A batch job, in The Elm Architecture
Summary
This article explores building a real batch job using The Elm Architecture, emphasizing a per-item state machine and the shift from a pipeline to a looped, stateful batch process. It argues that Elm enables reliable batch processing across multiple runtimes (Lamdera frontend/backend and a native batch binary) by sharing domain logic and exhaustive types. The piece highlights concepts like checkpoints, retries, and replaying failures to improve reliability in batch workloads.