In general I would not recommend trying to use a ROWID as a primary key. Although I'm not sure why your example is not working, it seems JDBC does not like to ROWID being bound back. You may wish to experiment with raw JDBC to find something that will work, then try configuring that in EclipseLink. You may be able to map to the attributes of the XML document and declare one of those as the primary key. You could also try defining the XMLType as the primary key, although you would still not be able to update it.
EclipseLink does support the Oracle XMLType, but only as a column type, not as a table. You may wish to log this enhancement request. Do you intend to map the XML to an object, or do you just want the XML? I would recommend you add a normal table that has a generated sequence id and a column for the XMLType.
dmitryerkin wrote:
Hi
I mean can I use a ROWID column as a primary key instead of real column which is sequenced in case I have not a primary key column in my table?
I work with xmldb object tables so I can not add extra columns.
Is it possible to retrieve into a bean's attribute the value of ROWID for the record which was inserted while executing of the EntityManager's flush method?
I use Oracle DB of course.
Maybe I can implement a custom stored procedure which returns a ROWID value ane set this procedure by descriptor.getQueryManager().setInsertCall?
Regards
Dmitry
