> From: Karl Řie [mailto:[EMAIL PROTECTED]] > > is it possible to "intercept" lucene's work at the > key-handling point? or > would this require a larger rewrite?
Not only would it be a large re-write, but I think it would make indexing slower. I have implemented full-text indexes a B-tree, and adding documents when the dictionary gets large is very slow. In Lucene terms, the index is always maintained in an optimized format, as a single segment. If you were to use multiple B-Trees, one per segment, then you would not take advantage of the B-Tree, and may as well use flat files, as Lucene already does. Doug -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>