On 8/5/11 15:00 , Niclas Hedhman wrote:
Well, so why is this different from an explicit UnitOfWork.dropLoadedEntities(), which is what I suggested? You objected to that, since any entities held in application code would become unusable.
Yes, something like that suggests that you can do it as part of normal UoW usage. To me, the pause/suspend suggests that it is not part of a "flow" of usage, so to speak, hence it would be more natural to demand that any entity instances held are released. But, it is very very subjective, either way.
What happens to entities that are held by application code and part of an UoW that has been suspended for a while (but not when the Entity is accessed) and the prune operation is not enabled? In my book, they must ALSO be killed, as otherwise the semantics are too fluffy IMNSHO. Whereas an explicit dropLoadedEntities() is far more obvious.
Maybe. I don't know. I think it's hard to argue either way about what is "obvious" with these things. Maybe an explicit name is good enough.
Again, this whole thing is due to workaround UoW's that are "large", but they don't necessarily have to be so. Falko, what do you think of the idea of using one UoW for tracking changes, and another for "view data"?
/Rickard _______________________________________________ qi4j-dev mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/qi4j-dev
