Insofar as KNN is a nuanced approach to these problems, forcing KNN to run on all the records in a dataset takes what I like to call a venerable and age-old approach. In other words, it requires a bit of a hack.
More on the general solution to using KNN within a function can be found in Alexandre Neto's post on the PostGIS users list at the following link:
http://lists.osgeo.org/pipermail/postgis-users/2012-May/034017.html
In SQL, the typical way to loop is to use a SELECT statement. For our case, we don't have a function that does KNN looping through the records in a table to use; we simply have an operator that allows us to efficiently order our returning records by distance from a given record. The workaround is to write ...
