Using the simple class name would be a likely strategy (as you don't use FQNs either to name your ORM-mapped SQL tables). I wished for that when demoing or even testing, where the FQN is painful. E.g. there could be a pluggable TypeNamingStrategy contract which returns the type name to be used for a given entity. If it doesn't return the FQN (which would be the default impl), we'd store the FQN in _meta. |