2014-05-14 4:51 GMT+02:00 Fernando Trebien <fernando.treb...@gmail.com>:

> For a long time I believed that the
> only practical reason for placing "capital=yes" or "state_capital=yes"
> on a node was to help the renderer decide how to render the label; the
> renderer could then avoid the trouble of handling relations, even
> though it's not thaaaat hard to handle them, and a good renderer would
> have to handle them to support multipolygons anyway.
>


It is a misconception to think either you support "relations" or you don't.
The relation concept is very flexible and open, allowing to map all kind of
relations between the member objects, which themselves could be other
relations etc., this means you'll have to write different code for every
type of relation (and often also for different ways of mapping the with the
same type of relation, think nested relations for instance). If you support
multipolygon relations (which are e.g. supported by osm2pgsql) this doesn't
give you automagically support for admin_centre roles as well. I agree that
it seems not too hard to support this particular role, but so far it simply
isn't there (AFAIK, maybe you can do it with lua?). "capital=yes" btw. is
the older concept, the admin_centre role and even the boundary relation
itself (of not the datatype relation per se) are more recent.

cheers,
Martin
_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Reply via email to