Am 27.09.2010 03:59, schrieb Andi Vajda: > In 3.x and trunk, I've been porting ICU-dependant Lucene contrib features to > use PyICU [1][2] (which depends on C++ ICU). I think that having PyLucene > depend both on C++ ICU and Java ICU is one ICU too many :-), though. > > I'm not sure at this point which should remain. There are advantages to > both... I'm open to arguments in favor of either. You can see examples in > the 3.x tree [3]. > > (disclaimer: I'm the author of PyICU)
It looks like you have far more experience and knowledge than we do. We will follow your lead here. ;) What are the advantages of ICU4J and ICU4C? We already found out that the default java.text.Collator is slower and not as efficient as the ICU collator. Christian
