Webhelp stemming and search indexer

Hi

    

    I am using the DocBook XSL stylesheets (version 1.78.1) to produce
    Webhelp, and my documents are being translated into French,
    Japanese, Korean, and Simplified Chinese.

    

    I have a couple of questions about configuring the Webhelp search
    which do not seem 100% obvious to me, having looked through the
    Webelp docs.

    

    (1) The Webhelp XSL templates always output a link to a _javascript_
    stemmer library. The file name of the library linked to is
    determined by the webhelp.indexer.language parameter. But Webhelp
    only includes stemmers for en, fr and de languages.

    

    Question 1: Is it OK to use the default "en" _javascript_ stemmer with
    non-English locales, or is it best to customize the template that
    outputs the stemmer link and remove the link for languages that do
    not have a stemmer?

    

    (2) The Java indexer command used with the Webelp build has the
    properties webhelp.indexer.language and enable.stemming.

    

    In trying to establish a list of languages that have Java stemmer
    support, the Webhelp docs have this:

    

    - In the section "Adding support for other (non-CJKV) languages")
    there is a list of non-CJKV languages that have stemmer support but
    no language codes.

    

    - In the section "Search indexing" it says look in the
    build.properties file for the language code, but the
    build.properties file says look in the docs.

    

    - In the section "New Stemmers" (in the developer docs) it seems to
    indicate a different list of languages with stemmers, with a list of
    language codes (including "cn" for Chinese?).

    

    Question 2: If the enable.stemming property if set to true, is the
    value of webhelp.indexer.language used to determine whether a Java
    stemmer is used?

    

    Question 3: Is there a definitive list of language codes that the
    Java indexer expects/accepts/supports for the language?

    

    Question 4: If a language has no Java stemmer, is it best to set the
    enable.stemming property to "false", or does it not really matter?

    

    

    Thanks