Josh Essman

For versions of CMake where FindCUDAToolkit is not available, https://gitlab.kitware.com/cmake/cmake/-/issues/17816 describes how similar functionality can be implemented without relying on `FindCUDA.cmake`.

Does this relate to #269 at all?

Is the conversion you were envisioning different than the `VDofToDof` conversion currently being done? Looking through the origin of this function it looks like that's the only transformation applied to...

Theoretically after the Spack issues blocking https://github.com/LLNL/serac/issues/179 are resolved, Mac users will be using the "official" build process that uses static libraries - though in the interim we could definitely...

In my experience this is most frequently a problem with TPLs for which we track hashes; it would be useful if Spack let us use commit SHAs as versions such...

Some caveats/gotchas/pitfalls that have come up as I've been working through the MCArray -> Array replacement: 1. MCArray functions like `size()`/`capacity()`/etc returned the *number of tuples* and not the *number...

@samuelpmishLLNL Your point on `len` is a good one - I did want to clarify that regardless of what `size()` does that "number of rank n-1 subtensors" is always accessible...

The decision to temporarily punt on removing MCArray from `mint` was made in the interest of not having that task hold up further development. I am not completely sure of...

We should also look into eliminating the requirement that meshes/fields be explicitly constructed as "owned" by the datastore, see https://github.com/LLNL/axom/pull/379#discussion_r551672264

Once reconstruction for parallel meshes is implemented we should look at ensuring coherency between the `mfem::Mesh` object and the contents of the datastore as refinement can cause them to diverge.