A self-contained Docker Compose stack that spins up MySQL, PostgreSQL, and SQL Server, each pre-loaded with English seed data so you can explore Tabularis features end to end.

tabularis_demo is the same logical schema on all three engines (departments, employees, products, customers, orders, order_items) — useful for testing the showcase notebook against any driver. The second database per engine highlights features specific to that ecosystem.

On MySQL and PostgreSQL the perf_demo database holds a single wide_table with 50 columns × 50 000 rows (a mix of int, bigint, decimal, double, varchar, text, date, datetime, boolean and JSON/JSONB columns). It reproduces slow-scroll reports with many columns/rows so DataGrid virtualization and memoization can be profiled — open perf_demo.wide_table and scroll both axes; it should stay fluid. This is a heavy seed, so the first container boot takes a little longer. The SQL is generated by generate-perf-sql.py (edit there and re-run to change the column/row counts).

• Docker Desktop or Docker Engine 24+ with the Compose plugin
• ~2 GB free disk space for the three image layers and volumes
• On Apple Silicon: Rosetta 2 enabled (SQL Server runs under linux/amd64)

cd demo
docker compose up -d

First boot takes ~60 seconds — the SQL Server sidecar (tabularis-mssql-init) waits for the server, then runs the init scripts via sqlcmd. MySQL and PostgreSQL run their init scripts automatically on first launch.

Check everything is healthy:

docker compose ps
docker compose logs -f mssql-init   # confirm SQL Server seeding succeeded

Tear down (preserve volumes):

docker compose down

Wipe all data and start fresh:

docker compose down -v

All three servers share the same password for simplicity:

The password meets SQL Server's complexity requirements; do not weaken it without updating docker-compose.yml accordingly.

Open Tabularis → Connections → Import and pick connections.json.

This adds a Tabularis Demo (Docker) group with pre-configured connections:

• Demo · MySQL — exposes tabularis_demo, blog_demo and perf_demo
• Demo · PostgreSQL — exposes tabularis_demo
• Demo · PostgreSQL (analytics_demo) — the JSONB analytics database
• Demo · PostgreSQL (perf_demo) — the wide-table scroll stress test

SQL Server is not in connections.json. Tabularis core currently ships drivers for MySQL, PostgreSQL, and SQLite only; the official plugin registry does not yet include a SQL Server plugin. The MSSQL instance is still useful for connecting external clients (Azure Data Studio, DBeaver, sqlcmd) and for testing future plugin work.

Once Demo · MySQL is connected and selected on tabularis_demo, import notebook-showcase.tabularis-notebook from the Notebooks panel and press Cmd/Ctrl + Shift + Enter to run all cells.

demo/
├── docker-compose.yml
├── connections.json                # Importable into Tabularis
├── generate-perf-sql.py            # Regenerates the 06-perf-wide.sql seeds
├── notebook-showcase.tabularis-notebook
├── init/
│   ├── mysql/
│   │   ├── 01-tabularis-demo.sql
│   │   ├── 02-blog-demo.sql
│   │   ├── ...
│   │   └── 06-perf-wide.sql        # 50 cols x 50k rows (perf_demo)
│   ├── postgres/
│   │   ├── 01-tabularis-demo.sql
│   │   ├── 02-analytics-demo.sql
│   │   ├── ...
│   │   └── 06-perf-wide.sql        # 50 cols x 50k rows (perf_demo)
│   └── mssql/
│       ├── run-init.sh             # Sidecar entrypoint
│       ├── 01-tabularis-demo.sql   # Idempotent
│       └── 02-finance-demo.sql     # Idempotent
└── README.md

SQL Server fails to start on Apple Silicon. Ensure Rosetta is enabled in Docker Desktop → Settings → General → "Use Rosetta for x86_64/amd64 emulation". The platform: linux/amd64 line in docker-compose.yml forces amd64.

Port already in use. Stop any local MySQL/PostgreSQL/MSSQL instance, or edit the ports: mapping in docker-compose.yml (left side is the host port).

Need to re-run SQL Server init. The sidecar (mssql-init) only runs on up. If you change the SQL scripts, run docker compose up -d mssql-init again — the scripts are idempotent and safe to re-execute.

Connection import does nothing. Tabularis merges connections by id; if you previously imported the file and want to reset, delete the existing "Demo ·" connections in Tabularis first, then re-import.