Michael Hunger wrote: > Thats really great. Again the benefits of simple and fast blob stores.
Indeed! I have put serialization of non-primitive types into the UoW layer, so the blob is really just a blob. > But as you already said making them indexable is crucial for accessing the > stuff. Agree. This is why it was so important to support ValueComposite for complex objects, so that the indexing has a "known structure" to work with, instead of having to do reflection. > What would be really interesting is using the values for modelling a medium > complex domain. > Especially I'd be interested to see the ease or complexitiy of handling > (building, populating, binding, validating etc.) > the entities with lots of nested values. > > So we're hoping for streamflow to evolve (no commits lately?) to show the > power of this. Yeah, I started doing modeling of persons in StreamFlow, immediately ran into a 3-level value (contact info), and hence had to fix that in Qi4j. This is precisely the kind of need-driven development I had hoped for when I started StreamFlow, and it is already paying off!! Now that Values work, from a modeling and persistence point of view, I want to go back to working on StreamFlow. If I understood things correctly Alin would take a look at indexing of Values. Alin, is that correct? All the metainfo should be there now I think. /Rickard _______________________________________________ qi4j-dev mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/qi4j-dev
