Hi, > Ok, I just implemented this and tested for back compat and it's working *in > the cases I have now*, i haven't dobne a complete test because I don't even > know all the possibilities of the configuration file. I implemented only the > cases I mentioned on the other post I referred earlier. > > I changed code in DatabasePersistenceManager and DatabaseFileSystem and on > the corresponding DDL. > > Should I post the code in here? Or in JIRA? Or not at all?
Could you please post your changes as an attachment in the JIRA issue? https://issues.apache.org/jira/browse/JCR-1180 It would be great if it's in the form of a 'patch'. If this is too complicated, just the source code is OK. > It is somewhat a ugly solution, thus... I agree. In my view this is an ugly hack. It will fail with quoted identifiers: Any database identifier can contain spaces and dots if it's quoted. Correct parsing and splitting the schemaObjectPrefix would be really ugly and database dependent (MS SQL Server supports [] quotes, MySQL backticks; but most database use double quotes, which need to be escaped in XML). I think it's better to use a distinct case sensitive property 'schemaName' as in the java.sql.DatabaseMetaData.getColumns and so forth. In my view, the schemaObjectPrefix should be kept as is (tableNamePrefix would actually be the right name). Regards, Thomas
