Note: this page is out-of-date.  

Issue owner:
Don Day <dond at us dot ibm dot com>

Statement of problem or objective:
The index sort today does not work based on language - it appears that all terms are sorted based on English. The sort should work for the appropriate language.

Is this architecture, function, fix, or other?
function/fix

Goals of this proposal:
DITA Maps that are not English will generated an index that is sorted based on the specified language.

Use cases:
If DITA content is translated into Swedish, words in the index may use the characters å, ä, and ö. In English, the default sort places these with the A and O characters. In Swedish, these three letters are sorted at the end of the alphabet, after Z.

Stakeholders:
List of those who request or need the proposed item.
Any consumer of the toolkit who generates indexes for non-English output.

Interested parties:
List of people interested in potential meetings or discussions.  Use obfuscated addresses, please.  For example, dond at us dot ibm dot com
robander .... us dot ibm dot com

Proposed design:

• The sort should be based on the first language encountered in the files. If there is no language set in the map, use the first language found in a topic.
• If more than one language is found, warn the user that the sort will be based on the first one found, and indicate which language that is. We could limit this message to index terms; that is, if one topic in the map uses an alternate language, but contains no index terms, do not warn.
• Check to see if ICU is available. If it is, use that for sorting. Otherwise, use the default Java sort.