I wrote this a long time ago as a proof-of-concept for a vaporware project. Recently I have had the need to something which seems easy at a high-level standpoint, but is actually quite hard to implement.
The original purpose for this concept was to easily maintain a strict order of data held within the rows, while still having the option to reorder the rows in an efficient manner (one table, no joins, no special embedded scripts etc).
A little insight into what I was doing: I once worked on a network security team for a tier-1 ISP. A coworker and myself had already written some really nice code to deal with the thousands (yes, thousands) of firewalls we dealt with on a daily basis. We created an API which generated abstracted structures of firewall configurations; by abstracting I mean creating a singular data-structure which represented rules, and rule-sets (a nice side-effect of this method was we could morph one firewall syntax structure to another. E.g., juniper filters -> netscreen rules. Needless to s