Currently, the whole simulation is written to a single file at the end of simulation. This has three potential drawbacks:
Sometimes you want to split your simulation into multiple files for easier handling after the simulation (eg. splitting on time)
Sometimes you want to split the simulation into different chunks which run independently, but might want to write the results into one file (eg. vis cpu is most easily run by sending each frequency to its own SLURM task. It would be nice if the results were all written back into one file though).
Sometimes you might want to "checkpoint" the simulation for large sims.
It would be good to think about what an extension to the "filing" configuration might look like that can enable this. Also, keep in mind that there should be some conceptual separation between the configuration and the implementation -- hera_sim for example uses the pyuvsim configuration definition, but not its implementation of the simulation itself.
Currently, the whole simulation is written to a single file at the end of simulation. This has three potential drawbacks:
It would be good to think about what an extension to the "filing" configuration might look like that can enable this. Also, keep in mind that there should be some conceptual separation between the configuration and the implementation -- hera_sim for example uses the pyuvsim configuration definition, but not its implementation of the simulation itself.