databuild/docs/narrative/why-not-sql-backend.md

If you look at the collection of state machines, objects, and state transitions that make up databuild, its tempting to ask, should this be implemented by a SQL database instead with constraint checks and foreign keys?

TODO