Hello William
> One challenge is that we still have no consensus on how to express
> this type of data and these operations in RDF, but that's a separate
> issue from support in the database itself.
IMHO, the lack of consensus of candidate users is a significant problem,
and the problem is "three-dimensional" itself.
1. What types of spatial things should be supported? Answers vary from
"lines and polygons" to "all 41 ESRI types of objects, and the
impementation of a subset is as useless as no objects at all".
2. What operations should be supported? Basic "In/NotIn", an
intersection with 9-bit config ("exterior/border/interior of A
intersects with exterior/border/interior of B"), distances, areas,
statistical calculations on fields of values over objects. anything
else?
3. What sorts of geometry should be supported? Rectangular, spherical
without polar areas, fully spherical, WGS-84 only or with arbitrary
conversions?
The problem is that costs of features are not additive, the total cost
of the implementation is rather a product of three sums :| So the list
of the needed features is much more important than the specific RDF
format.
Best Regards,
Ivan Mikhailov
OpenLink Software
http://virtuoso.openlinksw.com
