ADR-006: Semantic Layer & Metrics Framework (DRY)¶
Status: Proposed
Date: 2026-01-24
Context¶
Manufacturing costing involves 50+ inter-dependent metrics. Without a centralized semantic layer, logic becomes fragmented across reports, dashboards, and APIs, leading to inconsistent quotes.
Decision¶
Implement a Metadata-Driven Semantic Layer to enforce the DRY (Don't Repeat Yourself) principle.
- Centralized Definitions: All formulas (e.g.,
Material Cost) are defined in a YAML catalog. - Logic Separation: The calculation engine consumes these definitions dynamically.
Rationale¶
- Consistency: Eliminates "Excel Hell" where different sheets use different versions of the same formula.
- Agility: A 1-line update in the YAML catalog propagates to all quotes and dashboards.
- Auditability: Dependency graphs allow clear "Why this number?" drill-downs for auditors.
Consequences¶
- Positive: Single source of truth; 0% logic drift across the platform.
- Negative: Initial overhead of defining the 50+ metric catalog.