Standard names for multiple instances?

[gold2718]

Similarly to threading where we have standard names for the total number of threads and the current thread number, we require standard names for the total number of physics instances on a task and for the current instance number.

[gold2718]

@climbfuji, do you have names we could use?

[climbfuji]

Posted in NCAR/ccpp-framework#561 (comment):

So far, NEPTUNE uses

ccpp_num_instances
ccpp_ensemble_instance

which should probably be changed to, and I am happy to do that,

number_of_ccpp_instances
ccpp_instance_num

?

[mkavulich]

This is (unfortunately) yet another instance where the current names are inconsistent and need some attention. I would propose number_of_ccpp_instances and index_of_ccpp_instance respectively...this would follow the example set by most of the current names:

"number_of" (48 instances)

• number_of_openmp_threads
• number_of_chemical_species
• number_of_tracers
• etc.
  "index_of_X" (56 instances)
• index_of_ice_vegetation_category
• index_of_air_temperature_on_previous_timestep_in_xyz_dimensioned_restart_array
• index_of_air_temperature_two_timesteps_back_in_xyz_dimensioned_restart_array
• etc.

However there is one big exception: the existing "CCPP thread" variable, which is ccpp_thread_number. Are variables like threads and instances different enough from these other "index" examples that it calls for a different naming convention, or would it make sense to convert ccpp_thread_number --> index_of_ccpp_thread? It doesn't exactly align with common language usage, but it would come with the advantage of being consistent with other variables, and also standardizing the usage of the term number in standard names to mean a single thing.

I'm also open to other suggestions if there's some other option I've missed.

[climbfuji]

I personally really don't like the idea of index_of_ccpp_thread.

ccpp_thread_number is what people would use to describe it. We can use ccpp_instance_number for consistency.

[mkavulich]

@climbfuji That's fine by me if you have a strong opinion. I agree that would fit better with the human-readable description, and there is at least some subjective difference between these variables and existing "index" variables.

So number_of_ccpp_instances and ccpp_instance_number sounds good for these new variables?

[climbfuji]

@climbfuji That's fine by me if you have a strong opinion. I agree that would fit better with the human-readable description, and there is at least some subjective difference between these variables and existing "index" variables.

So number_of_ccpp_instances and ccpp_instance_number sounds good for these new variables?

That would be my preference, yes. Thanks very much!