It looks like you are only installing package ectrans?

@climbfuji Good question. I've asked, and @fmahebert told me majority of the work around this compiler is focused on ectrans. I am all for just adding it to the list and leaving the remaining packages there.

I've started a dialog with @fmahebert and @l90lpa to get some more clarification.

I pulled too much out for testing! I've added the original packages back into the spec and added ectrans to it.

This is more of a question for @fmahebert on requirements; Should we consider adding in jedi-fv3-env, soca-env and ewok-env? It would be nice to be able to run skylab small experiments with this, although (1) I don't know if that's outside the scope of the needs here and (2) maybe that would be better added in a followup change?

Thoughts? At minimum it would be nice to get ewok-env +ecflow ~cylc in this change since we would immediately have ecflow and could run some simple experiments (probably?)

@eap I (selfishly) view this PR's goal to be fixing issues in the current nvhpc spack-stack environment that are blocking our short-term code demonstration goals. Thus, I would recommend against the scope expansion of adding any new packages to the build as part of this PR.

To expand a bit — the demonstrations we're targeting will not involve skylab, workflows, or any of the soca/fv3 related code. Thus, adding the packages you've suggested is not immediately necessary, and risks introducing delays if we discover anything that doesn't build cleanly with nvhpc. In the longer term, it would certainly be nice to have more support for nvhpc across the system by working towards what you suggest. But I view that as a long-term project of building support (as and when warranted by JCSDA/partner needs), and beyond the scope of this PR.

Does that sound reasonable?