members of ::std::u128 module must not be stable

[est31]

Seems that #38482, while correctly making use ::std::u128; unstable, accidentally made the members of that module stable. This must not happen, and it may not leak to stable/beta. E.g. this compiles : println!("{}", ::std::u128::MAX);.

cc #35118 (tracking issue)
cc @nagisa

[nagisa]

Adding #![unstable(...)] on top of these modules should be sufficient, right? Why is that not the case then?

There’s even a

#[unstable(feature = "i128", issue = "35118")]
#[cfg(not(stage0))]
pub use core::i128;

in libstd/lib.rs…

[est31]

Adding #![unstable(...)] on top of these modules should be sufficient, right?

seems this only makes use ::std::u128 fail compilation, but not the example above.

[est31]

Edited the issue description to contain the new findings.

[ollie27]

Looks like int_module! and uint_module! need to take a stability attribute as an argument.

[est31]

See #38861 . Don't tell bors anything about it yet, I haven't run the test yet. EDIT: I've run it, it works. Ready for review.