Execution Model

DAG execution converts validated graph semantics into node outcomes and artifact evidence under explicit scheduler and policy controls.

Visual Summary

flowchart LR
    parse["parse and validate graph"] --> plan["lower to execution plan"]
    plan --> schedule["scheduler selects ready nodes"]
    schedule --> execute["engine invokes adapters"]
    execute --> persist["write run and artifact evidence"]
    persist --> classify["replay and diff classification surfaces"]

Execution Stages

1. parse, validate, canonicalize graph input
2. lower to runtime execution plan
3. scheduler computes dependency-correct frontier
4. engine executes nodes through adapter boundaries
5. runtime writes node traces and run outputs
6. replay/diff services consume persisted evidence

Code Anchors

• crates/bijux-dag-core/src/pipeline/parse.rs
• crates/bijux-dag-core/src/planner/planner.rs
• crates/bijux-dag-runtime/src/runtime_core/execution/scheduler.rs
• crates/bijux-dag-runtime/src/runtime_core/execution/engine.rs
• crates/bijux-dag-artifacts/src/lib.rs

Next Reads

• State and Persistence
• Operator Workflows
• Failure Recovery