Avoid more warnings with clang and C++20 #3719
Conversation
| KOKKOS_INLINE_FUNCTION | ||
| bool operator==(volatile const_value_type& val) const { return *ptr == val; } | ||
| bool operator==(volatile const AtomicDataElement& val) const { |
There was a problem hiding this comment.
While not a comment on this change, the volatile comparisons seem sketchy.
|
@ndellingwood should review. Readt from my side. More reviews/opinions welcome. Related to #3716. |
There was a problem hiding this comment.
lgtm @masterleinad I think intel is now PR tested via the github actions, but let me know if would you like me to do any additional testing
|
I should clarify, these changes were not made to address anything intel-related, but replacing enums with static constexpr variables in the past had caused intel compilation issues, though I don't think I've seen that since intel/15 or intel/16 |
It would probably be good to give it a full test run of the Sandia testers. The alternative to the switching to |
|
I just merge this and we check tomorrow if it broke anything and then we can revert. I think the likelyhood is very low. We did a similar change in some other places and it passed. |
Since all of the changed enums are in internal classes any ODR-usage should also be internal. Hence, I didn't provide extra definitions in namescape scope. Also, we don't have them for the corresponding
HIPvariables.The changes to the comparison operators in
core/src/impl/Kokkos_Atomic_View.hpprely on the existence of implicit conversion operators.