Mikael Simberg
Mikael Simberg
Newer CMake versions actually seem to error out on the wrongly spelled `-no-tests=error` (possibly all nonexistent flags; previously CMake silently ignores flags it doesn't know about). I'll try to prioritize...
Diff coverage failing is ok. The only uncovered lines are MPI continuations-related and we can't test those (easily) at the momeent in CI.
This and https://github.com/pika-org/pika/pull/1140 have issues building the HIP stack with the newer spack commit. Further investigation needed.
We still need https://github.com/NVIDIA/stdexec/pull/1398 to get things going with the newer commits of stdexec.
The nvhpc pipeline isn't happy now. This needs more time...
MPI continuations are not yet tested. All other modes are now tested since #1151.
Thanks for fixing the process mask regex! This looks good on the CI configuration side. However, the test failures remain. Have you had a chance to look into those?
We'd need to revive this quite soon as daint is going down. @aurianer do you think you will have time to continue this or should I have a look?
I've cherry-picked the thread binding regex into https://github.com/pika-org/pika/pull/1240.
> If I'm not mistaken, this is now covered by #1239 no? Yep, you're right. I suppose the addition in this PR is that it builds without CUDA on santis/todi/daint....