I'm working on a new relational / logic language inspired by datalog and Bruce MacLennan's pioneering work on relational programming, but adopting the open-world assumption to allow missing values / heterogeneous data and enforce data integrity. I think logic languages will become increasingly important in the LLM era to provide strong correctness guarantees, and as a source of reinforcement learning. And I'm super fascinated by relational data models -- I think they have many advantages over the object-oriented view.

https://github.com/roland-redmount/atom https://roland-redmount.github.io/atom/

It's work in progress, many features are not completed. Any feedback is appreciated!