On Wed, Jul 10, 2013 at 4:20 PM, Uwe Schindler <u...@thetaphi.de> wrote: > Hi, > > The "fast" replacement (means sorting works as fast without collating) is to > index the fields > used for sorting with CollationKeyAnalyzer ([snip]). The Collator you get > from e.g. the locale. [snip] > The better was is, as mentioned before, to index the collated sort keys in > the index > and do a binary sort on the (non human-readable) keys.
So... one field dedicated for sorting per locale. Or at least per unique comparator (maybe multiple languages share sorting rules somewhere.) That seems like it could eat some space... I wonder if it couldn't just apply the collator when caching the values for sorting. Then it would all be in memory so it should work at the same speed as what's being done now. TX --------------------------------------------------------------------- To unsubscribe, e-mail: java-user-unsubscr...@lucene.apache.org For additional commands, e-mail: java-user-h...@lucene.apache.org
