Christophe Leske wrote: > Shane Harrelson schrieb: >> Dennis- >> >> Your last "simplification": >> > I never got that email from Dennis, I would be very interested in it. > > Dennis, this is actually what i am currently doing. > > However: > > i see no speed up for large areas (half the globe, e.g.), but > considerable ones for small areas (a country like france for instance), > as well as very small areas (maximum zoom). > > I added an index on the ID field for the search in the city database, > that helped a bit, but i am dissapointed that the rtree search is not > faster than the normal search for bigger areas... >
Why would you expect it to be faster? Think about it for a second ... 1) if we do a search that covers the whole world, are we not doing the equivalent of a full table scan + the additional over head of the the rtree processing. This is just like doing a query the selects all entries in a btree indexed table. A full scan is the slowest. Then in a btree indexed table selecting a single record via the index is fastest, likewise with an rtree, if you narrow the search area to a very small area you eliminate most of the records and return just one I would expect that to be the fastest. -Steve W _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
