Comment by noworriesnate

I'm building an in-process, peer-to-peer, event sourced, categorical database in Rust. I think more and more people are going to start building their own personal apps, and people are not going to want to have to manage a server.

So the tradeoff this database makes is it doesn't work at high scale, but it has a lot of features not typically in a database, such as:

- The ability to embed non-turing-complete event source handlers in the database schema

- Categorical type system that makes migrations provably secure

- All schema changes take place through migrations

Example syntax:

  Inventory = aggregate(key = sku) {
      sku:       String
      name:      String
      available: crdt(merge = counter) I64

      command Stock(sku: String, name: String, qty: U64) {
          sku       = input.sku
          name      = input.name
          available += input.qty
      }

      command Reserve(sku: String, qty: U64) {
          available -= input.qty
      }

      command Release(sku: String, qty: U64) {
          available += input.qty
      }
  }

  command SubmitOrder(order_id: String, party_id: String, sku: String, qty: U64, total: U64) {
      Order::PlaceOrder(
          id       = input.order_id
          party_id = input.party_id
          sku      = input.sku
          qty      = input.qty
          total    = input.total
      ) compensate Order::Cancel()
      # `sku` is Inventory's key — the step must bind it so the engine can
      # resolve which Inventory row the command targets (task-212).
      Inventory::Reserve(sku = input.sku, qty = input.qty)
          compensate Inventory::Release(sku = input.sku, qty = input.qty)
  }