Chinese data problem

We have a customer who has a 8.1.7.3 database created with a db character set of WE8ISO8859P1. They have been using this database for a few years and while the data in there is 80% english they also have been putting Chinese (HK - traditional characters) into this database. Thing is, it works. They are able to enter data and view it through our application even though we do not specifically support this. The data is in regular VARCHAR2 columns, not NVARCHAR2.

I now have to upgrade them to a new version of our product which only runs on Oracle9i (9.2.0.4). I have used import/export to create a new database (with the same character set WE8ISO8859P1) - the English data is fine, but the chinese characters in 9i database have become corrupted in some places (about 60% of the time)- they are still chinese characters, but do not always match the characters that were in the original 8i database!

An example is a character which in the 8i database is made up of two bytes: Ascii 164 and 233. This character displays incorrectly when I select it from the 9i database. The strange thing is, even if I create a database link from the 9i to the 8i database and select it and get the ascii values of the two bytes, they have turned into 191 and 233.

It's obvious that the customer is in the wrong for using chinese data with a character set which doesn't support it - and frankly it's a miracle it works for them at all - but does anyone know why the ascii values are changing when selected from the 9i database? Is my only hope to recreate the 9i database in a character set which supports chinese? (would an import even work correctly for the chinese data in this case?)

Thanks in advance

Daniel Received on Sun May 22 2005 - 23:42:14 CDT