The Open Geospatial Consortium (OGC) has published a draft GeoPackage specification for comment. The GeoPackage specification attempts to create a non-proprietary means for packaging and exchanging all geospatial data in all its forms (vector, raster, and tiles). A couple of things that jump out at me:
- It calls out SQLite as the reference implementation of a GeoPackage container
- It calls out SpatiaLite 4 as the reference implementation of a vector feature store
- It does not call out a reference implementation for rasters or tiles
- It does not mention exchange of cartography.
Although the draft references the MBTiles specification and uses it as an example of a pure SQL method of storing tiles, it does not call out MBTiles as the chosen storage approach nor does it mention MBTiles as a reference implementation. The draft should not be read as “SpatiaLite + MBTiles.”
I think, in 2013, anything that attempts to do what GeoPackage sets out to do should have some provision for cartography. I would prefer CSS-like approach (such as CartoCSS) and will say so when I submit comments but I would suspect that future discussion of cartography, if any, will probably start with SLD. For now, tiles can fill the gap.
So comments are open, give it a look and tell OGC what you think.
GeoPackage sounds very promising.
IMO whatever provision for cartography should be *optional*, since so many different products have their own opinions about how spatial data should be rendered/stylized. Some don’t even currently support SLD. It would be a tall order for such products to support GeoPackage if you mandate that they must support SLD as well.
I agree that it should be optional. Really, all individual component should be optional since you should only ship what you need to ship. In truth, I’d rather it not be SLD.