Dokuwiki currently breaks Unicode string equivalence, where two sequences of characters are considered equivalent. For example "é" <U+0065 + U+0301> and "é" <U+00E9> are equivalent but Dokuwiki doesn't see them as such.
Dokuwiki should use something like UTFNormal.php of Mediawiki to normalize Unicode strings before any comparison (like in search), probably for indexing and when saving data as well. NFC is the form recommended by the W3C.
See
http://www.w3.org/TR/charmod-norm/
http://www.unicode.org/unicode/reports/tr15/