Tasks for the IOSS writer
General requests:
- For all tasks below, write mesh, point and cell data
- Output of the IOSS writer will target post processing. Currently it is envisioned that this output will not be used for input to simulation codes. Output files will read cleanly into ParaView.
- IOSS Writer will be able to seamlessly process restart files read into ParaView.
- IOSS Writer will never be expected to split datasets into restarts.
Here is a list of issues needed for a first pass of the IOSS writer.
-
Exodus file format. -
Parallel writer. Match one rank to one output file. (i.e., if we have 32 input files read by 4 ranks, write out 4 files.) -
Write dataset that was read in. -
Write out data and mesh that is a subset of dataset read in. No new cells will be created. Example is output of the threshold filter. -
Write out data and mesh that does not match the input mesh. Examples are output of the Slice, Clip, isoVolume, Contour and Extract Surface filter. -
Preseve blocks. -
Write global IDs correctly. Not necessary (yet) to create Pedigree IDs for created cells. -
#22146 (closed) IOSS Writer dialog cleanup
Here is a list of issues needed for the second pass of the IOSS writer.
-
CGNS file format. For all bullets below, write mesh, point and cell data. -
Allow writer to follow Redistribute Dataset filter. If we have 16 ranks, and want to move to 8 files out, run the Redistribute Dataset filter. -
Allow writer to write out global data. -
Allow writer to output side sets and node sets. This includes variable data on these sets. -
Write pedigree IDs for datasets that include created cells. Examples of filters are the list of Slice, Cliip, etc. above. -
Allow control over timesteps written. This includes start timestep, stop timestep and stride. (This is a secondary need as the ExtractTimeSteps filter replicates this feature.) -
Allow variable data to be selectively written.
Edited by W. Alan Scott