Oracle 8i, MS ODBC for Oracle and Shift JIS

Hi,

  We have SJIS data in a Oracle 8i database whose NLS_LANGUAGE is set to JAPANESE_JAPAN.JA16SJIS. This database is running on a Windows 2000 Pro (Japanese).

  We have a VC++ application which uses the Unicode API provided by ODBC for accessing the Oracle database. We use the MS ODBC for Oracle driver since it is the only driver that works with both MSDTC and ORACLE. Unfortunately MS ODBC for Oracle does not support Unicode.
(Now you know why we are using this weird combination. SJIS +
UNICODE).   According to the following document
(http://otn.oracle.com/docs/tech/windows/odbc/htdocs/817help/sqoraUnicode_Support.htm),
the MS ODBC Driver Manager converts data from the DB char set to Unicode if the application uses the Unicode ODBC API.

  The combination seems to work if the client and server reside on the same machine i.e. the W2KPro (Japanese). If the client is a English machine, it doesn't work, we generally end up with garbage. Some information is given in the above-mentioned document regarding this as well regarding local code page conversion and such.

  The following are my questions.

1. Will the conversion work if the database contains Shift-JIS data and the ODBC application uses the Unicode API? I mean, will the driver manager convert the SJIS data to UTF16 for the application?
   (Minor conversion errors are acceptable) Is this the correct and
   consistent behavior?
2. What will be the kind of changes we need to expect in the table datatypes and application code?
3. What if we need to run the server and client on English versions of W2K servers/professionals?

Regards

zrb Received on Mon Aug 26 2002 - 06:04:30 CDT