This is sort of a follow up on my post earlier post Traditional and modern approaches to GIS – short summary of FOSS4G 2011. I will try to sum up some of the projects I thought was most interesting from the sessions I went to.

On the conservative side, GeoTools is getting its own scripting API, GeoScript, which makes much of the GeoTools functionality available for Javascript, Python, Scala and Groovy. From the demo, it looks like a productive way to experiment with geometries (GeoScript ships with an integrated viewer), and maybe a way to generate SLDs programatically. Along the same lines, GeoTool’s Java API has been overhauled, so some of the powerful functionality can now be accessed in very few lines of code. Not everything must be related to OGC standards and XML, and that’s a great thing.

For what I call the modern approach, there’s just too much to talk about them all, but I’ll try to summarize some of the projects that I really want to look closer at.

Tilemill is a web application for designing web maps – basically, it lets you work out a design for your vector and raster data. This design is used to render the actual tiles using Mapnik. In contrast to all map design tools I’ve seen before, the focus in Tilemill is designing for the web - other tools I’ve seen have not been suitable for styling huge datasets and multiple zoom levels. Tilemill doesn’t use SLD, but uses Carto, a CSS like styling language. After seeing some of the designs AJ Ashton from MapBox has done in Tilemill, I’m convinced this is something we will have to try out.

For tiling, a lot of alternatives to GeoWebCache have been mentioned – I have no specifics on them, but we will check them out: TileStache (used by Tilemill, as I understand it), TileCache, MapProxy, MapCache. In the same area there’s also TileStream, a service that hosts and serves your tiles.

Two projects from Vizzuality with the Carto prefix seems really interesting. They’re building a stack with PostGIS as spatial database, which they have packaged as a cloud service called CartoDB, that can be accessed through a HTTP API. For the server application, they have CartoSet, a Ruby on Rails application available on Github. See UNESCOplaces.org for a neat example of what that might look like.

That was a few of my favourites from what I’ve seen. I hope to dig deeper into them, and perhaps some of the other things we’ve seen in future posts.

We’ve reached the end of FOSS4G 2011 in Denver, and I’m going to write down some thoughts after three days of sessions.

I’m going to try to do this without turning this post into a rant about some of the more traditional GIS software out there. It might be that I’m not originally coming from a GIS background, but digging through hundreds of lines of verbose XML config files is not really my idea of fun, and don’t get me started on SLDs. (<ogc:IsThisTagNameTooLongForYou>? Yes it is.). Don’t get me wrong, configuring and coding against GeoServer/GeoWebCache is way more productive than working with some of the legacy products we traditionally used, but it still feels too complicated.

The division between traditional and modern is hardly specific to GIS, but rather something we see all across the board: traditional, big, monolithic chunks of software, expecting your full attention and demanding complex setup and configuration to get started, against the new philosophy of sane defaults without configuration, the simplest thing that could possibly work, and customization through extensibility and integration. Compiled, strictly typed languages against the scripted, untyped ones. RDBMs against NoSql, and so on.

Since much of my prior contact with Open Source GIS has been with OGC standards, WMS, WFS, SLD and the software on the more traditional end of the scale – GeoServer, GeoWebCache, GeoTools – I had the picture of open source GIS as unnecessarily heavyweight and complex for many uses. (I don’t mean to be overly critic of the mentioned softwares – they are truely amazing tools, and great achievements in open source, my critique is more about how they and their APIs are packaged.)

From the projects being presented at FOSS4G 2011, there’s a huge push for the lightweight or modern approach. Every other session is talking about scripting, using Node.js, NoSql databases and it appears that even the core developers of GeoTools/Server are getting fed up with SLDs. That’s great news.

I’m going to follow up this with a post with specifics about some of the new projects I’ve run into during the conference.