While developing Dit, one of the significant challenges we faced was effectively tracking and managing database changes over time. Traditional methods of version control were not directly applicable to databases, primarily due to the complex nature of database schema and the need to track both structure and content changes efficiently.

Initially, we attempted to leverage existing file systems and version control techniques, but we quickly ran into limitations. Specifically, these methods struggled to handle the hierarchical and relational nature of databases with issues in performance, scalability, and the precision of change tracking, making it difficult to provide a reliable history of database states and changes.

To overcome these challenges, we developed our own custom file system, "rarc". This system is designed specifically for database version control, utilizing B+ trees to efficiently store database schemas and content. B+ trees offered several advantages, including the ability to handle large amounts of data with high efficiency, both in terms of speed and storage space. This was crucial for supporting large databases and complex schema changes.

Additionally, we employed topological sorting to track and order changes. This approach allowed us to accurately maintain the dependency relationships between different parts of the database schema, ensuring that changes could be applied, rolled back, or merged in a coherent and consistent manner. By understanding the dependencies between changes, Dit could intelligently manage the application and reversion of changes, even in complex scenarios with multiple concurrent branches of work.

Creating the "rarc" system and integrating it with B+ trees and topological sorting techniques was the breakthrough and we successfully demonstrated that our system could not only track changes more accurately than traditional methods but also perform operations like merges and rollbacks efficiently and reliably.