Bug(s) in Test_ComponentSpecParser

[tclune]

Building with NAG reveals compilation and fixes of those reveal run-time bugs and fixing those reveals failing tests.

1. Integer declared with ESMF_KIND_R4 (Should be I4)
   • Not surprising this works with Intel and GNU which have R4=I4=4. NAG does not.
2. expected_timestring and actual_timestring are declared allocatable but not allocated. ESMF expects allocated strings (unfortunately)
   • I do not know why this ran successfully with Intel & GNU in the CI. Clear runtime error message from NAG.
3. Fixing the strings results in a failing tests

Failure
 in: 
Test_MaxAction_suite.test_max_accumulate_R4
  Location: 
[Test_MaxAction.pf:39]
accumulated_field not equal to expected values
Arrays not conformable failure:
    Expected shape: [4]
      Actual shape: [16]
  
Failure
 in: 
Test_MinAction_suite.test_min_accumulate_R4
  Location: 
[Test_MinAction.pf:39]
accumulated_field not equal to expected values
Arrays not conformable failure:
    Expected shape: [4]
      Actual shape: [16]
  
 FAILURES!!!
Tests run: 219, Failures: 2, Errors: 0
, Disabled: 0

[darianboggs]

This may be related to MPI. The generic3g tests pass when compiled by NAG on bucy. The tests for the accumulator have no "npes" parameter.

[tclune]

@darianboggs - Just want to make sure I understand. You fixed (1) and (2) and then found the tests ran without error on bucy? Weird. Very consistently failing on my laptop with the message above.

[darianboggs]

Yes. Now that I added npes, it produces the error that you saw. I verified in the test I ran yesterday that it did not fail on bucy with NAG.

[darianboggs]

It appears that fixes for (1) and (2) are in `release/MAPL-v3' already. In my bugfix branch, I removed the expected_timestring and actual_timestring. They aren't needed for the test. I used them to make an error message string, but now that the test is working, the message is not needed. I replaced the error message with a simple message.

Nonetheless, I'll leave well enough alone regarding (2), and I won't push up my fix for (1) and (2).

[darianboggs]

Fixed by #3276

[darianboggs]

Closing

[tclune]

I think I still see the issue. I will reopen next week depending on what I see.