The short-term goal is to do the same as in EDS, which has an utility function for stripping accents. For the sake of consistency with case-sensitivity, accent-sensitivity should be "off" by default.

Long term it may be worthwhile to investigate better ways of searching, for example a method which also matches "ü" against "ue". See http://userguide.icu-project.org/collation/icu-string-search-service

Another, related request was to match foreign and Latin ways of writing the same text, for example 江 = jiāng (= Jiang when ignoring case and accents).

I've decided to treat this the same way as accent-insensitive search, meaning that the default will be as relaxed as possible: case-insensitive, accent-insensitive, and transliterated. I've finished the implementation (including updated tests) and now only need to check for regressions, then merge into master.

Here's the updated README:

    For text values, the default search without explicit flags is
    very tolerant, meaning that it ignores quite a few differences
    between search term and value. The default search:
    - transliterates any foreign script in search term and values
      to Latin before comparison, thus finding 江 when searching
      for Jiang and vice-versa
    - is case-insensitive
    - is accent-insensitive

    Case and accent differences get removed after the optional
    transliteration. Spaces between words always matter.

    This behavior can be modified by giving additional,
    optional flags after the search value:
    'case-insensitive' - force case-insensitive search (available for the sake
    of consistency and just in case, should the default ever change)
    'case-sensitive' - force case-sensitive search
    'accent-insensitive', 'accent-sensitive' - same for accents
    'transliteration' - force transliteration, i.e. explicitly choose the
    current default
    'no-transliteration' - disable transliteration

Implemented in "master" branch, included in 1.3.99.6.