Skip to content

DAG Architect (Data Lineage)

Architecture: Topological Sorting & Dead Logic Deflection

Summary: The DAG (Directed Acyclic Graph) Architect parses the structural intent of the repository to map Input/Output data flows. By analyzing which programs read and write to specific physical files, it mathematically calculates the perfect execution order for the entire mainframe cluster.

The Ghost Deflector

Before analyzing data flows, the DAG Architect queries the Intermediate Representation (IR) State Manager for a list of dead paragraphs (unreachable logic). It deliberately masks out these dead zones during its AST scan. This prevents the regex engine from detecting OPEN statements that mathematically will never execute, ensuring the generated dependency graph is completely free of hallucinated edges.

I/O Intent Mapping

The architect maps internal COBOL variables to their external physical boundaries (DD Names). It categorizes the execution intent based on standard legacy operations: * Read Operations: OPEN INPUT * Write Operations: OPEN OUTPUT * Mutation Operations: OPEN I-O and OPEN EXTEND (These act as both Read and Write dependencies).

Topological Execution Pipeline

Once all dependencies are mapped across the repository, the engine applies Kahn's Algorithm for topological sorting. * It calculates the in-degree (number of prerequisites) for every program. * It builds a strict, linear execution sequence where no program runs until its data dependencies are fully satisfied. * Deadlock Detection: If the algorithm detects a cyclic dependency (e.g., Program A waits on File 1 from Program B, but Program B waits on File 2 from Program A), it immediately halts the pipeline and isolates the deadlocked nodes for architectural review.



🌌 Powered by the blAST Engine

This documentation is part of the GitGalaxy Ecosystem, an AST-free, LLM-free heuristic knowledge graph engine.