On Tue, Jul 29, 2014 at 4:48 AM, Clemens Ladisch <clem...@ladisch.de> wrote:
> > Simplified for plain ol' SQLite as follows: > Thank you. That is exactly what we needed. The bug has now been characterized, studied, and fixed. See [ http://www.sqlite.org/src/info/d2889096e7bdeac6] for the ticket and [ http://www.sqlite.org/src/info/8cc41b0bf365] for a patch that fixes the problem. The problem is fixed on trunk and an amalgamation build of the latest trunk is available on the download page at [ http://www.sqlite.org/download.html] > > CREATE TABLE L (Lid INTEGER PRIMARY KEY); > INSERT INTO L VALUES(1); > CREATE TABLE P (Pid INTEGER PRIMARY KEY, Lid INTEGER); > INSERT INTO P VALUES(1,1); > INSERT INTO P VALUES(2,1); > CREATE VIRTUAL TABLE R USING rtree(Pid, xmin, xmax); > INSERT INTO R VALUES(1,0,0); > INSERT INTO R VALUES(2,0,0); > CREATE INDEX idx_P_Lid ON P (Lid); > > -- fails, result should be two rows: > SELECT * FROM P JOIN L USING (Lid) JOIN R USING (Pid) WHERE L.Lid = 1; > > > Regards, > Clemens > _______________________________________________ > sqlite-users mailing list > sqlite-users@sqlite.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > -- D. Richard Hipp d...@sqlite.org _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users