Hello,

The next step of removal of call to Perl code is for reproducible
transliteration.  It would be possible to translate the Perl code in C,
but I would prefer not to, as I think we should not implement
transliteration ourselves, but use existing libraries.  Instead, I would
like to simplify how transliteration is done in Perl.

Here is my proposal:

* in HTML, do not use transliteration for internal index and index
  letter targets
* only use Text::Unidecode for transliteration in Perl and always use it
  if there is some transliteration, remove the USE_UNIDECODE customization
* change tests such that transliteration is only tested for cases
  were all the iconv and Text::Unidecode give the same result

-- 
Pat

Reply via email to