We are migrating to a differenting Oracle database that is supposed to be identical to the old one, but after using Data Pump to move a schema containing an OLAP cube the cube view has several fields defined as VARCHAR2(60 BYTE ). On the old system they have VARCHAR2(60 CHAR ).
The odd thing is this is happening only with the cube view, no other tables or views have this issue after Data Pump.
I also tried exporting to an EIF file from the database where it is a CHAR and importing to the new database, but the view fields are still created with a datatype of BYTE.
Any thoughts would be greatly appreciated!
Edited by: Dave O on Dec 18, 2012 4:53 PM
You may be a victim of this bug Bug 10629770 - KEYWORD "CHAR" INDICATING CHAR SEMANTICS IS MISSING FROM OUTPUT DATA TYPE STRING
When an attribute is specified with CHAR or VARCHAR2 data type having char
semantics, e.g., CHAR(3 CHAR) or VARCHAR2(60 CHAR), the output data type
string for the attribute in the exported xml doesn't contain the CHAR
Without "CHAR" keyword, the RDBMS will assume the default "BYTE" semantics.
Datapump and EIF both internally generate XML, which is then imported into the new database.