Also: do I need to run a producer & consumer (and maybe some intermediary filters) on a stream of data, and want to run them in parallel? Shell pipelines are trivially easy to set up & test.