Hstore, and the supplied functions for converting to/from records has been very useful for making a "one size fits all" history table. Currently I'm working wi

Hstore, and the supplied functions for converting to/from records has been very useful for making a "one size fits all" history table. Currently I'm working with numeric range types to model order quantities and stock quantities, and forecasting which order is going to consume what stock by joining on rows with overlapping ranges, then taking the intersections of those overlaps. Again, Postgres supplies functions and operators for range overlaps, intersections, etc. In the absence of those datatypes, there'd be a lot more work required to achieve either of these.