Currently, filters operating on nested attributes (i.e. attributes of features that are joined to the queried type via feature chaining) are evaluated in memory, after all available features have been loaded. Translating these filters to SQL and thus doing the filtering directly in the database backend could bring significant performance improvements, especially when the total number of features in the database is high, but only a few of them would satisfy the filter. Related mailing list discussion: http://osgeo-org.1560.x6.nabble.com/App-Schema-Encoding-of-Filters-on-Nested-Attributes-td5285165.html |