vector-demo

This illustrates the workflow to serve tilesets based on the raw assets that are used by geoserver for the default_bathymetry layer. This is essentially:

• convert shapefiles (Australia, and world continents outlines) to GeoJSON
• convert Tifs to GeoTif (world bathymetry image and Australian-region bathymetry image)
• convert GeoJSON and GeoTifs to mbtile tileset files (sqlite)
• serve tilesets using docker
• point JS mapping APIs to a pre-prepared style which points to the tilesets being served

Requirements

• Docker

Note on resources

Resources in ./resources directory are as per the raw files rendered by geoserver-static. However...I couldn't find supporting .shx and .dbf files for the world continents shape file layer so for the puposes of this demo I've sourced a continent shape file from elsewhere then manually removed Australia.

Build and run Vectoriser

The Vectoriser is an docker container that has several GIS tools installed

Build the Vectoriser image

#from project root dir
docker build -t vectoriser ./docker/Vectoriser

Run the Vectoriser container (takes a few minutes)

#from project root dir 
docker run -v "$(pwd):/data" -i -t vectoriser ./create_tilesets.sh

This will output .mbtileset files to the local ./tilesets directory

Serve tilesets

Run docker-compose up in the project root dir. This will start up 4 containers each serving a different .mbtileset

run web demo

Serve the www directory e.g. using Python in another terminal

cd www
python -m SimpleHTTPServer

Browse to localhost:8000

Note on vector support in JS Mapping APIs

An examination of what different mapping apis are capable of doing with vector layers is still to be completed. For demo purposes however, you will see that

• Mapbox works very well
• OpenLayers (versions 2 and 4) both load tiles as rasters

Next Steps

• Implement a workflow for exporting .mbtileset -> disk. Once objects are on disk and are servable via HTTP, then logically, they can be served from an S3 bucket.
• Explore more JS options

Understand

Some Mapbox tools are used to produce this demo. While Mapbox.com is a subscription-based service (actually the free-tier is great), most of their suite of tools/specifications are open source https://www.mapbox.com/about/open/

Styles

• The style documents in www/styles conform to the Mapbox Style Specification
• These were drafted using Mapbox Studio but can be easily produced/edited manually

MBTiles

• The files output by the Vectoriser container are SQLite files.

TODO can always add more info about how it all works but let me know if there's anything that I can elaborate on.