Proposal: Native Event Sourcing Support (Event Store + Projections)

[YBazanPro]

SharpCoreDB is already a great fit for modern .NET apps (relational + vectors + graph work in progress). One big missing piece that would make it a truly multi‑model engine is native support for event sourcing.

Right now, you can of course implement event sourcing “on top” using a normal table, but having first‑class support in the engine would simplify patterns, improve performance, and make SharpCoreDB a very compelling alternative to dedicated event stores in local / embedded scenarios.

What I’m thinking of

I’m not suggesting a full-blown framework (aggregates, CQRS, etc.), just solid low-level primitives:

1. Append‑only Event Store

   • A table mode or option designed specifically for event streams:
     • StreamId (string / GUID)
     • Sequence (long, per stream)
     • EventType (string)
     • Payload (JSON/BLOB)
     • Metadata (JSON/BLOB)
     • Timestamp
   • Guarantees:
     • Immutable append‑only writes
     • Ordered events per stream
     • Optimized for sequential writes / reads
   • Simple SQL / API:
     • APPEND EVENT for one or many events
     • READ EVENTS FROM streamId [FROM sequence] [TO sequence]

2. Lightweight Snapshots

   • Optional snapshot table per stream to avoid replaying thousands of events:
     • StreamId, Version, SnapshotData, CreatedAt
   • No complex snapshot strategy in the engine, just efficient storage so apps can implement their own policies.

3. Projection‑friendly features

   • The ability to efficiently:
     • Read all events from a given stream
     • Read all events from all streams in global order (for building projections)
   • Maybe later: a simple notion of “projection checkpoint” helper, but that could stay on the application side.

Why this fits SharpCoreDB well

• The engine already has:
  • WAL, append‑only storage modes, triggers, procedures.
  • Strong focus on embedded/local scenarios where a heavy external event store (like EventStoreDB) is overkill.
• With event sourcing support, a single SharpCoreDB file could hold:
  • Events (write model)
  • Relational read models
  • Vector indexes for RAG
  • Graph structures for GraphRAG
  • (Maybe later) JSON documents

That would make SharpCoreDB an amazing “all‑in‑one” local engine for .NET apps that use CQRS/ES + RAG + graph, without needing multiple external services.

[MPCoreDeveloper]

Actually I have some close contact with developers who are huge fans of event sourcing , so yes that would be something that I would like to do