Inconsistent use of submatrix fields?

I'm not sure about this but I feel like there are inconsistent use of submatrix parameters (A->i and A->j field in the descriptor)

My understanding is that it should be used each time we need to resolve global matrices indices (i.e when we need to access the starpu data handle when starpu is used for instance) from sub-matrices indices, and that all 'pz' driver are supposed to work with sub-matrices indices.

Is it ok that here(MORSE_Task_dataflush(),RUNTIME_desc_iscached()) and here (get_rankof_2d()) (for instance) A->i and A->j are not used whereas it is used here (get_blk_addr_ccrb()) (and other get_blk_addr() functions in general) ?

(It seems that most of functions that are related with MPI does not use the fields.)

Is there record of MPI distributed tests with some of the algorithms that relies on the submatrices facilities?