In #245 I set the CI to build with Umpire, RAJA and Camp at their respective develop version. I did that to test CHAI in the release conditions, and because I needed to update those because I had to update BLT (so far using blt@develop).
Because Umpire@develop does not integrate LLNL/Umpire#857, in #245 we still run into "already defined targets" when cuda is involved.
The next step was to test LLNL/Umpire#857, which updates BLT in Umpire.
This PR is NOT to be merged: instead, it aims at showing that LLNL/Umpire#857 is ready.
Now, @adayton1 there are 2 remaining failures here that we need to address before moving forward. See the jobs on lassen with -stdlib=libc++ on one side for errors related to libFMT, with with XL on the other side for a runtime error.
This branch aims at testing LLNL/Umpire#857.
@adayton1 @davidbeckingsale @kab163.
In #245 I set the CI to build with Umpire, RAJA and Camp at their respective develop version. I did that to test CHAI in the release conditions, and because I needed to update those because I had to update BLT (so far using blt@develop).
Because Umpire@develop does not integrate LLNL/Umpire#857, in #245 we still run into "already defined targets" when cuda is involved.
The next step was to test LLNL/Umpire#857, which updates BLT in Umpire.
This PR is NOT to be merged: instead, it aims at showing that LLNL/Umpire#857 is ready.
Now, @adayton1 there are 2 remaining failures here that we need to address before moving forward. See the jobs on lassen with -stdlib=libc++ on one side for errors related to libFMT, with with XL on the other side for a runtime error.
Next steps are, in my opinion: