimati1992]

The future section in Multiple of 8 const assertion talks about creating a const_assert rfc.

The already accepted rfc for assertions in constants is this one,
which combined with const_wildcard would allow doing const assertions outside of any item.

Example of how it will look once both RFCS are fully implemented:

const _:()=assert!(SIZE % 8 == 0, "total size is required to be a multiple of 8 bits");

[dtolnay]

Thanks for bringing this up. From skimming those two RFCs, I don't get the impression that const _: () = assert!(...) would be a sufficient solution. I think it would hit the behavior described under First attempt of consts not being evaluated during cargo check.

[rodrimati1992]

Doing cargo check with this code:

#![feature(underscore_const_names)]
macro_rules! require_multiple_of_eight {
    ($e:expr) => {
        const _: () = [()][$e % 8];
    };
}

require_multiple_of_eight!{10}

Causes this error message (in nightly Rust):

error: any use of this value will cause an error
  --> src/lib.rs:4:23
   |
4  |           const _: () = [()][$e % 8];
   |           --------------^^^^^^^^^^^^-
   |                         |
   |                         index out of bounds: the len is 1 but the index is 2
...
8  | / require_multiple_of_eight!{
9  | |     10
10 | | }
   | |_- in this macro invocation
   |
   = note: #[deny(const_err)] on by default

Are you saying that it might stop producing that error at some point?

Also,const_assert is one of the motivations for the const_wildcard RFC.

This is what panicking in constants looks like in nightly(just doing cargo check):

Code:

#![feature(underscore_const_names)]
#![feature(const_panic)]

const _: () = panic!("Hello world at compile-time!");

Error

error: any use of this value will cause an error
 --> src/main.rs:4:15
  |
4 | const _: () = panic!("Hello world at compile-time!");
  | --------------^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^-
  |               |
  |               the evaluated program panicked at 'Hello world at compile-time!', src/main.rs:4:15
  |

[dtolnay]

Please see #6. Let's reopen if the compiler implementation goes in the right direction to make const _ usable for this purpose.