We've had some data flows configured to give self-service management over what used to be a complex ETL process.
Several of these "broke" recently, generating errors and all kinds of issues. After extensive research, we've found that the option to Save to a Dataset Table used to save the mentioned table as its own dataset. It now writes a table into the save-to dataset. This is then an "unlinked" table in the data set design and causes errors and triggers multiple indications that the data needs to be reprofiled.
This leaves us (1) trying to fix the initial data flows and then updating all of the downstream dataflows which had been designed using the dataset tables as inputs; (2) with very low confidence about the long-term stability of using data flows.
Has anyone else experienced this? Was there something in an announcement that we've overlooked?
