Hi all,
In Skrooge (http://skrooge.org), we use extensively views, and views based on
views. However, it seems that in such a case, the linked views are computed
several times. To better explain the issue, Stephane, our main developper,
created a small use case, explaining the issue :
CREATE TABLE t(x);
INSERT INTO t (x) VALUES ('A');
INSERT INTO t (x) VALUES ('B');
INSERT INTO t (x) VALUES ('C');
CREATE VIEW v_t1 AS SELECT x, random() as "Y" from t;
CREATE VIEW v_t2 AS SELECT x, Y as "Y1", (Y+1) as "Y2" from v_t1;
SELECT x, Y2-Y1 from v_t2;
The result is:
A|3495515542145967690
B|-2850173736877580107
C|5413870427642078391
instead of
A|1
B|1
C|1
==> WRONG RESULT
While the problem illustrated here leads to a wrong result, the real problem
for us is on the perfo side : underlying views are recomputed several times
for a top level view.
Do you think there is room for improvement here ? Does this sound feasible ?
Thanks :)
Guillaume
--
Skrooge, a free, Open Source, personal finances software for linux, Mac OS,
Windows
http://skrooge.org
_______________________________________________
sqlite-users mailing list
[email protected]
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
