classes which query an in-memory set of datasets is not that easy as it might seem: there is no query engine. This means that I have to do the query execution too. But perhaps there is another way, I have to look into this.
It is a complex problem. I can create a caching mechanism which will intercept retrievals of data and which will then return in-memory (cached) objects. This also will not be as fast as a database, because it will of course not be a real RDBMS with indexes etc. PK selects will be fast, filters will not be that fast when the data is huge.
I'll look into both solutions and see which one is better for your and others situation, when you have a slow db connection and a lot of read-only data: the only situation caching will not be harmful and the only situation caching can help.
It will be in the first half of november when this will/can be realized. In early november a new template set is released which uses separate DAO objects for persistence, and then a cache can be included more easily (otherwise you have to specify for each entity when and how it will be cached which is cumbersome).