We are thrilled to announce the alpha release of TopGun v2. This isn’t just a version bump—it’s a completely new data platform built from the ground up.
The Journey: From v1 to v2
I spent two years developing TopGun v1 as a TypeScript implementation of the gun.js protocol. While we achieved a lot, we hit fundamental architectural walls:
- Query Limitations: We couldn’t easily request subsets of data. Pagination, complex filters, and sorting were nearly impossible to implement efficiently.
- Full Replication: Every connected server node required a full copy of the dataset, which made “smart” horizontal scaling difficult.
- Protocol Constraints: The graph synchronization protocol, while powerful, was hard to optimize for the predictable performance required by modern applications.
The Search for a Better Way
Before building v2, I explored using PostgreSQL with logical replication for real-time updates. The idea was tempting: use a battle-tested SQL database for storage and querying, and just stream changes to clients.
However, reality struck:
- Latency: Logical replication introduced unacceptable delays for real-time interaction.
- Vendor Lock-in: The architecture became too tightly coupled to Postgres specific features.
- Schema Rigidity: Pre-defined schemas (Organizations -> Users -> Roles) limited the flexibility that local-first apps often need.
- Auth Complexity: Offline authorization with a traditional SQL backend is a nightmare of edge cases.
Enter TopGun v2: The In-Memory Data Grid
For v2, we took a radical new approach. TopGun v2 is an In-Memory Data Grid that serves as the source of truth for your active application state.
Key Innovations:
- Zero-Latency Realtime: Because data lives in memory, reads and subscription updates are instantaneous.
- Agnostic Persistence: You can use any cold storage (Postgres, SQLite, S3). TopGun acts as the high-speed cache and synchronization layer.
- Smart Sharding: Unlike v1, nodes don’t need the full dataset. They can dynamically load and unload parts of the graph based on demand.
- True Offline Support: By moving authorization and logic to the protocol level, we support secure offline operations without custom fragile code.
We are currently in Alpha. The core is stable, but APIs might change. We invite you to try it out and help shape the future of local-first development.