I'm using Zope 2.1.4 on Linux with environment LC_ALL set to fi_FI and Zope started with -L fi_FI, and ZCatalog indexing & searching seems to work fine. (well, there's a problem with some documents having ä etc. spelled out, but if ä is used everything works OK.) BUT, ZDConfera & Squishdot don't work right. For some reason, at some phase of the indexing characters like ö (ö, octal 366) in words are converted to their octal codes. This can be seen by looking at the raw Data.fs file. The result is that when I have a word like kökköviesti in the ZDConfera or Squishdot article, it can be found by search as 366viesti It is converted in the db storing/indexing phase to something like k\366kk\366viesti, which is then broken into words k, 366kk, 366viesti which are stored into the textindex. Any ideas on how to fix this? I've looked at Splitter.c, TextIndex.py, ZDiscussions.py and Squishdot.py, but I can't find anything which would convert characters to their octal codes. //Jyrki