Eliminate remaining uses of Cray pointers

[tclune]

We (NASA GMAO) would like to build GEOS using the NAG Fortran compiler to perform various aggressive compile-time and runtime checks provided by that compiler. Unfortunately, NAG does not support Cray pointers, so a workaround is needed for FMS.

Now that there is a firm integration path, I will happy to attempt the necessary changes and then issue a pull-request.

[underwoo]

Thank you. We look forward to your pull request. However, please note that internally at GFDL we are in the process of a major rewrite of the fms and mpp sections of the FMS code.

[tclune]

Understood. We ended up with less time to work on this than anticipated. So other than drafting a rough process on how we would do the conversion, no work has actually happened. We'll now wait until the major rewrite is completed.

Note: one of our stumbling blocks was that we have no good way to test our changes to FMS. Our FV3 probably only exercises a small subset of the functionality and would be a poor diagnostic in any event. Do you have a testing methodology for FMS? Or do you just verify that it works with the GFDL model(s)?

[underwoo]

Currently, we run several tests manually using several GFDL model configurations. This also only tests portions of the FMS code, but enough for us to have confidence it will work for the GFDL models. We are working on a automated process that will take use of one (or more) of the cloud providers (e.g. Google Compute, Amazon's AWS). Once complete, we will release the documentation that others can use to set up their own build/test instances. However, the government shutdown and a system update have delayed our progress.

[tclune]

Great. Looking forward to the improved testing.

[edhartnett]

I have just put up a PR with an autotools based build system for FMS.

This supports testing very easily.

I am looking at the existing test code now to see if it can be turned into automated tests.