I guess points 1 & 2 are common through out the Open Source world. I think that people traditionally have their GI data in .shp,.tab or RDBMS formats. There are a few steps that have to taken to feed this data out via an open source solution and this is perceived as a problem

Just my two cents:)

That said, I do believe that there is untapped potential in the spatially-enabled relational databases on the market that could provide the power behind a fairly lightweight Web-based data editing frontend. Similarly, spatial databases have a huge potential to provide analytical capabilities that are currently locked up in heavy middle- or client-tier applications.

Regarding the connectivity between Arc* datasources and Open Source geospatial tools…. I believe that FDO, used in MapGuide Open Source (and others?), can provide direct connectivity to SDE.

poke around on the GRASS and GMT mailing lists a bit for an idea of what people who are new to FOSS/GIS are complaining about. I think that you have identified one of the major points: hard-copy map production, but this can be a bit misleading. People producing small-scale maps (scale is a fraction, i.e. 1:1,000,000 scale maps are small scale) have been using open source tools for a long time now: nearly all of the figures in any publication of AGU (http://www.agu.org/) look as if they were produced with GMT (http://gmt.soest.hawaii.edu/). It is mostly the people wanting to make large-scale maps that are having a tough time. there should be some solutions in the near future, as much discussion on the GRASS-DEV mailing list has covered postscript output from the display drivers, along with better integration with GMT.

Dylan

ESRI has done a great job of capturing the traditional GIS market that communicated with the rest of the world largely through a paper map. Advanced cartographic functionality is quite difficult, and the ability to create beautiful maps is one I tip my cap to.

However in 2007, it’s clear that the web is/will shortly be the dominant communications medium for geospatial information, and I think it’s no coincidence that the momentum for open source has coincided with the explosion in web mapping. Suddenly, powerful server-side software e.g. MapServer, PostGIS can be harnessed to web front ends as not merely providers of raster and vector content but, with a little scripting, analytics as well (that used to be the sole domain of the desktop). Bringing geoanalytics to a wider audience in easy-to-use bite-sized chunks is a degrees-of-magnitude greater opportunity for OS than implementing a complex feature label algorithm in order to keep up with ArcMap.

Finally, it bears remembering that the most valuable commodity in the world today is time. Why do I use MapServer? Because it takes 25 minutes to set up and serve content. Why OpenLayers? Because I can roll out a web map with multiple layers from multiple sources in 10 minutes with a layer-toggling and overhead map. Etc., etc. And GRASS? I’d love to feed a server module a bunch of points and attribute values and get back a simple IDW grid on the fly that I could throw on to a web interface–any takers?

In short, keeping the focus on the web, usability, and helping the not-overly-technical use these tools is where the big growth is.

Brian Timoney

Open Source tools often fail to provide a compelling pathway from using ESRI tools to using them. Some projects (ZigGIS comes to mind) are addressing this need, but other seem to ignore the user-base numbers. Perhaps OS GIS needs a single-install ArcGIS plug-in bundle for subverting the ESRI community.

One of the weak points in the OS GIS offerings is the ability to easily compose a good looking map, complete with all the trappings that map makers adore. As that capability matures, more heads will turn towards OS.