In-Memory Runtimes

Both in-memory runtimes store workflow state in memory without persistence. They differ in their concurrency model:

	Synchronized	Concurrent
Module	workflows4s-core	workflows4s-cats-effect
Effect requirements	MonadThrow	Async (cats-effect)
Under contention	Blocks thread	Suspends fiber
On cancellation	Lock may leak	Releases lock cleanly
Best for	Tests, simple/synchronous usage	Concurrent/async usage with fibers

Concurrent Runtime

The concurrent runtime is built on top of cats-effect and uses functional concurrency primitives (Ref, Semaphore). It is fiber-safe and cancellation-safe.

Example

import MyWorkflowCtx.*
val workflow: WIO.Initial                                       = ???
val engine: WorkflowInstanceEngine[IO, Ctx]                     = ???
val runtime: InMemoryConcurrentRuntime[IO, Ctx]                 = InMemoryConcurrentRuntime
  .default(workflow, InitialState(), engine)
  .unsafeRunSync()
val wfInstance: IO[InMemoryConcurrentWorkflowInstance[IO, Ctx]] = runtime.createInstance("my-workflow-1")

Synchronized Runtime

The synchronized runtime uses JVM-level synchronization primitives. It is thread-safe but not fiber-safe — under contention it blocks the carrier thread rather than suspending the fiber.

Example

import MyWorkflowCtx.*
val workflow: WIO.Initial                                           = ???
val engine: WorkflowInstanceEngine[Try, Ctx]                        = ???
val runtime: InMemorySynchronizedRuntime[Try, Ctx]                  = InMemorySynchronizedRuntime.create(workflow, InitialState(), engine)
val wfInstance: Try[InMemorySynchronizedWorkflowInstance[Try, Ctx]] = runtime.createInstance("my-workflow-1")