Added test cases for bug #7351.
item_cmpfunc.cc:
Fixed bug #7351: incorrect result for a query with a
subquery returning empty set.
If in the predicate v IN (SELECT a FROM t WHERE cond)
v is null, then the result of the predicate is either
INKNOWN or FALSE. It is FALSE if the subquery returns
an empty set.
item_subselect.cc:
Fixed bug #7351: incorrect result for a query with a
subquery returning empty set.
The problem was due to not a quite legal transformation
for 'IN' subqueries. A subquery containing a predicate
of the form
v IN (SELECT a FROM t WHERE cond)
was transformed into
EXISTS(SELECT a FROM t WHERE cond AND (a=v OR a IS NULL)).
Yet, this transformation is valid only if v is not null.
If v is null, then, in the case when
(SELECT a FROM t WHERE cond) returns an empty set the value
of the predicate is FALSE, otherwise the result of the
predicate is INKNOWN.
The fix resolves this problem by changing the result
of the transformation to
EXISTS(SELECT a FROM t WHERE cond AND (v IS NULL OR (a=v OR a IS NULL)))
in the case when v is nullable.
The new transformation prevents applying the lookup
optimization for IN subqueries. To make it still
applicable we have to introduce guarded access methods.
to behave well on 5.0 tables (well now you can't use tables from 4.1
and 5.0 with 4.0 because former use utf8, but still it is nice to have
similar code in acl_init() and replace_user_table()).
This also will make such GRANTs working in 5.0 (they are broken now).
STR_TO_DATE() function if there is another format specifier after %f
in format string". Also small cleanup of STR_TO_DATE() implementation.
(After review version.)
(back to behaviour of 4.1.7). Warning was not fatal: mysqldump continued. And the good thing is that it helped spot that starting from 4.1.7,
SHOW CREATE DATABASE failed (if --single-transaction and first db has non-empty InnoDB table and there is a second db) and thus mysqldump
produced CREATE DATABASE statements missing the CHARACTER SET clause. Removing the bug which was in the server, and the warning reporting in
mysqldump (compatibility with old servers).
For numeric constants we only need to add, since the parser doesn't produce
negative numbers.
For strings we only add (we actually could substract 1 if given string is a constant
and it has '-number' form but we're not doing that because
* we set max_length bigger then necessary in other cases as well.
* the current solution is simpler and safer (bigger max_length is better then cutting out)
which Heikki fixed in 4.1.8 and 4.0.23. I verified that without Heikki's patch the test fails (7 gets inserted).
Test added to 4.1 because in testsuite of 4.0 it's impossible to start slave with InnoDB.
ps-modify1 used the user variables @1, @2, @100 set within ps_query and
ps_modify. That architecture was wrong, because the dependence
of ps_modify1 on ps_query and ps_modify makes the test script
maintenance and the use of these test cases during bug fixing/
debugging of single sub test cases very uncomfortable.
Therefore these user variables (@1, @2, @100) are also set within ps-modify1.
The result files of the test cases ps_2myisam, ps_3innodb, ps_4heap, ps_6bdb,
ps_7ndb will be affected by that change and show 3 additional lines, but
nothing else will change.
