Ah, here's where this spill over to/from the "traditional vs. in-line" ML debate came to and from.