General Administration

Internationalization I18N - Background Installation Information

 

Background Installation Information

 
  • DataView Client
    The DataView client needs a new registry entry name EP_CLIENT_CHARSET. It contains the canonical ICU name of the codepage used by the client. The DataView client always uses the Windows codepage Cp1252, which has the ICU name ibm-5348_P100-1997.
    This client codepage is read by the server, but only used if server side character conversion has been activated (see also e6 Server).
  • Java Client
    The Java client (and all other Java ECI clients) scans the configuration parameter EDB-CHR-ENC-JVM for a standard environment (currently set to ISO-8859-15).
    If NLS_LANG is set to AMERICAN_AMERICA.WE8MSWIN1252 (see below), this parameter should be set to windows-1252.
    For an I18N environment it must always be set to UTF-8, regardless of the NLS_LANG setting.
    Each user in an I18N environment needs to configure a proper font that is able to display non-Latin characters.
  • Web Client
    Same as Java client. Additionally, the encoding defined in the webplm properties file need to be adapted to use UTF-8 instead of ISO-8859-x in an I18N environment.
  • J2EE
    The Java property oracle.jdbc.defaultNChar must be set to "true" in an I18N environment that accesses an Oracle database. This is currently done in the run.bat/run.sh scripts and in the J2EE wrapper configuration BS_wrapper.conf.
  • e6 Server
    Three new configuration entries exist to control how characters are mapped from the client to the database charset:
    • If EDB-EPQ-USE-CHAR-CONVERSION is active the EPQ performs the client side character set conversion.
    • If EDB-EPQ-USE-SUBST-CHARACTER is set invalid characters will be replaced with a substitution character. Otherwise, an error message is issued and the conversion is canceled.
    • If EDB-EPQ-SUBST-CHARACTER is set its content is used as substitution character. Otherwise, the ICU's default substitution character (usually 0x1a) is used.
    In the standard dump the character conversion is deactivated. This configuration is needed when upgraded dumps are used because they can contain illegal characters (e.g. €).
  • Oracle Database
    For upgraded environments NLS_LANG should not be changed (AMERICAN_AMERICA.WE8ISO8859P15).
    In this environment no character conversion is performed and the bytes are written to the database as they are.
    For new environments NLS_LANG can be set to AMERICAN_AMERICA.WE8MSWIN1252.
    In this environment a character conversion is performed by the Oracle library (preferred method).