Features
- Dual-engine architecture: Pure Python (
fdag.dag) and High-performance Cython (fdag.fast_dag) - Automatic fallback to the Python engine if Cython is unavailable
- Reactive lazy evaluation via dynamic dependency tracking
- Intelligent caching and selective node invalidation
- Node overrides at runtime with
@node(CAN_OVERRIDE) - Modern Python type hints (3.12+) and PEP 8 compliance
- Seamless integration with Pandas and NumPy for heavy math operations
- Advanced Graph Management
- Strict proactive cycle detection (Topological Sort)
- Iterative traversal for invalidation to prevent stack overflow
- Temporary graph overrides to calculate deltas in what-if scenario analysis
- Persistence & Visualization
- Persistent node caching (MongoDB/SQLite)
- Visual graph export (Graphviz/Mermaid)
- Interactive web-based graph visualization
- Quality Assurance & Distribution
- Automated performance benchmark suite
- Cross-platform CI/CD pipeline and pre-compiled wheel distribution