Recently several of the expected results in the test e_select.test were
changed to results that look very odd. Basically the numbers that were
returned by the query were replaced with lots of "#". For example, test
4.10 went from:
# EVIDENCE-OF: R-14926-50129 For the purposes of grouping rows, NULL
# values are considered equal.
#
do_select_tests e_select-4.10 {
1 "SELECT group_concat(y) FROM b2 GROUP BY x" {0,1 3 2,4}
2 "SELECT count(*) FROM b2 GROUP BY CASE WHEN y<4 THEN NULL ELSE 0
END" {4 1}
}
To:
# EVIDENCE-OF: R-14926-50129 For the purposes of grouping rows, NULL
# values are considered equal.
#
do_select_tests e_select-4.10 {
1 "SELECT group_concat(y) FROM b2 GROUP BY x" {/#,# 3 #,#/}
2 "SELECT count(*) FROM b2 GROUP BY CASE WHEN y<4 THEN NULL ELSE 0
END" {4 1}
}
The change was in a change set meant to fix test failures after changes
to the query optimizer, but I fail to see how changing the optimizer
would change the results of queries in such an odd way. Was this change
intentional, and if so, why are such odd results being printed out now?
You can view the specific change set at
http://www.hwaci.com/cgi-bin/sqlite/fdiff?v1=69013a64b4694588&v2=07e8d81268ba1ffc
Thanks for your time.
Lauren
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users