You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Compiler panic playing with "Err" and lifetimes #67072
Locally I can't reproduce the ICE with rustc built from HEAD with just rustc program.rs. Maybe I'm not passing the right parameters? (not sure what parameters play.rust-lang.org is passing)
@hellow554 is this output from play.rust-lang.org or locally? How do you run rustc exactly to get this? I'm still not getting any ICEs locally, I wonder how you do it.
Interesting, I get an ICE on another tree (I wonder what the difference is ...). The exact ICE message is different than the one in play.rust-lang, but the panic is in the same module (rustc::ty::subst).
Activity
osa1 commentedon Dec 6, 2019
Code copied from play link to make this easier to read:
Locally I can't reproduce the ICE with rustc built from HEAD with just
rustc program.rs. Maybe I'm not passing the right parameters? (not sure what parameters play.rust-lang.org is passing)hellow554 commentedon Dec 6, 2019
minified:
Backtrace:
osa1 commentedon Dec 6, 2019
@hellow554 is this output from play.rust-lang.org or locally? How do you run rustc exactly to get this? I'm still not getting any ICEs locally, I wonder how you do it.
hellow554 commentedon Dec 6, 2019
@osa1 locally by using either
cargoorrustcdirectly.osa1 commentedon Dec 6, 2019
Interesting, I get an ICE on another tree (I wonder what the difference is ...). The exact ICE message is different than the one in play.rust-lang, but the panic is in the same module (
rustc::ty::subst).hellow554 commentedon Dec 6, 2019
This is a regression from 1.38.0 to 1.39.0 and was introduced in ef906d0
Centril commentedon Dec 6, 2019
This should have been fixed by #67044.
pnkfelix commentedon Dec 12, 2019
triage: P-high, removing nomination
hellow554 commentedon Dec 12, 2019
Has already been fixed on f6840f3.
So I guess like Centril already said, fixed by #67044.
So this can be closed?
estebank commentedon Dec 14, 2019
Nightly output:
Closing as fixed by that PR.