i128 is being stabilized

[mark-i-m]

cc rust-lang/rust#35118 rust-lang/rust#49101

[est31]

You'd most likely have to do #[cfg_attr(stage0, feature(i128_type))] here, as on stage0 it is still needed.

[mark-i-m]

Hmm... It seemed to work fine when I made this change locally?

[alexcrichton]

Thanks! I think it's ok if this is here, right? Does this fail the build?

[alexcrichton]

(in that I'd prefer to land the rustc change first and then land the change here after that's gone through)

[PlasmaPower]

It's sort of a deadlock. Rustc fails without this, but this fails without rustc.

[alexcrichton]

The "failure" I think is deny(warnings), right? Can that be removed?

[PlasmaPower]

Yeah I think that's the best solution. If we want we can re-add if after the rustc change goes through.

[mark-i-m]

Moving the feature(i128_type) after allow(unused_features) seems to make the compiler build for me locally.

[kennytm]

We could use the old external tool update workflow here:

1. Push this PR into a new, non-master branch on this repository
2. Update the submodule in E0263 needs to be updated to new format rust#35518 to point to that new branch
3. Wait for E0263 needs to be updated to new format rust#35518 to merge and a new nightly is released
4. Merge this PR

[PlasmaPower]

That's what I originally suggested in the rustc PR but just ignoring the warning temporarily seems easier.

[mark-i-m]

Umm... does anyone understand the current Travis failure:

thread 'rustc' panicked at 'slice index starts at 4294967295 but ends at 30759529', libcore/slice/mod.rs:791:5
note: Run with `RUST_BACKTRACE=1` for a backtrace.
error: internal compiler error: unexpected panic
note: the compiler unexpectedly panicked. this is a bug.
note: we would appreciate a bug report: /~https://github.com/rust-lang/rust/blob/master/CONTRIBUTING.md#bug-reports
note: rustc 1.26.0-nightly (adf2135ad 2018-03-17) running on x86_64-apple-darwin

[mark-i-m]

It looks like the current change in the PR might be sufficient if we fix the build failures... alternately, I could do @kennytm's suggestion. Just let me know :)

[alexcrichton]

Ok this strategy seems fine by me yeah

[mark-i-m]

Hmm... this doesn't seem to have worked... I don't understand why.

[PlasmaPower]

Oh this is #![allow(unused_features)], but you need #[allow(stable_features)].