On 4/20/2014 5:21 AM, Robert Muir wrote: > There is no bug here. the positions are correct. > > If you want to use phrase queries, i wouldnt try to be so tricky with n-grams. > > This never works well, and there is nothing to fix...
OK, if it's not a bug, would it be reasonable at all to have a config option that preserves positions for all unigrams? The customer (a Japanese company) wants to be able to search for single characters, but they do not want to have single-character "words" match when they put in longer strings. This means that outputUnigrams needs to be there on the index analysis, but not the query analysis. This combination breaks phrase queries. The config option I just mentioned would fix this. Although it's possible it might break something else, off the top of my head I can't think of any examples. Thanks, Shawn --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
