The current database schema uses a type based table, with id, version, order, xml as columns. The orginal editor software also completely denormalised the individual attributes as unique columns.
I think we must consider if the conversion should use a single object table approach, using class, id, version, order, xml. This would allow joins between tables for example embedded and referencing, and also allows to build a graph. The downside is an even more complex compound key, and a single table may not have the same performance as the table per object type.
The current database schema uses a type based table, with id, version, order, xml as columns. The orginal editor software also completely denormalised the individual attributes as unique columns.
I think we must consider if the conversion should use a single object table approach, using class, id, version, order, xml. This would allow joins between tables for example embedded and referencing, and also allows to build a graph. The downside is an even more complex compound key, and a single table may not have the same performance as the table per object type.