From: Wolfgang M. <wol...@ex...> - 2012-02-04 10:31:05
|
Hi Jens, > In all the string-related tests I have been able to think up, these characters are treated as they should in eXist - first of all, their string-length is consistently 1 and fn:contains() gives the expected results. To me, this shows that Java understands how to treat these characters and that the problem lies in the generation of the index. That's good news because it shows Java does indeed treat those characters as single characters, so my fears were wrong. I'm not sure why the ngram index fails then. Maybe it happens when it tries to convert characters to lower case. I'll do some debugging on this later. Wolfgang |