Add rmsByLevel function

[mo-joshuacolclough]

Description

Adds rmsByLevel function in Increments.h. Required by OOPS PR JCSDA-internal/oops/pull/1893. Note: It seems this does not affect any of the test YAMLs.

Issue(s) addressed

• Closes Add rmsByLevel method to Increment #15

Acceptance Criteria (Definition of Done)

• All tests pass. bb output: http://fcm1/cylc-review/taskjobs/jcolclou/?suite=add_rms_by_level_orca.

SPICE intel builds are failing but I don't think it is due to these changes.

Dependencies

• waiting on JCSDA-internal/oops/pull/1901 (Old: JCSDA-internal/oops/pull/1893)

[twsearle]

I am curious what are the merits of using oops abort as opposed to throwing an exception? Elsewhere I have been throwing exceptions, but perhaps this is contrary to the jedi style?

[mo-joshuacolclough]

I am curious what are the merits of using oops abort as opposed to throwing an exception? Elsewhere I have been throwing exceptions, but perhaps this is contrary to the jedi style?

Hmm, I don't know the details but perhaps @tf17270 can answer more fully? By looking at the code for the ABORT function (JCSDA-internal/oops/blob/develop/src/oops/util/abor1_cpp.cc), it looks like it's ensuring that any other MPI processes also abort with oops::mpi::world().abort(EXIT_FAILURE);.

[tf17270]

I am curious what are the merits of using oops abort as opposed to throwing an exception? Elsewhere I have been throwing exceptions, but perhaps this is contrary to the jedi style?

Hmm, I don't know the details but perhaps @tf17270 can answer more fully? By looking at the code for the ABORT function (JCSDA-internal/oops/blob/develop/src/oops/util/abor1_cpp.cc), it looks like it's ensuring that any other MPI processes also abort with oops::mpi::world().abort(EXIT_FAILURE);.

Let me preface this by saying its just by looking and extrapolating from the fact I was asked to use abort. I have seen within oops that abort is the preference for where functionality doesn't exist but and except (often wrapped up in oops::EXPECT statements) are used in cases where there are yaml validation failures or other errors associated with values not being what they were meant to be

[twsearle]

Thanks very much @tf17270 and @mo-joshuacolclough that makes sense! I will add that little bit of refactoring to my list!