Detailed Description

Writer TRUST -> CGNS for post-processing outputs.

This class centralizes all CGNS writing strategies used by TRUST postprocessing. The implementation supports several execution contexts and file layouts:

sequential single file,
parallel single file,
parallel over-zone output,
linked files (grid + one solution file per post time + one final link file),
communicator-group outputs.

The current design intentionally keeps some filename construction and mode dispatch explicit in the implementation. For this code, readability of the different CGNS workflows is preferred over a stronger abstraction layer.

High-level behavior:

In single-file modes, the main CGNS file stays open as long as possible and is flushed and/or closed according to Option_CGNS::FLUSH_EVERY_N and Option_CGNS::CLOSE_EVERY_N.
In linked-file modes, the mesh file is separated from solution files. A final link file is rebuilt at each post time so the result remains directly visualizable during the simulation.
For deformable or lagrangian cases, additional constraints apply because coordinates and/or topology evolve with time.

Important design note about the final link file: the file is currently rebuilt at each post time instead of being incrementally updated. This is a deliberate trade-off. In practice, CGNS/HDF5 append-style updates on iterative nodes (BaseIterativeData_t, ZoneIterativeData_t) were found to be more fragile and not clearly faster than rebuilding this small index file.

Definition at line 58 of file Ecrire_CGNS.h.

The documentation for this class was generated from the following file:

/home/docs/trust-code/src/Kernel/Postraitement/CGNS/Ecrire_CGNS.h