Stream: t-compiler

Topic: #58610


csmoe (Mar 05 2019 at 11:52, on Zulip):

@pnkfelix I narrowed the ice into a smaller case last week, it seems an issue related to const-eval & closure (cc @oli )

oli (Mar 05 2019 at 12:02, on Zulip):

accidentally promoting things will cause very random ICEs that are hard to make sense of. I think your example causes the crash due to the a.b field access having messed up local variable ids and using the type from the closue instead of the struct

csmoe (Mar 05 2019 at 12:08, on Zulip):

@oli yep, removing either the closure or the place expr, this ice will disappear.

oli (Mar 05 2019 at 12:10, on Zulip):

also changing the closure to be a u32, will cause the ICE to mention a field access on a u32 :D

oli (Mar 05 2019 at 12:10, on Zulip):

ok, let's just wait until the promoted fix is through

Last update: Nov 16 2019 at 01:05UTC