This thesis deconstructs Datalog from a categorical and type theoretic perspective to determine what makes it tick.
Datalog’s semantic guarantees are provided by brute syntactic restrictions, such as stratification and the absence of function symbols.
In place of these, we find compositional semantic properties such as monotonicity, which we capture using types.
We show that this permits integrating Datalog’s features with those of typed functional languages, such as algebraic data types and higher order functions.
Jack Rusher 2022-11-15 17:26:27
"We have not attempted to embed Datafun into an existing language, as this would greatly complicate the context-management operations needed to ensure monotonicity." 😿