Klaus

@wuxf99 I think this can be closed?

Not sure whether divergence constraint will be required. My understanding is that for FLASH5, face variables will not be used for representing magnetic field components in an MHD solver.

Now that both this Issue and the other one that referred to it (#424) have been closed, I would just like to note (for FLASH) that we may still be...

It seems this was primarily about interpolation. @chaw0023 gave details in September 2019, see above (in Issue #327), I did not see any followup. I don't know details about the...

Let me explain what happens when we have "ghost faces" filled when using PARAMESH, which has shaped our expectations. @chaw0023 can jump in if his requirements are different or more...

No connection specific to face variables, as far as I am aware. As far as guard cell filling more generally, its relationship to orchestration is... poorly understood by me so...

Ideally this would all happen inside the iterator, hidden from the iterator-using code. The iterator's next() method would ship out the next block if and when there is one available,...

Actually, we have a higher-level iterator (implemented in Fortran), which is called throughout the higher-level code wherever iteration is needed. That one is currently (internally) just looping over levels (when...

What we don't have is an asynchronous iterator. (Except if somehow the underlying AMReX iterator, either `octree` or `mfiter`, can be operating asynchronously in some sense, and our wrapper layer...

I'm not so sure myself what I'm asking... But basically with "asynchronous" I meant that client code of the iterator can already process some blocks, while the iterator itself is...