[OSM-dev] API 0.5 is on the way

Martijn van Oosterhout kleptog at gmail.com
Sat Sep 15 19:24:58 BST 2007

On 9/15/07, michael J <michael_j at email.de> wrote:
> Independently from the issue above I have below an example, how the DTD
> could be extended to cover the complete structure of the payload too.
> It also help defining the "Advanced Object Access Methods".

I don't understand. The output is described here:

You don't need to introduce any more elements since the osm tag can
already contain arbitrarily many of them.

> However, if used it has the draw back that we would need to introduce
> another child element node below the OSM root node. As it would require
> code changes, I am interested to hear if such code changes  are useful
> enough to get into the code basis.

If you mean to have an actual description of each specific result. ie.
that the get_nodes requests can only produce nodes, then I don't see
the point. Adding new elements would cause a lot of pain for no gain.
Every program would have to be taught to ignore these new elements. We
have a single DTD that describes all the possible results, I think
that's good enough.

Note: it's possible that more advanced structure languages, like XML
Schema or RelaxNG could do all this without the hassle, but I think
DTDs are too limited for this.

Have a nice day,
Martijn van Oosterhout <kleptog at gmail.com> http://svana.org/kleptog/

More information about the dev mailing list