Interview With Tyler Mitchell of OSGeo

Linux Journal's James Gray recently spoke with Tyler Mitchell, Executive Director of OSGeo, about his organization's efforts to promote open geospatial technologies.

OSGeo, short for the Open Source Geospatial Foundation, is a "not-for-profit organization whose mission is to support and promote the collaborative development of open geospatial technologies and data."



Linux Journal: Hello, Tyler, and thanks for agreeing to speak with us about OSGeo, the Open Source Geospatial Foundation. I am a user of proprietary GIS (Geographic Information Systems) user who is just starting to explore the wonderful world of open-source geospatial technologies, such as GIS, Web mapping, geospatial libraries and others. I was fully impressed to see the sophistication of your work at OSGeo.

First of all, I see that OSGeo's mission is to "support the development of open source geospatial software, and promote its widespread use." What kinds of applications fall under the rubric of "geospatial software"?

Tyler Mitchell: When we talk about geospatial software we generally mean any software that interacts with geographic information. This could be a tool that helps you draw maps or it could do more advanced work like analysing spatial scenarios and building more software. This software is also often used to provide various real-time services such as data sharing or remote processing.

OSGeo software is grouped into four general categories, which represents the broader realm of geospatial applications as well. These categories include web-based mapping, desktop applications, programming libraries and metadata catalogues. The other major area of development is around spatially-aware databases, though OSGeo does not represent any database platforms directly at this time.

These categories parallel kinds of general open source applications that allow users to communicate, write programs, find datasets, visualise information, etc. Except here the primary focus is on spatial data that describes the geography and attributes of the real world.

LJ: In what sorts of ways do users and developers find themselves working with geospatial software? And how does the open source aspect affect the way the software is used and perceived?

Mitchell: The nature of geospatial software requires it to have several levels of interaction -- such as data access and sharing services between applications. So an enduser might be merely using a desktop application to view information, but behind the scenes, common data access APIs are supporting the user's interaction. As well, like many Web-based applications, there are several layers -- data access, Web
server integration, user-interface presentation and interaction, etc. Geospatial software is crosses many of the similar barriers, like the many other types of networking applications.

When we talk about open source geospatial technology, we are usually talking about freedom to choose things such as -- target operating system, scripting or programming languages, target browsers, data formats required, analytical tools required, etc. Because applications can interact at several different levels, it is possible to build a stack of geospatial technology that uniquely meets your given requirements. This includes the common need to interoperate with proprietary platforms at a data or services level. By having
these choices, a development team, integrator or user can choose to focus on future goals instead of ongoing limitations. Just because these are open source doesn't mean they are without limitations, but the possibility of overcoming the limitations is very real.

LJ: Could you please provide us with an overview of the state of development of open-source geospatial applications vs. proprietary ones?

Mitchell: This is always a hot topic, especially when you dig into the motivation and purpose of various projects. Some aim to replace proprietary products while others are simply best-of-breed and stand on their own. Regardless of the perspective that a project takes, the reality is that open source projects are increasingly being used behind proprietary applications, not only in place of them. For many, there also remains a need to interoperate with existing proprietary platforms and services.

Proprietary solution providers are increasingly required to consider whether or not they can justify building a new tool, from scratch, while ignoring the open source options that are available -- especially if the tool is mature and well supported already. We are not talking about simple things like writing a parser or manipulating some text or graphic files. With geospatial applications we are talking about some areas that are highly mathematical or time-intensive to implement. Such things as representing complex vector or raster graphics, transforming coordinate data between coordinate systems, reading and writing obscure data formats, high-performance user interfaces, etc.

One good example is the GDAL/OGR data translation library project. Many end users will never know it exists, or why it is even helpful, yet it sits behind some of the most popular proprietary products,
such as ArcGIS and Google Earth. Of course anyone who uses a Web mapping application may also be benefiting from GDAL/OGR but you might never know.

Open source geospatial applications are also helping fuel various free-to-use Web-based systems. This presents challenges similar to proprietary software development, yet because the software is hidden
on the server, you may never truly know what lies behind it. Just as open source developers tend toward focusing on providing services, rather than trying to sell software products, so it seems with the
proprietary developers using the Web. Development trends in proprietary services, as well as products, is well worth watching. Either way, there is open source uptake behind both.

The current state of development is not easy to define. Some of the open source options are still relatively young, but there is no doubt that several are quite mature and are sought after by users who would traditionally use proprietary products. Web mapping is a good example of where the disruption of open source server products like MapServer, MapGuide, GeoServer and deegree have given more options to
developers and users than had been previously available in proprietary products.

The area of standards is another one that is important to mention, because so many of the open source tools adopt them earlier than proprietary software. Some even become the industry reference implementation of certain specifications put forth by standards bodies such as the Open Geospatial Consortium.

LJ: Let's talk specifically about GIS for a bit. I became familiar with GIS through using ESRI's ArcGIS, which is perhaps the most popular commercial/proprietary GIS package. While ArcGIS's desktop
applications run only on Windows, several back-end applications run on Linux. What are the core open-source GIS applications, and how do they stack up vs. ArcGIS?

Mitchell: I can't speak about the current state of their applications and their Linux support, but when I started down the Web mapping road in the late 90's, there wasn't any option. I nervously watched as UNIX support for these applications decreased in favor of Windows NT. For many of us, moving to Linux servers naturally meant moving to open-source GIS. This had the powerful side-effect of helping me see the benefit of server-side applications in an enterprise. Cross-platform availability sure has a lot to do with adoption of OSGeo applications.

The core of the open source GIS applications is likely the handful of APIs available (OSGeo's - GDAL/OGR, GeoTools, GEOS, FDO, OSSIM). They provide, for example, access to data and geometry handling functions. I can say these are the core because they are pervasive, with very few other competitors in both open source and proprietary. So many of the other open source geospatial applications depend on these libraries for their success that if we didn't have them, we'd all be writing our own data access drivers and geometry handling functions in isolation still! Instead we have several mature toolkits to chose from.

The next tier of applications splits into two branches -- desktop and Web centric applications. The Web applications are serving up data, maps and various services. The breadth and stability of these applications is very good. Milestone releases for many applications are regular and only a couple are hovering around the 1.0 mark.

The desktop applications are usually the ones that people moving from a proprietary model struggle with. In some cases the concepts and tools are different enough to scare away users who only know how to operate in a certain paradigm. In other cases the applications may yet be relatively young. Some of the more advanced cartographic functions are usually described as lacking when compared to proprietary systems, but they are the focus of ongoing improvement. That's not to say that applications like Quantum GIS, GRASS GIS and gvSIG are not ready to use - for many users they are ready, but it depends on your requirements.

This past year saw significant improvements across all our projects. They all have some interesting features and continue to capture the attention of many users and developers alike on an increasing number
of platforms.

LJ: If someone wants to get started with open-source GIS, such as QuantumGIS and GRASS, what resources are available to them?

Mitchell: There is the typical open source support options: forums, wikis and/or mailing lists. Several are active in IRC as well. I've seen some screencasts, usually geared more at developers, and of course there are blogs and Web tutorials. We've also started an international Service Provider Directory for users looking for commercial support -- you can query by geography, language and technology.
[Search the Service Provider Directory here: http://www.osgeo.org/search_profile]

One area that has seen a lot of activity this past year is in developing a wide range of introductory articles and case studies from volunteers. We've been publishing them in our OSGeo Journal, along with community reports and news. There are also more general discussion articles that will be of interest to GIS professionals and developers. [Visit OSGeo Journal - http://www.osgeo.org/journal]

There are also some educational institutions that are offering courses or even open course material to help train on these products. Our Education Committee has been pulling together pointers to these resources so they are easier to find. It's still early on in the effort but is a good starting point.
[See http://wiki.osgeo.org/index.php/Educational_Content_Inventory]

LJ: On your Web site, OSGeo is called "a nascent organization". How developed is OSGeo today, what do you consider your biggest accomplishments, and where do you hope to be 2-3 years from now?

Mitchell: We've just celebrated our second anniversary, so I don't think we can claim to be nascent much longer! With each year our community has become stronger and more involved in our mission. One of our biggest accomplishments is the running of our annual conference event. This year we had over 700 participants from around the world meet in Victoria, Canada. It was buzzing with energy and very memorable -- having Damian Conway speak was a real treat.

Of course, we wouldn't have much to talk about if our projects weren't being actively developed. Many of our projects have graduated from our incubation program and others are lined up hoping to join. Several important projects have come to join OSGeo over the past two years and this encouraging trend has continued.

It has also been encouraging to watch people and organizations meet and start working together, including our sponsors. From a wide set of backgrounds they have come together under OSGeo's banner to help further our common mission and that has been very satisfying to see.

LJ: Tell us more about the event you mentioned above, the annual Free and Open Source Software for Geospatial, or FOSS4G, conference. It sounds cool.

Mitchell: The primary purpose of our international event is to introduce users and decision-makers to the many projects, applications, case studies and tools that our communities use. This goes well beyond OSGeo, as
open-source geospatial projects in general are also well represented. Hands-on workshops are a major feature as well - for both new users and seasoned developers. They are always sold out and it is a great way to meet face-to-face with various project teams.

FOSS4G is also geared as a community meeting place for OSGeo and its friends. Those that do not come to learn, come to help and meet others. The networking opportunities are tremendous, especially as lead developers get to meet their users from business, government and other organizations. For the developers themselves we also hold a Code Sprint day were teams can meet to plan and program. All around, there is something for everyone at the event. I hope to see you in Cape Town, South Africa at the end of September!
[FOSS4G's Web site is: http://www.foss4g2008.org]

LJ: Thanks, Tyler! We wish you and your team at OSGeo all the best and keep up the important, fascinating work!

Mitchell: You're welcome!

About Tyler Mitchell
Tyler Mitchell is the Executive Director of the Open Source Geospatial Foundation, a.k.a. OSGeo. He is the author of Web Mapping Illustrated: Open Source GIS Toolkits from O'Reilly Media.

Mitchell has over 10 years of industrial geospatial and GIS experience in
natural resource management and forestry in western Canada. In looking for tools in his career work, Mitchell found that open-source tools could dramatically improve enterprise, corporate-wide, geospatial data management and communication.

Finally, Mitchell is an avid proponent of open source technologies including Linux, Web-based mapping applications such as MapServer, PostgreSQL and its
spatial extension PostGIS, desktop mapping applications such as Quantum GIS and more. His work and interests include geospatial and tabular data management, analysis, manipulation and visualization through maps.

Load Disqus comments