Uhoh, someone acquired a patent on o/r mapping... but... that's an age old invention...

The text is also very vague, everything is possible and can only include a fraction of what they claim and it still falls under the patent. Surree...

One word - Ludicrous.

The guy who acquired a patent forget to rename namespaces in XML from "http://www.microsoft..." to something else.

I think that the patent is not about "simple" O/R mapping: the keyword in the title is "arbitrary". What they describe - without pronouncing the magic word "morphism" (as in group theory) - could be summed up as follows: a query language can be decomposed into atomic operators f1, f2, ... fn (like scan, project, join). A mapping schema can be viewed as a function M such that M(f1 o f2 o ... fn) = M(f1) o M(f2) o ... M(fn). This means that an "arbitrary" mapping schema can be described out of atomic operator mappings:
f1' = M(f1), f2' = M(f2), ... fn' = M(fn)

IOW, to "scan" over the object domain, you just have to get a SQL view over the relational domain. More complex queries over the object domain can then be composed from the SQL views: if you want to "filter" a "scan", i.e. M(filter o scan) = M(filter) o M(scan), you just have to know how to scan and filter in the relational domain, then combine the results in the object domain.

This principle is illustrated in http://eldorado-net.sourceforge.net/