some query fails with local errors

/root/genii/mysql/queries/working_tpch1/misc

[root@srvregtest misc]# more bug2954.sql.log
ERROR 1690 (22003) at line 1: DOUBLE value is out of range in 'cot(0)'

[root@srvregtest misc]# more bug3783.sql.log
ERROR 1815 (HY000) at line 5: Internal error: IDB-2015: Sorting length exceeded. Session varia
ble max_length_for_sort_data needs to be set higher.

[root@srvregtest misc]# more bug5267.sql.log
ERROR 1178 (42000) at line 1: The storage engine for the table doesn't support IDB-1001: Funct
ion 'xor' can only be used in the outermost select or order by clause and cannot be used in co
njunction with an aggregate function.

/root/genii/mysql/queries/working_dml/misc

pwd
/root/genii/mysql/queries/working_dml/misc
[root@srvregtest misc]# grep -i error vtabledmlcol2col.sql.log
71 65916ERROR 1815 (HY000) at line 128: Internal error: IDB-2004: Cannot connect to ExeMgr.

pwd
/root/genii/mysql/queries/working_dml/misc
[root@srvregtest misc]# grep -i error vtabledmlcol2col.sql.log
71 65916ERROR 1815 (HY000) at line 128: Internal error: IDB-2004: Cannot connect to ExeMgr.

working_tpch1/qa_fe_cnxFunctions/NULLIF.DM.sql, line 2: fails the compare. NULLIF on date vs datetime returns as though the compare was falsem yet select with = returns true

MariaDB [tpch1]> select cidx, CDATE, CDATETIME, NULLIF(CDATE,CDATETIME), CDATE=CDATETIME from datatypetestm order by cidx;
-------------------------------------------------------------------------

-------------------------------------------------------------------------

-------------------------------------------------------------------------
11 rows in set (0.06 sec)

NULLIF should return NULL if the two values compare as =. For row 1, we see in the last column that they do compare =, yet NULLIF returns the value, which it should not do.

Re: NULLIF(CDATE,CDATETIME) the return type requested by MariaDB is DATE, where in mysql, it requested a string object. Columnstore has specific code for the date vs datetime NULLIF compare when a string is requested. When the DATE type return is requested, it had no such special processing.

Added the special processing for this case.

Re: working_tpch1_compareLogOnly/misc/bug4433.sql
This test is for the change made to InfiniDB to increase the maximum size of database and table names from 64 to 128. This change has not been ported to columnstore, so this test is invalid at this time. Should we decide to increase the maximum size to 120 char, I suggest we do it for all of MariaDB, as it takes a change in the Server code to effect this modification.

Re: working_tpch1_compareLogOnly/onClauseJoins/bug4031.sql query one gets a different answer than InfiniDB. It looks like another case where InfiniDB's optimizer is conflicting with MariaDB's optimizer and moving things to the inside of a join when they should stay outside.

This was indeed the same bug that optimized when it shouldn't. The shortcut I took before wasn't good enough. A full search of the list is needed.

		// MariaDB 10.1: cached_table is never available for derived tables.

		// Find the uncached object in table_list

		TABLE_LIST* tblList = ifp->context->table_list;

		while (tblList)

		{

			if (strcasecmp(tblList->alias, ifp->table_name) == 0)

			{

				if (!tblList->outer_join)

				{

					sc->derivedTable(derivedName);

					sc->derivedRefCol(cols[j].get());

				}

				break;

			}

			tblList = tblList->next_local;

		}

David.Hall regarding your comments on working_tpch1_compareLogOnly/onClauseJoins/bug4031.sql, it is not showing as failure in test001 test report . If this test script is outside of test001 testsuite, let us file a separate Jira item for it.

test001 as is now working, see MCOL-25 for details.

InnoDB comparison will fail, but test001 doesn't do this.