In the current implementation of start_pipeline, the param files from before selfcal are deleted and selfcal (or any other later step) would create a new one.
This means that the summary functions of preflag, crosscal, and convert cannot be used to at the end of the pipeline to get information on the performance of these steps (for example for the QA). Or if a beam fails preflag or crosscal, convert and all subsequent steps would still run.
It would be good to think of a way to keep them, maybe by moving or renaming if deleting them is necessary.
In the current implementation of start_pipeline, the param files from before selfcal are deleted and selfcal (or any other later step) would create a new one. This means that the summary functions of preflag, crosscal, and convert cannot be used to at the end of the pipeline to get information on the performance of these steps (for example for the QA). Or if a beam fails preflag or crosscal, convert and all subsequent steps would still run. It would be good to think of a way to keep them, maybe by moving or renaming if deleting them is necessary.