On Wed, 26 Feb 2014, Alexander Wagner wrote: > My concern was, that if I have to use regexp all the time, it will get > terribly slow. At the moment our bean counting takes about 1-2 min using > intbitsets on the backend and "" queries, regexping wouldn't be nice > here. Additionally, we have similar searches for frontend related stuff. > > But I think with the till now unknown tokenizer assignments everything > is fine.
OK. > Sorry for being incorrect. If I want to allow exact match for fields > that use fancy stuff by default I'd need another index. Also I have to > select another index and not just place it in "". Yes. One could imagine standardising the "exact..." technique to pre-create exact indexes for everything, but often it would not make sense, e.g. for that "doi" index. Best regards -- Tibor Simko
