On Mon, Jul 9, 2018 at 1:52 PM Zachary Turner wrote:
> makeArrayRef() isn't necessary, but when I was first looking at this I had
> to stare at the code for a bit to see that there was an implicit conversion
> happening. So I put the makeArrayRef() just for the benefit of the person
> reading th
makeArrayRef() isn't necessary, but when I was first looking at this I had
to stare at the code for a bit to see that there was an implicit conversion
happening. So I put the makeArrayRef() just for the benefit of the person
reading the code. But no, it's not necessary.
This did not fail on exis
Did this fail on an existing regression test, or is there a need for more
test coverage? (guessing it failed on existing tests)
Also, is the makeArrayRef necessary? Looks like if the original code
compiled (implicitly converting from vector to ArrayRef) then the new code
wouldn't need a makeArrayR
Author: zturner
Date: Tue Jul 3 11:12:39 2018
New Revision: 336219
URL: http://llvm.org/viewvc/llvm-project?rev=336219&view=rev
Log:
Fix crash in clang.
This happened during a recent refactor. toStringRefArray() returns
a vector, which was being implicitly converted to an
ArrayRef, and then the
