On 21 Mar 2011, at 5:46pm, Enrico Thierbach wrote: > On 21.03.2011, at 06:34, Simon Slavin wrote: > >> Suppose you have a TABLE employees with ten thousand rows and no indexes, >> and you execute >> >> SELECT id,firstname,surname FROM employees WHERE firstname='Guilherme' AND >> age=46 >> >> The query optimizer has a choice. It can first pick out all the Guilhermes, >> and then find those who are the right age, or it can first pick out all the >> 46 year olds, then find any with the right firstname. [snip] > > Is it really that way? I somehow assumed in that non-index case there is just > a single table scan involved? Fetching each record, calculating > firstname='Guilherme' AND age=46 for each, and selecting the matching > records?
Well Richard corrected me that the results of ANALYZE actually comes into play when two INDEXes do exist, and I am no longer certain about what I wrote above. It would probably become apparent if we looked at the results of an EXPLAIN QUERY PLAN for that situation. Sorry about the wrong steer, everyone. Simon. _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
