Development

Data experts can now become mapmakers without picking up GIS expertise first. This session will walk through how to leverage emerging open source tools to design highly custom maps for use online and on mobile devices - without GIS experience.

This session will start by looking at the current ecosystem of open source mapping technologies and the possibilities with the growing open data movement and market demand for location awareness. Then it will look at one tool specifically - TileMill, an open source map editor that makes it easy to design and generate beautiful custom map tiles for users with basic web HTML/CSS fluency.

This session will walk through how to use TileMill to create custom maps, starting from combining spreadsheet data with open data shapefiles to editing the map design using Carto, a CSS like language to adding new data overlay using your own or open data sets. It will then show how to take a custom interactive map, host it, and use it either on a website or on a mobile device. By the end of the session, participants will have seen at least one map created from scratch using TileMill in use on a website and ipad.

This session will also discuss the open source technologies at TileMill's core - Mapnik, which renders the map, and Carto, the map styling language. These projects and others are the basis of TileMill, so the presentation will also focus on their individual characteristics as well as how they fit into a user friendly package. The centerpiece, however, will be teaching participants how to use the tools to create and use their own custom maps.

The current open source tile cache implementations all use the file system for managing tiles. Each tile is a single file stored into a common directory hierarchy. It is a pragmatic approach, easy to implement and with good performance, but it is limited. Managing millions of small files is inefficient with most file systems. The time to remove tile sets, calculate cache sizes or sync caches between differens hosts quickly becomes large enough to be impractical.

This talk will show other existing and new approaches. It will present benchmarks comparing the existing file system based approach with prototypes based on SQLite and other storages as a backend.

It will also discuss the requirements for a new tile store format: What are possible use cases? How could a format be extended? Can we share a format/implementation between tile servers?

The case is that we have tables with GB's of data. This tables are updated regularly by different users and applications. Mapserver is used for WMS and is connected to Oracle.

We can not change all the applications to use postgresSQL and we can not except all user's to switch from Oracle to postgresSQL overnight.

One place to start would be to use mapserver against Postgis. To do this we only need to make a copy of the data from Oracle to PostgreSQL. The Oracle database will master and PostgreSQL the slave.

In some cases it's easy to copy data from Oracle to PostgreSQL. For instance if we have small tables and non frequent changes.

But it this case we have big tables (More 10 GB and millions of rows pr table) and some of them changes many times a week.

In our case we need a simple program that can find all the differences and transfer this changes to PostgresSQL while both databases are in use.

We have made such program that handles this case. We want describe the algorithm behind this program and the code.

Key words here are :

• Split in small chunks by using grids.
• Keep track of neighbors
• Remove all equals rows from work mem.
• Push update, delete and read in single SQL transaction using Hibernate Spatial.