Good info, interesting. It does add a layer of complexity that takes a bit more to handle. The point of moving away from stateful/naming was to prevent global state and bugs but you can also run into those in other ways as it is somewhat like going from keyed data to offsets/positions binary data. Rebuilding pipelines also seems heavy.

Maybe the solution is a lower level API (current) and a higher level API (somewhat stateful/named but not so global). That does add extra weight though.