sammccall added a comment.
Thanks, and sorry for sitting on this so long.
Addressed comments. I think the failing windows bots were implicit
`-fms-extensions` or so that sometimes makes diagnosis more lazy. Moving the
test to recovery-expr-type.cpp should take care of this as that test case
specifies `-triple`. But waiting for the windows bot to finish before landing.
================
Comment at: clang/test/Sema/enum.cpp:8-9
+// When initializers contain errors, enumerators are non-type-dependent zeros.
+static_assert(Circular_A != 0, ""); // expected-error {{static_assert failed}}
+static_assert(Circular_B != 0, ""); // expected-error {{static_assert failed}}
----------------
aaron.ballman wrote:
> Should we be static asserting this? It seems like emergent behavior more than
> something we intentionally want anyone to rely on.
Yeah, the thing we're testing is that we can const-evaluate these, and it
neither crashes nor propagates dependence further. The static_assert checks the
latter.
Replaced with a static assert that isn't sensitive tot he actual value.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D108451/new/
https://reviews.llvm.org/D108451
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
