I'm currently doing something similar. Our database is USASCII7 and we're now seeing records being loaded with some Western Euro characters (accents and tildes). csscan reports them and the migration utility won't run until they are cleaned up. Especially fun is that some of those chars are in the data dictionary now as part of column histograms (thanks to the default GATHER_STATS_JOB in 10g). Then we plan to migrate to WE8ISO8859P1. Don. On 7/17/07, Peter McLarty <p.mclarty@xxxxxxxxxx> wrote:
Be very careful and definitely run csscan. I worked with a customer in Thailand that had an applicaion where the front end was cobol. The database had USASCII7 as the characterset and for whatever reason the cobol had allowed the use of Thai Characters, which had become stored in the database. Oracle was storing in effect garbage, that only the cobol application could extract, and it was going to take some work to extract that data. It became apparent when the people were testing and ran a couple of queries in sqlplus and got garbasge in product descriptions or details lines. This data couldnt go into a datawharehouse and in fact it was useless in that database to anything except the cobol application. The only solution I could see in this case would have been to extract that data using the cobol and then creating a new database with a suitable characterset and running the data back into it. I only hope its since been fixed but I doubt it.
-- Don Seiler oracle: http://ora.seiler.us ultimate: http://www.mufc.us -- http://www.freelists.org/webpage/oracle-l