Inconsistent types of processor IDs: int vs. t8_gloidx_t

[spenke91]

Currently, some functions in t8code are using int for process IDs, others t8_gloidx_t. On the long run, it would be cleaner to always use the same to avoid problems (i guess only for very large numbers of processes).

In #2026, a new type t8_procidx_t is / has been added. No matter whether it is set to int or t8_gloidx_t, using it everywhere for process IDs will make sure to be consistent.

Estimated priority: Low
Estimated workload: Medium

[Davknapp]

I think we should use an int for that. The process id will be most likely used in a call to an MPI-Function, which uses int for sending/receiving processes.

[Davknapp]

Sorry, I think in the PR I said suggested to use an t8_gloidx_t, because we use it elsewhere. I think it is not a good argument after having a more detailed look into it

[spenke91]

Thanks for the reply. Sure, than it might still make sense to use t8_procidx_t everywhere and just set it to int - or at least change the existing usage of t8_gloidx_t...

[Davknapp]

Yes, consistently using t8_procidx_t everywhere should be the right way to go. We have to take care to cast it into an t8_gloidx_t when using the formula to compute the process boundaries.