On 8 Oct 2010, at 16:57 , Ivo Ladage-van Doorn wrote: > After analyzing the code of the CassandraPersistenceManager, Angelo and I > discussed the existence of the ColumnFamilyAvailable service. This service is > instantiated automatically when a certain ColumnFamily is available and some > bundle wants to define a service dependency on its availability. This could > be improved however, similar to the keyspace approach already used for the > CassandraPersistenceManager. > A CassandraPersistenceManager instance should not only exist per keyspace but > also per set of ColumnFamily?s (the ColumnFamily?s available in that > keyspace). So each keyspace will have one associated > CassandraPersistenceManager, with service properties that hold the keyspace > and all ColumnFamily?s contained by it. Service dependencies can then be > defined on a CassandraPersistenceManager for not only a certain keyspace but > also for one or more ColumnFamily?s. This would be a better approach. Anyone > that does not agree?
Sounds okay. I would consider using a different name than CassandraPersistenceManager though. Greetings, Marcel -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.amdatu.org/pipermail/amdatu-developers/attachments/20101008/40a6b529/attachment.html
