[geos-devel] Issues with relate not handling GeometryCollections?

[chodgson at refractions.net]

Union semantics are certainly what I (an admitted 2nd-class geometryologist) 
would expect.

Chris

> Good point, though, that the percentage of cases which really are incorrect
> will be very low.  My original question was "What is the appropriate
> semantics?"  If there is a strong answer of "union semantics", then it may be
> worth implementing this even if a few times it's going to be wrong.