[PATCH] D25249: [libc++] Many any test fixes

[Eric Fiselier via cfe-commits]

EricWF added a comment.

Addressed the issues about `in_place` SFINAE constraints in an inline comment.



================
Comment at: test/libcxx/utilities/any/any.class/any.assign/value.pass.cpp:25
+// Test that any& operator=(ValueType&&) is *never* selected for:
+// * std::in_place type.
+    {
----------------
EricWF wrote:
> CaseyCarter wrote:
> > This behavior is not required by N4606, so I moved this test to the libcxx tree.
> I thought this behavior was subject to a LWG PR, but I can't seem to find one.
> 
> I'll either create a LWG issue to standardize this behavior or remove the tests and fix libc++.
OK, so the `ValueType` ctor was constrained by http://cplusplus.github.io/LWG/lwg-active.html#2744 but that didn't touch the assignment operator. I think that's a mistake, and I'll look into adjusting the PR for 2744.


https://reviews.llvm.org/D25249