Currently, the cell->vertices to face->cells and face->vertices connectivity conversion is done in the proprocessor. Migrating this to the solver requires modifying the preprocessor reader, but would have several advantages:
Lower memory usage (about half) in the non-parallel preprocessor
Migrate all postprocessing and mesh checking aspects to the solver
prerequisite to allow adding parallel mesh readers directly in solver
This could also be done in a separate order, adding specific readers to the solver (reading into an FVM mesh for formats such as MED or CGNS, allowing interpolation operations) first.
Note: this is an update from an issue on the old (Mantis) issue tracker, which is a long-term goal.
Currently, the cell->vertices to face->cells and face->vertices connectivity conversion is done in the proprocessor. Migrating this to the solver requires modifying the preprocessor reader, but would have several advantages:
This could also be done in a separate order, adding specific readers to the solver (reading into an FVM mesh for formats such as MED or CGNS, allowing interpolation operations) first.
Note: this is an update from an issue on the old (Mantis) issue tracker, which is a long-term goal.