Symptom: multi-rank chemistry post-processing silently drops or garbles species (Y_*) output on ranks != 0.
Mechanism: chem_wrt_Y is namelist-read (rank 0 only) and consumed on all ranks by s_save_data (src/post_process/m_start_up.fpp:292 on current master), but it is absent from s_mpi_bcast_user_inputs in src/post_process/m_mpi_proxy.fpp — its sibling chem_wrt_T is in the logical broadcast list, chem_wrt_Y is not. Non-root ranks always see the default.
Introduced: #544 (17eb3def8, 2024-09-16, @henryleberre) added the namelist binding and consumption but only updated the simulation proxy. #829 later fixed chem_wrt_T and missed chem_wrt_Y. Tagging @henryleberre for context as the original author.
Fix: #1552 replaces the hand-maintained broadcast lists with registry-generated ones (generated_bcast.fpp), which closes this by construction — a registered namelist parameter is broadcast or the generator is wrong.
Symptom: multi-rank chemistry post-processing silently drops or garbles species (
Y_*) output on ranks != 0.Mechanism:
chem_wrt_Yis namelist-read (rank 0 only) and consumed on all ranks bys_save_data(src/post_process/m_start_up.fpp:292on current master), but it is absent froms_mpi_bcast_user_inputsinsrc/post_process/m_mpi_proxy.fpp— its siblingchem_wrt_Tis in the logical broadcast list,chem_wrt_Yis not. Non-root ranks always see the default.Introduced: #544 (
17eb3def8, 2024-09-16, @henryleberre) added the namelist binding and consumption but only updated the simulation proxy. #829 later fixedchem_wrt_Tand missedchem_wrt_Y. Tagging @henryleberre for context as the original author.Fix: #1552 replaces the hand-maintained broadcast lists with registry-generated ones (
generated_bcast.fpp), which closes this by construction — a registered namelist parameter is broadcast or the generator is wrong.