Xpetra: issue with MultiVector::getLocalView in uvm_off build

[jewatkins]

Bug Report

@trilinos/xpetra

Description

The uvm_off build is broken in PR #9107 because of the use of Xpetra::MultiVector::getLocalView(). See: https://testing.sandia.gov/cdash/viewBuildError.php?buildid=9186703
There's some logic there that I don't think works for cuda space. Is someone planning to work on this soon? I imagine the fix would be similar to what's done in Tpetra.

I'm able to get passed this issue by specifically calling getDeviceLocalView() but it's probably better to fix getLocalView().

Steps to Reproduce

Build the uvm_off build with https://github.com/jewatkins/Trilinos/tree/jewatkins/muelu-semicoarsen-kokkos

[jhux2]

@jewatkins Yes, we are planning to rewrite parts of Xpetra to use Tpetra's new interfaces. If you can work around this, that might be better for the moment, but leaving this issue open as a reminder that we need to revisit this. Would that be ok with you?

@trilinos/xpetra

[jewatkins]

Yes, I think that's okay for now. Let me see if the work around works for all cases.

[csiefer2]

Not sure why we're building Xpetra/MueLu in that PR build at all.

[github-actions]

This issue has had no activity for 365 days and is marked for closure. It will be closed after an additional 30 days of inactivity.
If you would like to keep this issue open please add a comment and/or remove the MARKED_FOR_CLOSURE label.
If this issue should be kept open even with no activity beyond the time limits you can add the label DO_NOT_AUTOCLOSE.
If it is ok for this issue to be closed, feel free to go ahead and close it. Please do not add any comments or change any labels or otherwise touch this issue unless your intention is to reset the inactivity counter for an additional year.

[jewatkins]

Xpetra multivector has been updated and I don't currently see any issues with what's currently being used so I think we can close this.