David Wells

I don't think so. CMake prefers that we set up MPI like other dependencies (e.g., provide a root directory for it) and we have some additional code in place to...

It might also help to use the diagonal (unconstrained) mass matrix as a preconditioner.

I checked as a part of #1049 that we do get the same results with the current implementation as we do with `QTRAP`.

Its hard to know without more testing. Do we have any simple test code for this class lying around? I don't see it in any of the examples.

Looking at this again I think this may have been fixed by #1029, but without an explicit test case I don't know. @alexkaiser have you noticed this again over the...

@boyceg @amneetb @abarret @cpuelz This issue was tagged with 0.7 but didn't happen. How do you suggest we proceed?

AFAIK there hasn't been any progress on this - its not critical and not likely to get done this week so lets remove the milestone. @amneetb any news on this?...

I would very much prefer this to the current interface. @amneetb is this something we can get done in time for the next release?

> do you mean implement everything in one class and select the form as an input option? Yes. > I would like to have different classes doing this to make...

It still isn't clear to me how we can register new data outside the variable database: even `SAMRAIVectorReal::cloneVector()` uses the variable database. We might want to ultimately write our own...