GML just makes sense; that’s always been my view. In the ten years I worked at the UK’s national mapping agency, Ordnance Survey, we were always big fans of GML. During this time I was very fortunate to provide technical leadership on a project that eventually resulted in OS MasterMap, the UK’s large-scale GML dataset - all 450 million features of it. The decision to choose GML as the supply format gave rise to a lot of head scratching within the UK GIS industry. So much so that in 2001 I decided to leave Ordnance Survey and, along with one of my colleagues, Eddie Curtis, found my own company, Snowflake Software (see Company’s View, GIM International January 2008).
Back in 2001 we believed that GML would rapidly spread throughout the UK and across Europe. After all, GML is only XML, the language of IT. Rather than create a product only for OS MasterMap GML, Snowflake developed a suite of generic tools for loading and publishing any GML 2 or GML 3 dataset. Looking back, we were a little over-enthusiastic regarding the rollout of GML. Despite the pace at which the IT industry adopted XML, the GI industry dug in with its propriety GIS formats and it was not until GML became an ISO standard (ISO 19136) that we started to see widespread acceptance.
To date there isn’t a national mapping agency in Europe that doesn’t have a GML product on the market or in the pipeline. At the moment I’m spending a lot of time in the Netherlands, primarily due to the large numbers of GML schemas being developed under the NEN3610 standard. It’s a great approach and has been very successful for Geonovum (the National Spatial Data Infrastructure executive committee in the Netherlands), which is driving the concept forward. I like to think of NEN3610 as a GML toolbox. At its core is the GeoObject, an abstract GML feature-type that defines concepts of identity, versioning and history. Implementers using NEN3610 then extend this core GeoObject to create their own feature-types that reuse common GeoObject principles. This not only gives implementers a head start in developing their own application schema; it also delivers a consistent approach across GML application schemas.
GML Makes Sense
Many of the schemas developed under the NEN3610 umbrella are being used to facilitate shared services and “joined up” government-infrastructure projects involving hundreds of organisations sharing and supplying data over the internet. To achieve this goal, these projects are implemented using proven off-the-shelf, mainstream IT technologies such as SOAP web services, for which the language of choice is XML. In the GI industry we like to think that spatial data is special, but the reality is that it’s just another data type. The important thing to remember about GML is that it’s not special; it’s just an extension of XML, the language of web services. Given this, and the drive towards “joined up” web services, GML just makes sense!