Strong Towns Madison Backend

Prerequisites

• Rust setup
  • Install Rust
  • Setup Rust-Analyzer
  • Setup rustfmt
  • Setup clippy
• Install Docker
• Rust (book)

Getting Started

1. Clone the repo

2. Copy .env.example to .env and update values

3. Start the database

# todo

4. Run migrations

# todo

5. Start the dev server

DUCKDB_DOWNLOAD_LIB=1 cargo run

6. Run tests

DUCKDB_DOWNLOAD_LIB=1 cargo test

Code Quality

DUCKDB_DOWNLOAD_LIB=1 cargo fmt
DUCKDB_DOWNLOAD_LIB=1 cargo clippy
DUCKDB_DOWNLOAD_LIB=1 cargo test

Parcel Tiles

This repo includes a static parcel tile pipeline for the full silver.parcels dataset (82,152 parcels). It extracts parcel geometries from DuckDB/GCS, writes newline-delimited GeoJSON, and builds a single PMTiles archive for MapLibre.

Local Build

Prerequisites:

• duckdb CLI
• tippecanoe
• GCS_KEY_ID and GCS_SECRET in your environment or .env

Run:

set -a
source .env
set +a
./scripts/generate_parcel_pmtiles.sh

Output:

• build/tiles/parcels.geojson.ndjson
• build/tiles/madison-parcels.pmtiles

Defaults:

• minimum zoom: 10 (citywide Madison view)
• maximum zoom: 14

Override those if needed:

MIN_ZOOM=9 MAX_ZOOM=15 ./scripts/generate_parcel_pmtiles.sh

GitHub Action

The scheduled workflow is in .github/workflows/parcel-tiles.yml. It runs weekly and can also be triggered manually.

Required GitHub secrets:

• GCS_KEY_ID
• GCS_SECRET
• CLOUDFLARE_R2_ACCESS_KEY_ID
• CLOUDFLARE_R2_SECRET_ACCESS_KEY

Required GitHub repository variables:

• CLOUDFLARE_R2_BUCKET
• CLOUDFLARE_R2_ENDPOINT

Optional GitHub repository variable:

• CLOUDFLARE_R2_OBJECT_KEY (defaults to madison-parcels.pmtiles)

Tech Stack

• Axum
• Tokio
• ThisError
• Serde, Serde JSON
• Tower Middleware
• env_logger
• iter_tools
• chrono
• dotenvy
• sqlx
• validator
• bcrypt
• jsonwebtoken
• uuid