SqlCAD Documentation

Welcome to the documentation for SqlCAD (SQL Computer Aided Design).

SqlCAD is a compiler for data warehouse modeling. It allows you to define your data model as a graph of Entities and Relationships, and then automatically generates optimized, flattened SQL for your target data warehouse.

Why SqlCAD?

• Graph-First: Define joins once, use them everywhere. No more copy-pasting LEFT JOIN clauses.
• Auto-Flattening: Build wide "Gold" tables (Marts) automatically by traversing the relationship graph.
• Automated DDL: Automatically generate CREATE VIEW (for sources) and CREATE TABLE (for entities) statements, managing file reading and schema creation.
• Point-in-Time Correct: Built-in support for SCD Type 2 history and correct temporal joins (AS OF logic).
• Strongly Typed: Validates schemas and column dependencies before you run any SQL.

Getting Started

• Tutorial: Build a complete data warehouse from scratch.
• Concepts: Understand the Source -> Silver -> Gold architecture.
• Schema Evolution: Using the Diff Engine and declarative schema management.
• Reference: Syntax guide for .sqlcad files.

Guides

• Loading Strategies: Learn about SCD2 and Incremental Marts.
• Features & ML: How to build Point-in-Time correct features.
• Data Lineage: Visualize dependencies and trace column provenance.
• Observability: Statistics collection and data quality assertions.
• Warehouses: Configuration for DuckDB, BigQuery, and Snowflake.
• Comparison: How SqlCAD differs from dbt and SQLMesh.

Installation

git clone https://github.com/your-repo/sqlcad.git
cd sqlcad
uv sync