Skip to content
learn
Global menu

LadybugDB — Mission

Why I’m learning this

Section titled “Why I’m learning this”

The DuckDB course answers questions about rows: how many runs, how long, which step. Other questions about this site are about connections. How many clicks from the home page to a lesson? Which files change together? Which commits did a failing workflow run on? In SQL, each of these is a chain of self-joins, or a recursive CTE whose depth I have to guess.

A graph database stores the connections themselves, and its query language describes paths directly. LadybugDB is one that runs inside your process, like DuckDB and SQLite: no server to install.

Who this course is for

Section titled “Who this course is for”

You know SQL: SELECT, JOIN, GROUP BY, maybe a recursive CTE. You write C# or Java. You don’t need to know anything about graph databases or Cypher.

What LadybugDB is, in one table

Section titled “What LadybugDB is, in one table”
SQL Server graph tables Neo4j LadybugDB
Runs as a server as a server inside your process
Schema tables created AS NODE and AS EDGE optional: labels and properties appear when you write them required: node tables and relationship tables, with typed columns
Query language T-SQL with MATCH Cypher Cypher
Paths of any length SHORTEST_PATH variable-length patterns variable-length patterns, shortest paths
From .NET Microsoft.Data.SqlClient Neo4j .NET driver NuGet package LadybugDB
From Java JDBC driver Neo4j Java driver Maven package com.ladybugdb:lbug

Sources: SQL Graph architecture, Neo4j Cypher manual, differences between LadybugDB and Neo4j.

LadybugDB was formerly known as Kuzu. Kuzu’s authors have archived its repository; LadybugDB continues the code base under the MIT license, with new names: the CLI is lbug, and the documentation names database files .lbdb.

The data

Section titled “The data”

code/ladybugdb/data/extract.py reads this repository at commit cbcbb42 (2026-09-14) and writes six CSV files to code/ladybugdb/data:

  • pages.csv: the 319 pages of the site (117 in English, 101 in French, 101 in Spanish), with their locale, course, title and number of lines;
  • links.csv: the 712 links from one page of the site to another, found in Markdown links and href attributes outside code blocks;
  • external_links.csv: the 3,712 links to 85 other sites;
  • commits.csv, parents.csv and changes.csv: the 74 commits up to cbcbb42, the parent of each one, and the 1,019 file changes they made to 710 files.

Lesson 4 adds the 125 CI runs of code/duckdb/data/runs.json, the snapshot the DuckDB course queries.

Lessons 5 and 6 query another repository: the .NET projects of GuitarAlchemist/ga at commit a26a7893, extracted by code/ladybugdb/data/ga/extract.py: 111 projects, 266 project references and 478 package references.

By the end of this course, I will be able to

Section titled “By the end of this course, I will be able to”
  • model data as node tables and relationship tables, and query it with Cypher patterns;
  • load CSV and JSON files, and find the rows that don’t fit the schema;
  • write variable-length paths and shortest paths, and know the path semantics they use;
  • recognize LadybugDB’s bugs and limits, and check a result another way;
  • use LadybugDB from C# and from Java.

Outline

Section titled “Outline”
# Lesson If you know SQL
1 A first graph: tables, CREATE, MATCH, MERGE CREATE TABLE, INSERT, MERGE, JOIN
2 Loading files: LOAD FROM, COPY, warnings OPENROWSET, BULK INSERT
3 Paths: variable length and shortest paths recursive CTEs
4 Git history and CI runs as a graph self-joins, junction tables
5 LadybugDB from C# ADO.NET
6 LadybugDB from Java JDBC
7 Graph algorithms and full-text search full-text indexes, CONTAINSTABLE
8 Persistence, transactions and concurrency isolation, locks, the transaction log
9 A graph lab: translations, prerequisite cycles, bounded reachability test fixtures, a recursive CTE with a bound
— Journal

Resources

Section titled “Resources”