Thanks for the report! However, there is no bug here. This is expected. const { ... } is an expression, and you can't write expressions outside of functions.

If you want to just declare a constant outside a function for the purpose of checking an assertion, you can declare an anonymous const item as follows:

const _: () = assert!(size_of::<u32>() <= size_of::<usize>());

The diagnostic should say that, then.

@rustbot label: -C-discussion +C-bug +A-diagnostics +D-papercut +D-terse

It would be nice if this was allowed if the const block evaluates to (), like in this example. The _: () = syntax feels more like a workaround for missing functionality than something that's supposed to still be required now that we have const blocks.

while a nice wishlist idea, given that we don't even have a finished story for const {} in patterns, where we already know we want to be able to write it (as normally we can put a NAME there, and notionally using const NAME: T = {}; then NAME should "always" be possible to replace by const {}), I don't think it's in anyone's interest to special-case item generation here yet. maybe later.

note that it also would be somewhat unusual for us to start accepting any expression in item context, as normally Rust is quite the hardass about misplacing expressions. for instance, not only is const {} supposed to be equal to this, in an expression context:

{
    const NAME: T = expr;
    NAME
}

but this is rejected in item context, because blocks are expressions, not items:

{
}

and part of the reason rustc is persnickety about this sort of thing is that it allows us to be really certain what is supposed to go where, and from there easily infer what was likely intended, and then issue diagnostics.