[OSM-dev] .osm tile naming convention

Robert Scott lists at humanleg.org.uk
Sat Jun 5 01:14:03 BST 2010


On Friday 04 June 2010, Frederik Ramm wrote:
> Richard Fairhurst wrote:
> > It's not really suitable for vector data, though. Firstly, the concept of
> > zoom levels doesn't apply - the vector data is the same at every zoom.
> 
> Of course; the zoom level would just be a parameter for the conversion 
> of x/y to minlat/minlon maxlat/maxlon.
> 
> I'd think that a sophisticated server could make tiles of all sizes 
> available - i.e. if you request something at z10 then you get a large 
> area, if you request at z18 you get a small area. A server that has only 
> prepared tiles for download would only carry one fixed zoom level.

This is exactly the approach I was considering taking recently, only I was going to use geojson files instead.

In the end I decided it would be quicker to write a dirty nasty WFS implementation than to write a new protocol for openlayers vector layers.

The tile-style scheme appealed to me because I could have had different level of detail features at different zoom levels. This would have worked well for geojson's (traditional gis) data model but I'm unsure how well it would adapt to OSM's data model.

It would indeed be very useful to have a mechanism for statically serving feature data. Not everyone has access to good dynamic hosting.


robert.




More information about the dev mailing list