Please help - french accents not correctly displaying/storing in database

Question

Hello -
I have 2 databases running in 2 different character sets.
I am able to insert the following french accents in both databases and stored correctly also in db and displayed correctly in sql plus -
à À Ä ä Â â é É È è Ë ë Ê ê î Î ï Ï ç Ç Û û ô æ ö « »
But the following 2 charactes donot get displayed/stored correctly in any of the databases -
1. For db running in UTF8 character set it stores/displays as follows -
2. For db running in WE8ISO8859P1 character set it displays as 2 filled black squares. But when I do 'copy and paste'. It is pasteing correctly in the editor as follows -
I have another one database running in same WE8ISO8859P1 character set , DBA out there sent me the results from sql plus for the above 2 characters and they are getting displayed as well as stored fine.
I have the following questions -
1. Besides database characterset what else parameters or environment variables I need to check the for these characters to be displayed/stored correctly ?
2. Why would Databases with same characterset gives different results for the same characters ?
3. UTF8 is superset of WE8ISO8859P1 characterset, then why in that UTF8 db is unable to store that characters
which were able to be stored in WE8ISO8859P1 db.
By these various results, I am very confused and not able to troubleshoot further.
Any help would be very much appreciated...
Thanks
Rama

Answer 1

Ah... an interesting NLS problem ;)
Main problem or misunderstanding could be that the characters small and capital ligature OE (œ Œ) is not part of the ISO 8859-1 repertoaire. That is, you need another characterset to store and use those characters properly.
However, and sometimes unfortunate, if both db and session is using the same characterset setting, the binary value is "passed through" without conversion and stored - even if actual char is not valid!
1. DB charset should be able to store all possible chars your apps/users need. That's a first. Is that the case?
For sessions, the OS/app charset used should be indicated to Oracle with NLS_LANG parameter (registry or env var).
2. Are you sure that (valid) character codes are stored correctly. Note that storing values directly could work technically but be wrong semantically. Values > 127 are not valid in UTF-8 (needing further explanation)!
3. How are "characters" stored? What does this tell you?
SQL> select dump(column,1016) from dual;
Regards,
Fredrik