02 — Passing data with Context

A Context is the typed, persisted container that flows through a workflow. One task writes; a downstream task reads.

Two dependent tasks

produce writes a value; consume depends on it and reads it.

Rust

#[workflow(name = "pipeline", description = "Pass data downstream")]
pub mod pipeline {
    use super::*;

    #[task]
    pub async fn produce(ctx: &mut Context<serde_json::Value>) -> Result<(), TaskError> {
        ctx.insert("numbers", serde_json::json!([1, 2, 3]))?;
        Ok(())
    }

    #[task(dependencies = ["produce"])]
    pub async fn consume(ctx: &mut Context<serde_json::Value>) -> Result<(), TaskError> {
        let nums = ctx.get("numbers").cloned().unwrap_or_default();
        ctx.insert("sum", serde_json::json!(/* sum of nums */ 6))?;
        Ok(())
    }
}

Python

with cloaca.WorkflowBuilder("pipeline") as builder:
    builder.description("Pass data downstream")

    @cloaca.task()
    def produce(context):
        context.set("numbers", [1, 2, 3])
        return context

    @cloaca.task(dependencies=["produce"])
    def consume(context):
        nums = context.get("numbers")
        context.set("sum", sum(nums))
        return context

The dependencies list makes consume run after produce, so the value is present when it reads. The context is persisted, so it survives a restart and is returned as result.final_context.

Next

• 03 — Dependencies & parallelism
• Reference: Context