There is a limitation when using Japanese character sets in Replication Server. Neither the eucjis nor the sjis character set can be converted; this issue affects both Adaptive Server and Open Client™ and Open Server libraries.
In general, Japanese character sets are compatible. However, Hankaku Katakana characters, although they exist in both the eucjis and sjis character sets, cannot be converted. Converting data that contains Hankaku Katakana characters between eucjis and sjis does not work. This conversion problem occurs with character datatypes and the text datatype and is documented in the Adaptive Server Enterprise System Administration Guide Volume 1 > Configuring Client/Server Character Set Conversions.
This conversion problem affects both Adaptive Server and the Sybase Open Client and Open Server libraries. Because Replication Server uses these libraries for all conversions, this problem also affects Replication Server.
In Replication Server, this type of failure is treated in the same way as is the case of a single character missing from the target character set. The remainder of the conversion succeeds and replication proceeds, and problem characters are replaced by question marks in the target data area. There is currently no way to escape this restriction with the Sybase connectivity libraries. However, in Adaptive Server, if you turn on trace flag number 2402, you can remove this restriction.
Generally, Sybase recommends that you set up your replication system so that Replication Server handles all character set conversions at the replicate Replication Server and prevents the replicate data server from performing any conversions. In this case, you can work around the Hankaku Katakana restriction if you set up your system so that the replicate data server performs the conversion.
Primary Replication Server |
sjis |
Replicate Replication Server |
sjis |
Primary data server |
sjis |
Replicate data server |
eucjis |
The primary and replicate Replication Servers are configured to use the same character set as the primary data server. (If only one Replication Server manages the primary and replicate data servers, configure it with the character set of the primary data server.)
In this configuration, when the replicate Replication Server connects to the replicate data server with character set sjis, the replicate data server detects this condition and converts data into its own character set, eucjis. If trace flag 2402 is activated in the replicate data server, then the conversion includes the Hankaku Katakana characters.
If you modify the common.loc file to change the default date format for a given language, make the corresponding change to the syslanguages table on all affected Adaptive Servers.