Ironflow crate

ironflow is the runtime crate that powers the Ironflow workflow engine. It exposes the core workflow state machine, Postgres-backed persistence, outbox processing, and timers, and it integrates with the ironflow-macros helpers such as HasWorkflowId.

NB: Ironflow is in active development and testing. Expect breaking changes, bugs, and potential concurrency or data issues in the current release and between releases; use in production only with caution.

Highlights

• Event sourcing with SQLx-backed persistence for workflows, timers, outbox, and projections.
• Async effect support with idempotent retries and dead-letter tracking.
• Optional SQLx/Postgres feature set that pulls in sqlx, tokio, and serde to keep the runtime lean when the database integration is not needed.

Quick start

1. Follow the workspace-level README.md to configure your environment, start Postgres, and run SQLx migrations.

2. Add the crate and the postgres feature to your workspace:

   [dependencies]
   ironflow = { version = "0.1", features = ["postgres"] }
   

3. Import the runtime pieces you need:

   use ironflow::{Decider, Workflow, WorkflowId};
   

If you build without a database, set SQLX_OFFLINE=true and rely on the shipped .sqlx/ metadata (generated via cargo sqlx prepare).

Migrations

The database schema lives under migrations/20251230000001_create_ironflow_schema.sql. Downstream consumers should apply the migrations for their desired version before running the runtime (for example, cargo sqlx migrate run --source crates/ironflow/migrations from the repository root).

Testing

See tests/postgres/README.md for the integration suites that exercise the runtime against Postgres.