Home

	Expand Search

Configuring Adaptive Server Enterprise for UNIX Platforms

Appendix C: Character Sets, Sort Orders, and Message Language

Appendix C

Character Sets

By default, Adaptive Server and Backup Server have character set definition files for the character sets shown in Table C-1.

Table C-1: Platforms and their native character sets

Platform	Native Character Set
Digital UNIX	ISO 8859-1
Digital Open VMS Alpha	ISO 8859-1
HP-UX	Roman8
IBM PC	Code Page 437 Code Page 850
IBM RS/6000	ISO 8859-1
Macintosh	Mac
NCR System 3000	ISO 8859-1
Sun Solaris	ISO 8859-1

These character sets support the Western European languages on the specified platforms. Each platform has a native character set. By default, srvbuild installs Adaptive Server and Backup Server with your platform's native character set.

Selecting the Default Character Set

The default character set is the character set in which data is encoded and stored on the Adaptive Server databases. You can select any character set as the default on your Adaptive Server or Backup Server, including character sets that are not native to the platform. Determine Adaptive Server's default character set, based on the character set used by your client. For example, if most of your clients use Code Page 850, specify Code Page 850 on your server to minimize the amount of code conversion that has to occur.

Conversion Between Character Sets

In a heterogeneous environment, Adaptive Server and Backup Server may need to communicate with clients that run on different platforms using different character sets. To maintain data integrity, the server converts the code between the character sets. To enable automatic code conversion, you need to install character set definition files on the server for all the character sets that your clients are using.

Backup Server returns messages to Adaptive Server in the client's language and Adaptive Server's character set. Adaptive Server then converts the messages and issues them in the client's language and character set.

If Adaptive Server or Backup Server does not support a client's language or character set, it issues a warning. Errors also occur if the Backup Server character set is not compatible with the Adaptive Server character set.

Note: Code conversion is supported between character sets for the same language or group of languages only. For example, code conversion is supported between the character sets for the Western European languages (ISO 8859-1, Code Page 437, Code Page 850, Mac, and Roman 8). Similarly, code conversion is supported between the character sets for the Japanese language (EUC-JIS, Shift-JIS, Dec-Kanji). However, code conversion is not supported between the Western European language character sets and the Japanese character sets. For more information about supported conversions, see the System Administration Guide.

Sort Orders

The sort order determines the collating sequence Adaptive Server uses to order data. Each character set comes with one or more sort orders. Sort orders are located in sort order definition files (.srt files) that accompany your character set definition files.

The available sort orders vary, depending on the available character sets. To see what sort orders are available, refer to the sort order= field in the charset.loc file in $SYBASE/charsets/charset_name for each character set you plan to install.

All character sets have a binary sort order. They may also have one or more of the sort orders listed below:

• Binary sort order sorts all data according to ASCII numeric byte value, is the default sort order. Binary order sorts all uppercase letters before lowercase letters. It is compatible with pre-10.0 SQL Server releases.
• Dictionary order, case sensitive, accent sensitive sorts uppercase letters before their lowercase counterparts. Dictionary order recognizes the various accented forms of a letter and sorts them after the associated unaccented letter.
• Dictionary order, case insensitive, accent sensitive sorts data in dictionary order but does not recognize case differences. Uppercase letters are equivalent to their lowercase counterparts and are intermingled in sorting results.
• Dictionary order, case insensitive, accent sensitive, order with preference does not recognize case difference in determining equivalency of items. A word written with uppercase letters is equivalent to the same word written with lowercase letters. Uppercase and lowercase letters are distinguished only when you use an order by clause. The order by clause sorts uppercase letters before lowercase.

  Do not select "case-insensitive order with preference" unless your installation requires that uppercase letters be sorted before lowercase letters in otherwise equivalent strings for order by clauses. Using "case-insensitive with preference" could cause poor performance in large tables when the columns specified in an order by clause match the key of the table's clustered index.

• Dictionary order, case insensitive, accent insensitive treats accented forms of a letter as equivalent to the associated unaccented letter. It intermingles accented and unaccented letters in sorting results.

By default, each Western European character set that comes with Adaptive Server and Backup Server comes with the following sort orders:

Table C-2: Sort orders for Western European languages

Language	Available Sort Orders
All Western European languages	Binary sort order
English, French, and German	Dictionary order, case sensitive, accent sensitive Dictionary order, case insensitive, accent sensitive Dictionary order, case insensitive, accent insensitive Dictionary order, case insensitive, accent sensitive, order with preference
Spanish	Dictionary order, case sensitive, accent sensitive Dictionary order, case insensitive, accent sensitive Dictionary order, case insensitive, accent insensitive

Selecting the Sort Order

By default, your server is configured with the binary sort order. The server can support only one sort order at a time, so be sure to select a sort order that will work for all of your clients.

Message Language

By default, U.S. English messages are installed as the default language on Adaptive Server. If your clients require software messages in a language other than English, you must purchase and install the Language Module for those languages.

During the installation of Adaptive Server, you specify a default language for Adaptive Server. If messages are not supported in the client's language, they will receive their messages in the default language. For example, if your client's language is Portuguese, a language for which Sybase does not provide translated messages, and the Spanish Language Module is installed, and Spanish is specified as the Adaptive Server default language, the client receives software messages in Spanish.

Japanese Language Module

The Japanese Language Module includes the following multibyte character sets:

• eucjis - Extended UNIX Code for JIS-X0201 and JIS-X0208
• deckanji - DEC-Kanji Code for JIS-X0208
• sjis (Shift-JIS) - IBM/Microsoft Code for JIS-X0201 and JIS-X0208

Hankaku Katakana Conversion and Its Limitations

• Hankaku Katakana (half-width phonetic Japanese) character set conversion between Shift-JIS and EUC-JIS is not currently built into Adaptive Server. However, it is possible to enable Hankaku Katakana character set conversion by turning on trace flag 2402 when starting the Adaptive Server. Include "-T2402" on the command line in the script used to start Adaptive Server.

  Note: Hankaku Katakana conversion is enabled only if Adaptive Server is using EUC-JIS and clients are using Shift-JIS.

• Once conversion is enabled, you cannot change the conversion on a per-client basis. To turn off the conversion, either remove "T2402" from the script or use a different script with the trace flag disabled; then restart Adaptive Server.
• If conversion is disabled for Hankaku Katakana, the characters will be replaced with "?" to indicate unconverted characters.
• The client application using Shift-JIS must recognize different character sizes when defining a table. Data that is longer than the column length is truncated; no warning message appears.
• In an Adaptive Server using EUC-JIS, characters are stored in EUC code set. Zenkaku characters come after Hankaku characters in the EUC code set; hence, the Adaptive Server sorts Hankaku first, followed by Zenkaku.
• If the column contains Hankaku Katakana, the Adaptive Server using EUC-JIS sends the client application using Shift-JIS the converted characters followed by nulls.
• Hankaku Katakana can be used in logical expressions, for example:

  select * from table
  where column="Hankaku_Katakana_here" 

• The Transact-SQL search functions treat Hankaku Katakana and Zenkaku Katakana as different character sets. Katakana can be either half-width (Hankaku) or full-width (Zenkaku) in both EUC-JIS and Shift-JIS. For example, if the data in a table is stored in Hankaku Katakana, the following SQL statement returns no rows, even if both Hankaku and Zenkaku data have the same meaning:

  select * from table
  where column="Zenkaku_Katakana_here" 

• Hankaku Katakana cannot be used for procedure names or parameter names because the internal memory allocated for these names may be less than the length required by EUC.
• Hankaku Katakana can be used only in char, varchar, nchar, nvarchar, and text datatypes.
• The char, varchar, nchar, nvarchar, and text datatypes stored in an Adaptive Server using EUC-JIS can be used for a Shift-JIS client application. A client application using Shift-JIS can insert a maximum of 127 Hankaku Katakana characters, rather than 255 characters of an Adaptive Server using Shift-JIS. Inserting more than 127 characters during conversion to EUC-JIS generates an error message.
• Adaptive Server is not certified to use Hankaku Katakana for the sysname datatype such as database names, table names, column names, etc.
• Do not use Hankaku Katakana in text datatypes that are larger than one data page (typically 2K). If the text column contains Hankaku Katakana, and the character spans two pages in the database, the partial byte in the first page is not converted. A warning message advises that the conversion did not succeed.
• Even though you can use bcp from a client application using Shift-JIS to an Adaptive Server using EUC-JIS, you cannot run bcp from Shift-JIS client applications containing Hankaku Katakana characters.

For more information on character sets, see the System Administration Guide and the Adaptive Server Reference Manual.