Here is what we are trying out:
- Make JDBCDataStore2 final, configured by a Stratagy object
- One Stratagy Objects to Rule them all: make an Object for "Table", "View" removing conditional code based on case by case GeometryColumn use, FID mapping stratagy etc... If we need PostGIS version specific behaviour it will also be handled in this fashion. This approach was called TypeRef and comes from the AbstractDataStore2 effort.
- ResultSet: rather then build on layers and layers of abstractions (FeaureSource < FeatureCollection < FeatureReader < AttributeIO < QueryData < ResultSet < Statement < Connection) we are going to get as close to ResultSet as we can; making use of paged or random access abilities as needed to directly implement a FeatureCollection.
- SQLEncoder based on Filter 1.1 specification (combines rolls of SQLEncoder and SQLBuilder - basically splits the Filter and builds in a single pass)
Gold Star Quality Assurance Check
There are also tips and tricks that should be noted:
- the createSchema method does not yet work
- a hsql plugin will be needed before content can be created
If you are a volunteer we could use help improving the following: