Symptom: post_process misreads simulation restart/output layouts for bubbles_lagrange combined with mhd, elasticity/stress, or cont_damage — the void-fraction field lands at a different index than post expects. No checker forbids the combinations.
Mechanism: on master, post folds beta into sys_size mid-build, BEFORE the mhd/elasticity/cont_damage blocks (src/post_process/m_global_parameters.fpp:480-483 vs :485+), while simulation writes beta at sys_size + 1 after the complete system (src/common/m_mpi_common.fpp:115-126).
Introduced: the original lagrange PR #749 was consistent; the divergence came with the MPI refactor #831 (8e94d0a5f, 2025-06-13, @wilfonba), which moved post's beta into sys_size ahead of the pre-existing feature blocks. Tagging @wilfonba for context.
Fix: #1552's unified equation-index builder (one s_initialize_eqn_idx shared by all three executables) makes the layouts agree by construction.
Symptom: post_process misreads simulation restart/output layouts for
bubbles_lagrangecombined with mhd, elasticity/stress, or cont_damage — the void-fraction field lands at a different index than post expects. No checker forbids the combinations.Mechanism: on master, post folds beta into
sys_sizemid-build, BEFORE the mhd/elasticity/cont_damage blocks (src/post_process/m_global_parameters.fpp:480-483vs :485+), while simulation writes beta atsys_size + 1after the complete system (src/common/m_mpi_common.fpp:115-126).Introduced: the original lagrange PR #749 was consistent; the divergence came with the MPI refactor #831 (
8e94d0a5f, 2025-06-13, @wilfonba), which moved post's beta into sys_size ahead of the pre-existing feature blocks. Tagging @wilfonba for context.Fix: #1552's unified equation-index builder (one
s_initialize_eqn_idxshared by all three executables) makes the layouts agree by construction.