Description
Currently, the persistence adapter attached to the broker service is simply ignored by the scheduler. The scheduler always uses KahaDB, instead.
I see two ways to go about this:
- Creating a SchedulerPersistenceAdapter akin to (and possibly extending from) PersistenceAdapter, as well as a corresponding factory class and BrokerService property. This seems clumsy, but is in line with the approach currently taken, separating scheduler-related data from non-scheduler-related data - see BrokerService.setDataDirectoryFile() vs. BrokerService.setSchedulerDirectoryFile(). This approach is probably unnecessary, since the scheduler can clearly use existing PersistenceAdapters (or at least the KahaDB adapeter).
- Depracating or removing the BrokerService.schedulerDirectoryFile property and having the scheduler use the one and only persistence adapter attached to the BrokerService (if it's a journaling adapter - BrokerService.dataDirectoryFile will be used, rather than BrokerService.schedulerDirectoryFile). This seems like the reasonable approach.