Description Nicolas Bros 2010-08-17 08:53:31 EDT

At the moment, to provide an implementation of AbstractModelQuery you must implement basicEvaluate, which takes a single context element.
When evaluating a query on several context elements, basicEvaluate is called repeatedly with each context element.
For some implementations that make use of a system with which you must establish a connection (eg a database) for each call, this is costly.

To solve this, we can add 2 methods: startEvaluate and endEvaluate in AbstractModelQuery, so that the implementation can be optimized.