In prior releases of DITA Open Toolkit, index entries were sorted based on the Java JDK collator. For example, Swedish words beginning with umlauted a and o vowels appeared with the equivalent non-umlauted vowels, ignoring the linguistic significance in Swedish.
In release 1.3, the index sort is based on the paradigm of the language specified for the files. In the case of Swedish umlauted vowels, they now appear at the end of the alphabet, after Z.
If there is no language set in the map, DITA Open Toolkit uses the first language found in a given topic. If more than one language is found , the user is warned that the sort will be based on the first one found, along with the name of that language.
If the ICU (International Components for Unicode) is available (that is, the icu4.jar file is in your classpath), it is used for sorting. Otherwise, the default Java sort is used.