It'd be nice to see documentation of how it works. Is it using a replication channel for Postgres? The write-ahead log? It looks like it does schema introspection and then works off that. How does it handle schema changes? How does this work in scenarios where you have multi-master replication? What's the compatibility with popular Postgres extensions like PostGIS? What happens if the database is rebooted or crashes? Yes I realize the source is there.