> Plus, of course, index will only ever be used for operations where you have > overridden the default collating sequence for the operation, for example by > specifying collate nocase in the join expression, or adding the collate > nocase to > the order by or group by.
I assume this explains why the change in the table def made a difference from not specifying the collation whereas the index did. I did not override the default of the table in the query so the index was not used. I've encountered another issue as I was running my tests in sqlitestudio when I realized the query against the tables with the collation specified returned all rows in less than a minute. Running the query against the db in the sqlite shell is still bad. I know sqlitestudio enables certain non-default pragmas, but I wonder which ones could result in this speed difference. Thanks, jlc _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
