The indexes being built in the kernel which map a journal entry to the rolons which it effected is a two-way mapping, so from a timestamp you can determine the effected rolons and from a uuid you can determine the effecting journal entries. These indexes work only with timestamps and UUIDs, so the effected rolons don't actually need to be present in the database. So we can have virtual rolons. :-)
Journal entry factories are now virtual rolons, and their class names are used as UUIDs. And when a transaction is processed, the journal entry factory rolon is indexed as being effected by the journal entry which it created. So, given the class name of a journal entry factory, you can retrieve all the journal entries created by it. This is an important capability and will be used extensively in the application layer. And it is much lighter weight than what was done in AW4.
Bill