database/mariadb-columnstore-engine
1
0
Fork 0
mirror of https://github.com/mariadb-corporation/mariadb-columnstore-engine.git synced 2025-07-29 08:21:15 +03:00
Code Activity

MCOL-4234 and MCOL 5772 cherry-picked into [stable 23.10] (#3226)

Browse Source 
* MCOL-4234: improve GROUP BY and ORDER BY interaction (#3194)

This patch fixes the problem in MCOL-4234 and also generally improves
behavior of GROUP BY.

It does so by introducing a "dummy" aggregate and by wrapping columns
into it. This allows for columns that are not in GROUP BY to be used
more freely, for example, in SELECT * FROM tbl GROUP BY col - all
columns that are not "col" will be wrapped into an aggregate and query
will proceed to execution.

The dummy aggregate itself does nothing more than remember last value
passed into it.

There also an additional error message that tries to explain what types
of expressions can be wrapped into an aggregate.

* MCOL-5772: incorrect ORDER BY ordering for a columns not in GROUP BY (#3214)

When ORDER BY column is not in GROUP BY, is not an aggregate and there
is a SELECT column that is also not an aggregate, there was a problem:
ordering happened on the SELECTed column, not ORDERed one.

This patch fixes that particular problem and also performs some tidying
around newly added aggregate.

---------

Co-authored-by: Leonid Fedorov <79837786+mariadb-LeonidFedorov@users.noreply.github.com>
This commit is contained in:
Sergey Zefirov
2024-06-27 23:31:53 +03:00
committed by GitHub
parent 9f4231f87f
commit db4cb1d657
18 changed files with 370 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
dbcon/execplan/aggregatecolumn.h
View File

@ -75,6 +75,7 @@ class AggregateColumn : public ReturnedColumn
BIT_XOR,
GROUP_CONCAT,
JSON_ARRAYAGG,
SELECT_SOME,
UDAF,
MULTI_PARM
};

4
dbcon/joblist/joblistfactory.cpp
View File

@ -541,10 +541,12 @@ void checkGroupByCols(CalpontSelectExecutionPlan* csep, JobInfo& jobInfo)
// Not an aggregate column and not an expression of aggregation.
if (dynamic_cast<AggregateColumn*>(orderByCols[i].get()) == NULL &&
orderByCols[i]->aggColumnList().empty())
{
csep->groupByCols().push_back(orderByCols[i]);
}
}
}
}
if (csep->groupByCols().size() > 0)
{
@ -604,8 +606,10 @@ void checkGroupByCols(CalpontSelectExecutionPlan* csep, JobInfo& jobInfo)
}
if (csep->groupByCols().size() != uniqGbCols.size())
{
(csep)->groupByCols(uniqGbCols);
}
}
}
void checkAggregation(CalpontSelectExecutionPlan* csep, JobInfo& jobInfo)

11
dbcon/joblist/tupleaggregatestep.cpp
View File

@ -124,6 +124,8 @@ inline RowAggFunctionType functionIdMap(int planFuncId)
case AggregateColumn::MULTI_PARM: return ROWAGG_MULTI_PARM;
case AggregateColumn::SELECT_SOME: return ROWAGG_SELECT_SOME;
default: return ROWAGG_FUNCT_UNDEFINE;
}
}
@ -1204,6 +1206,7 @@ void TupleAggregateStep::prep1PhaseAggregate(JobInfo& jobInfo, vector<RowGroup>&
{
case ROWAGG_MIN:
case ROWAGG_MAX:
case ROWAGG_SELECT_SOME:
{
oidsAgg.push_back(oidsProj[colProj]);
keysAgg.push_back(key);
@ -1766,6 +1769,7 @@ void TupleAggregateStep::prep1PhaseDistinctAggregate(JobInfo& jobInfo, vector<Ro
{
case ROWAGG_MIN:
case ROWAGG_MAX:
case ROWAGG_SELECT_SOME:
{
oidsAgg.push_back(oidsProj[colProj]);
keysAgg.push_back(aggKey);
@ -2176,6 +2180,7 @@ void TupleAggregateStep::prep1PhaseDistinctAggregate(JobInfo& jobInfo, vector<Ro
case ROWAGG_BIT_AND:
case ROWAGG_BIT_OR:
case ROWAGG_BIT_XOR:
case ROWAGG_SELECT_SOME:
default:
{
AGG_MAP::iterator it = aggFuncMap.find(
@ -2832,7 +2837,7 @@ void TupleAggregateStep::prep1PhaseDistinctAggregate(JobInfo& jobInfo, vector<Ro
f->fAggFunction == ROWAGG_STATS || f->fAggFunction == ROWAGG_BIT_AND ||
f->fAggFunction == ROWAGG_BIT_OR || f->fAggFunction == ROWAGG_BIT_XOR ||
f->fAggFunction == ROWAGG_CONSTANT || f->fAggFunction == ROWAGG_GROUP_CONCAT ||
f->fAggFunction == ROWAGG_JSON_ARRAY))
f->fAggFunction == ROWAGG_JSON_ARRAY || f->fAggFunction == ROWAGG_SELECT_SOME))
{
funct.reset(new RowAggFunctionCol(f->fAggFunction, f->fStatsFunction, f->fInputColumnIndex,
f->fOutputColumnIndex, f->fAuxColumnIndex - multiParms));
@ -3127,6 +3132,7 @@ void TupleAggregateStep::prep2PhasesAggregate(JobInfo& jobInfo, vector<RowGroup>
{
case ROWAGG_MIN:
case ROWAGG_MAX:
case ROWAGG_SELECT_SOME:
{
oidsAggPm.push_back(oidsProj[colProj]);
keysAggPm.push_back(aggKey);
@ -4044,6 +4050,7 @@ void TupleAggregateStep::prep2PhasesDistinctAggregate(JobInfo& jobInfo, vector<R
{
case ROWAGG_MIN:
case ROWAGG_MAX:
case ROWAGG_SELECT_SOME:
{
oidsAggPm.push_back(oidsProj[colProj]);
keysAggPm.push_back(aggKey);
@ -5079,7 +5086,7 @@ void TupleAggregateStep::prep2PhasesDistinctAggregate(JobInfo& jobInfo, vector<R
f->fAggFunction == ROWAGG_MIN || f->fAggFunction == ROWAGG_MAX ||
f->fAggFunction == ROWAGG_STATS || f->fAggFunction == ROWAGG_BIT_AND ||
f->fAggFunction == ROWAGG_BIT_OR || f->fAggFunction == ROWAGG_BIT_XOR ||
f->fAggFunction == ROWAGG_CONSTANT)
f->fAggFunction == ROWAGG_CONSTANT || f->fAggFunction == ROWAGG_SELECT_SOME)
{
funct.reset(new RowAggFunctionCol(f->fAggFunction, f->fStatsFunction, f->fInputColumnIndex,
f->fOutputColumnIndex, f->fAuxColumnIndex - multiParms));

135
dbcon/mysql/ha_mcs_execplan.cpp
View File

@ -474,6 +474,22 @@ bool sortItemIsInGrouping(Item* sort_item, ORDER* groupcol)
if (sort_item->type() == Item::SUM_FUNC_ITEM)
{
found = true;
return found;
}
{
// as we now can warp ORDER BY or SELECT expression into
// an aggregate, we can pass FIELD_ITEM as "found" as well.
Item* item = sort_item;
while (item->type() == Item::REF_ITEM)
{
const Item_ref* ref_item = static_cast<const Item_ref*>(item);
item = (Item*)*ref_item->ref;
}
if (item->type() == Item::FIELD_ITEM || item->type() == Item::CONST_ITEM || item->type() == Item::NULL_ITEM)
{
return true;
}
}
// A function that contains an aggregate function
@ -3189,6 +3205,9 @@ CalpontSystemCatalog::ColType colType_MysqlToIDB(const Item* item)
if (item->field_type() == MYSQL_TYPE_BLOB)
{
// We default to BLOB, but then try to correct type,
// because many textual types in server have type_handler_blob
// (and variants) as their type.
ct.colDataType = CalpontSystemCatalog::BLOB;
}
}
@ -4902,6 +4921,70 @@ static void processAggregateColumnConstArg(gp_walk_info& gwi, SRCP& parm, Aggreg
}
}
void analyzeForImplicitGroupBy(Item* item, gp_walk_info& gwi)
{
if (gwi.implicitExplicitGroupBy)
{
return;
}
while (item->type() == Item::REF_ITEM)
{
Item_ref* ref = static_cast<Item_ref*>(item);
item = *ref->ref;
}
if (item->type() == Item::SUM_FUNC_ITEM)
{
// definitely an aggregate and thus needs an implicit group by.
gwi.implicitExplicitGroupBy = true;
return;
}
if (item->type() == Item::FUNC_ITEM)
{
Item_func* ifp = static_cast<Item_func*>(item);
for(uint32_t i = 0;i<ifp->argument_count() && !gwi.implicitExplicitGroupBy;i++)
{
analyzeForImplicitGroupBy(ifp->arguments()[i], gwi);
}
}
}
ReturnedColumn* wrapIntoAggregate(ReturnedColumn* rc, gp_walk_info& gwi, SELECT_LEX& select_lex, Item* baseItem)
{
if (!gwi.implicitExplicitGroupBy)
{
return rc;
}
if (dynamic_cast<AggregateColumn*>(rc) != nullptr || dynamic_cast<ConstantColumn*>(rc) != nullptr)
{
return rc;
}
ORDER* groupcol = static_cast<ORDER*>(select_lex.group_list.first);
while (groupcol)
{
if (baseItem->eq(*groupcol->item, false))
{
return rc;
}
groupcol = groupcol->next;
}
cal_connection_info* ci = static_cast<cal_connection_info*>(get_fe_conn_info_ptr());
AggregateColumn* ac = new AggregateColumn(gwi.sessionid);
ac->timeZone(gwi.timeZone);
ac->alias(rc->alias());
ac->aggOp(AggregateColumn::SELECT_SOME);
ac->asc(rc->asc());
ac->charsetNumber(rc->charsetNumber());
ac->expressionId(ci->expressionId++);
ac->aggParms().push_back(SRCP(rc));
return ac;
}
ReturnedColumn* buildAggregateColumn(Item* item, gp_walk_info& gwi)
{
// MCOL-1201 For UDAnF multiple parameters
@ -5009,8 +5092,9 @@ ReturnedColumn* buildAggregateColumn(Item* item, gp_walk_info& gwi)
if (ord_col->type() == Item::CONST_ITEM && ord_col->cmp_type() == INT_RESULT)
{
Item_int* id = (Item_int*)ord_col;
int64_t index = id->val_int();
if (id->val_int() > (int)selCols.size())
if (index > (int)selCols.size() || index < 1)
{
gwi.fatalParseError = true;
@ -5020,8 +5104,8 @@ ReturnedColumn* buildAggregateColumn(Item* item, gp_walk_info& gwi)
return NULL;
}
rc = selCols[id->val_int() - 1]->clone();
rc->orderPos(id->val_int() - 1);
rc = selCols[index - 1]->clone();
rc->orderPos(index - 1);
}
else
{
@ -7483,6 +7567,32 @@ int getSelectPlan(gp_walk_info& gwi, SELECT_LEX& select_lex, SCSEP& csep, bool i
}
#endif
// analyze SELECT and ORDER BY parts - do they have implicit GROUP BY induced by aggregates?
{
if (select_lex.group_list.first)
{
// we have an explicit GROUP BY.
gwi.implicitExplicitGroupBy = true;
}
else
{
// do we have an implicit GROUP BY?
List_iterator_fast<Item> it(select_lex.item_list);
Item* item;
while ((item = it++))
{
analyzeForImplicitGroupBy(item, gwi);
}
SQL_I_List<ORDER> order_list = select_lex.order_list;
ORDER* ordercol = static_cast<ORDER*>(order_list.first);
for (; ordercol; ordercol = ordercol->next)
{
analyzeForImplicitGroupBy(*(ordercol->item), gwi);
}
}
}
// populate returnedcolumnlist and columnmap
List_iterator_fast<Item> it(select_lex.item_list);
Item* item;
@ -7509,6 +7619,7 @@ int getSelectPlan(gp_walk_info& gwi, SELECT_LEX& select_lex, SCSEP& csep, bool i
// @bug 5916. Need to keep checking until getting concret item in case
// of nested view.
Item* baseItem = item;
while (item->type() == Item::REF_ITEM)
{
Item_ref* ref = (Item_ref*)item;
@ -7533,8 +7644,6 @@ int getSelectPlan(gp_walk_info& gwi, SELECT_LEX& select_lex, SCSEP& csep, bool i
if (sc)
{
boost::shared_ptr<SimpleColumn> spsc(sc);
string fullname;
String str;
ifp->print(&str, QT_ORDINARY);
@ -7550,10 +7659,14 @@ int getSelectPlan(gp_walk_info& gwi, SELECT_LEX& select_lex, SCSEP& csep, bool i
sc->alias(itemAlias);
}
gwi.returnedCols.push_back(spsc);
// We need to look into GROUP BY columns to decide if we need to wrap a column.
ReturnedColumn* rc = wrapIntoAggregate(sc, gwi, select_lex, baseItem);
SRCP sprc(rc);
gwi.returnedCols.push_back(sprc);
gwi.columnMap.insert(
CalpontSelectExecutionPlan::ColumnMap::value_type(string(ifp->field_name.str), spsc));
CalpontSelectExecutionPlan::ColumnMap::value_type(string(ifp->field_name.str), sprc));
TABLE_LIST* tmp = 0;
if (ifp->cached_table)
@ -8400,6 +8513,7 @@ int getSelectPlan(gp_walk_info& gwi, SELECT_LEX& select_lex, SCSEP& csep, bool i
{
if ((*(ordercol->item))->type() == Item::WINDOW_FUNC_ITEM)
gwi.hasWindowFunc = true;
// XXX: TODO: implement a proper analysis of what we support.
// MCOL-2166 Looking for this sorting item in GROUP_BY items list.
// Shouldn't look into this if query doesn't have GROUP BY or
// aggregations
@ -8411,10 +8525,10 @@ int getSelectPlan(gp_walk_info& gwi, SELECT_LEX& select_lex, SCSEP& csep, bool i
getColNameFromItem(osr, *ordercol->item);
Message::Args args;
args.add(ostream.str());
string emsg = IDBErrorInfo::instance()->errorMsg(ERR_NOT_GROUPBY_EXPRESSION, args);
string emsg = IDBErrorInfo::instance()->errorMsg(ERR_NOT_SUPPORTED_GROUPBY_ORDERBY_EXPRESSION, args);
gwi.parseErrorText = emsg;
setError(gwi.thd, ER_INTERNAL_ERROR, emsg, gwi);
return ERR_NOT_GROUPBY_EXPRESSION;
return ERR_NOT_SUPPORTED_GROUPBY_ORDERBY_EXPRESSION;
}
}
@ -8464,6 +8578,8 @@ int getSelectPlan(gp_walk_info& gwi, SELECT_LEX& select_lex, SCSEP& csep, bool i
else
{
rc = buildReturnedColumn(ord_item, gwi, gwi.fatalParseError);
rc = wrapIntoAggregate(rc, gwi, select_lex, ord_item);
}
// @bug5501 try item_ptr if item can not be fixed. For some
// weird dml statement state, item can not be fixed but the
@ -8495,6 +8611,7 @@ int getSelectPlan(gp_walk_info& gwi, SELECT_LEX& select_lex, SCSEP& csep, bool i
gwi.orderByCols.push_back(SRCP(rc));
}
}
// make sure columnmap, returnedcols and count(*) arg_list are not empty
TableMap::iterator tb_iter = gwi.tableMap.begin();

4
dbcon/mysql/ha_mcs_impl_if.h
View File

@ -141,6 +141,9 @@ struct gp_walk_info
std::vector<execplan::CalpontSystemCatalog::TableAliasName> correlatedTbNameVec;
ClauseType clauseType;
execplan::CalpontSystemCatalog::TableAliasName viewName;
// we can have explicit GROUP BY and implicit one, triggered by aggregate in pojection or ORDER BY.
// this flag tells us whether we have either case.
bool implicitExplicitGroupBy;
bool aggOnSelect;
bool hasWindowFunc;
bool hasSubSelect;
@ -186,6 +189,7 @@ struct gp_walk_info
, subSelectType(uint64_t(-1))
, subQuery(0)
, clauseType(INIT)
, implicitExplicitGroupBy(false)
, aggOnSelect(false)
, hasWindowFunc(false)
, hasSubSelect(false)

23
mysql-test/columnstore/basic/r/MCOL-5772-hidden-order-by.result Normal file
View File

@ -0,0 +1,23 @@
DROP DATABASE IF EXISTS MCOL5772;
CREATE DATABASE MCOL5772;
USE MCOL5772;
CREATE TABLE products (
product_id INT,
product_name VARCHAR(100),
category VARCHAR(50),
unit_price DECIMAL(10, 2),
stock_quantity INT
) ENGINE=Columnstore;
INSERT INTO products VALUES
(1, 'Laptop', 'Electronics', 1200.00, 50),
(2, 'Smartphone', 'Electronics', 800.00, 100),
(3, 'Coffee Maker', 'Appliances', 50.00, 30),
(4, 'Backpack', 'Fashion', 40.00, 80),
(5, 'Desk Chair', 'Furniture', 150.00, 20);
SELECT product_name, SUM(stock_quantity) AS total_stock FROM products GROUP BY category ORDER BY stock_quantity;
product_name total_stock
Desk Chair 20
Coffee Maker 30
Backpack 80
Smartphone 150
DROP DATABASE MCOL5772;

3
mysql-test/columnstore/basic/r/mcol-4525.result
View File

@ -80,7 +80,8 @@ SET columnstore_select_handler=ON;
SELECT SUM(col1) AS col2 FROM t1 GROUP BY col2;
ERROR 42000: The storage engine for the table doesn't support MCS-2016: Non supported item 'col2' on the GROUP BY list.
SELECT col1 c FROM t1 ORDER BY AVG(col1);
ERROR HY000: Internal error: MCS-2021: 'c' is not in GROUP BY clause. All non-aggregate columns in the SELECT and ORDER BY clause must be included in the GROUP BY clause.
c
10
SET columnstore_select_handler=AUTO;
SELECT SUM(col1) AS col2 FROM t1 GROUP BY col2;
col2

4
mysql-test/columnstore/basic/r/mcs35_select_group_by.result
View File

@ -56,7 +56,9 @@ NULL NULL NULL NULL NULL NULL NULL NULL NULL NULL
NULL NULL NULL NULL NULL NULL NULL NULL NULL NULL
NULL NULL NULL NULL NULL NULL NULL NULL NULL NULL
SELECT * FROM t1 GROUP BY t1_tinyint;
ERROR HY000: Internal error: MCS-2021: '`mcs35_db1`.`t1`.`t1_int`' is not in GROUP BY clause. All non-aggregate columns in the SELECT and ORDER BY clause must be included in the GROUP BY clause.
t1_tinyint t1_int t1_bigint t1_double t1_float t1_blob t1_text t1_char t1_varchar t1_datetime
0 NULL 403685477580676 54.797693231 8.40287 222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222222 dddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd d hello world! 4387-11-08 11:22:30
NULL NULL NULL NULL NULL NULL NULL NULL NULL NULL
SELECT COUNT(*) FROM t1 GROUP BY t1_tinyint;
COUNT(*)
11

23
mysql-test/columnstore/basic/r/mcs36_select_order_by_group_by.result
View File

@ -69,7 +69,11 @@ spID userid MIN(t1.score)
3 3 3
SELECT t1.spID, t2.userid, MIN(t1.score)
FROM t1, t2 WHERE t1.userID=t2.userID GROUP BY t2.userid, t1.spID ORDER BY t1.date;
ERROR HY000: Internal error: MCS-2021: 'mcs36_db1.t1.date' is not in GROUP BY clause. All non-aggregate columns in the SELECT and ORDER BY clause must be included in the GROUP BY clause.
spID userid MIN(t1.score)
1 1 1
2 1 1
2 2 2
3 3 3
SELECT t2.userid, MIN(t1.score)
FROM t1, t2 WHERE t1.userID=t2.userID GROUP BY unknown ORDER BY t2.userid;
ERROR 42S22: Unknown column 'unknown' in 'group statement'
@ -78,5 +82,20 @@ FROM t1, t2 WHERE t1.userID=t2.userID GROUP BY t2.userid ORDER BY unknown;
ERROR 42S22: Unknown column 'unknown' in 'order clause'
SELECT t2.userid, MIN(t1.score)
FROM t1, t2 WHERE t1.userID=t2.userID GROUP BY t2.userid ORDER BY NULL;
ERROR HY000: Internal error: MCS-2021: 'unknown db.unknown table.unknown field' is not in GROUP BY clause. All non-aggregate columns in the SELECT and ORDER BY clause must be included in the GROUP BY clause.
userid MIN(t1.score)
1 1
2 2
3 3
SELECT * FROM t1 GROUP BY spID, userID ORDER BY score ASC, spID, userID;
spID userID score lsg date
1 1 1 0000-00-00
2 1 1 0000-00-00
2 2 2 0000-00-00
3 3 3 0000-00-00
SELECT * FROM t1 GROUP BY spID, userID ORDER BY score DESC, spID, userID;
spID userID score lsg date
3 3 3 0000-00-00
2 2 2 0000-00-00
1 1 1 0000-00-00
2 1 1 0000-00-00
DROP DATABASE IF EXISTS mcs36_db1;

4
mysql-test/columnstore/basic/r/mcs76_having.result
View File

@ -23,7 +23,9 @@ col1 col2
3 sss
4 ooo
SELECT col1, col2, SUM(LENGTH(col2)) FROM t1 GROUP BY col1 HAVING col1 > 1 AND col2 LIKE '%o%' ORDER BY col1;
ERROR HY000: Internal error: MCS-2021: '`mcs76_db`.`t1`.`col2`' is not in GROUP BY clause. All non-aggregate columns in the SELECT and ORDER BY clause must be included in the GROUP BY clause.
col1 col2 SUM(LENGTH(col2))
2 oooooooooooooooooooo 40
4 ooo 6
CREATE TABLE t2(col1 INT, col2 DATETIME)ENGINE=Columnstore;
INSERT INTO t2 VALUES(1, '2020-2-2'),(2, '2020-3-3'),(5,'2020-6-6'),(6, '2020-7-7');
SELECT t1.col1, SUM(t1.col1*t2.col1) AS a FROM t1 JOIN t2 ON t1.col1 = t2.col1 GROUP BY t1.col1 HAVING a>1 ORDER BY t1.col1;

28
mysql-test/columnstore/basic/t/MCOL-5772-hidden-order-by.test Normal file
View File

@ -0,0 +1,28 @@
# Order by a column that is not in GROUP BY and SELECT parts
# should be correct.
--disable_warnings
DROP DATABASE IF EXISTS MCOL5772;
--enable_warnings
CREATE DATABASE MCOL5772;
USE MCOL5772;
CREATE TABLE products (
product_id INT,
product_name VARCHAR(100),
category VARCHAR(50),
unit_price DECIMAL(10, 2),
stock_quantity INT
) ENGINE=Columnstore;
INSERT INTO products VALUES
(1, 'Laptop', 'Electronics', 1200.00, 50),
(2, 'Smartphone', 'Electronics', 800.00, 100),
(3, 'Coffee Maker', 'Appliances', 50.00, 30),
(4, 'Backpack', 'Fashion', 40.00, 80),
(5, 'Desk Chair', 'Furniture', 150.00, 20);
SELECT product_name, SUM(stock_quantity) AS total_stock FROM products GROUP BY category ORDER BY stock_quantity;
DROP DATABASE MCOL5772;

1
mysql-test/columnstore/basic/t/mcol-4525.test
View File

@ -85,7 +85,6 @@ INSERT INTO t1 VALUES(10,'bob',10);
SET columnstore_select_handler=ON;
--error 1178
SELECT SUM(col1) AS col2 FROM t1 GROUP BY col2;
--error 1815
SELECT col1 c FROM t1 ORDER BY AVG(col1);
SET columnstore_select_handler=AUTO;
SELECT SUM(col1) AS col2 FROM t1 GROUP BY col2;

2
mysql-test/columnstore/basic/t/mcs35_select_group_by.test
View File

@ -52,7 +52,7 @@ INSERT INTO t1 VALUES(NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NUL
--sorted_result
SELECT * FROM t1;
#Without aggregate function
--error 1815
--sorted_result
SELECT * FROM t1 GROUP BY t1_tinyint;
--sorted_result

10
mysql-test/columnstore/basic/t/mcs36_select_order_by_group_by.test
View File

@ -60,8 +60,7 @@ SELECT t1.spID, t2.userid, MIN(t1.score)
FROM t1, t2 WHERE t1.userID=t2.userID GROUP BY t2.userid, t1.spID ORDER BY t1.spID, t2.userid;
# columns in ORDER BY clause not included in the GROUP BY clause.
#Suspected bug. Innodb succeeds.
--error 1815
--sorted_result
SELECT t1.spID, t2.userid, MIN(t1.score)
FROM t1, t2 WHERE t1.userID=t2.userID GROUP BY t2.userid, t1.spID ORDER BY t1.date;
@ -76,10 +75,13 @@ SELECT t2.userid, MIN(t1.score)
FROM t1, t2 WHERE t1.userID=t2.userID GROUP BY t2.userid ORDER BY unknown;
# order by null
#Suspected bug. Innodb succeeds.
--error 1815
# Innodb succeeds - it is a workaround to speed up GROUP BY.
--sorted_result
SELECT t2.userid, MIN(t1.score)
FROM t1, t2 WHERE t1.userID=t2.userID GROUP BY t2.userid ORDER BY NULL;
SELECT * FROM t1 GROUP BY spID, userID ORDER BY score ASC, spID, userID;
SELECT * FROM t1 GROUP BY spID, userID ORDER BY score DESC, spID, userID;
#Clean up
DROP DATABASE IF EXISTS mcs36_db1;

2
mysql-test/columnstore/basic/t/mcs76_having.test
View File

@ -18,7 +18,7 @@ SELECT col2 FROM t1 GROUP BY col2 HAVING col2 LIKE '%o%' ORDER BY col2;
SELECT col1 FROM t1 GROUP BY col1 HAVING col1 > 1 ORDER BY col1;
SELECT col1, col2 FROM t1 GROUP BY col1, col2 HAVING col1 > 1 AND col2 LIKE '%o%' ORDER BY col1;
SELECT col1, col2 FROM t1 GROUP BY col1, col2 HAVING col1 > 1 OR col2 LIKE '%o%' ORDER BY col1;
--error ER_INTERNAL_ERROR
--sorted_result
SELECT col1, col2, SUM(LENGTH(col2)) FROM t1 GROUP BY col1 HAVING col1 > 1 AND col2 LIKE '%o%' ORDER BY col1;
CREATE TABLE t2(col1 INT, col2 DATETIME)ENGINE=Columnstore;

2
utils/loggingcpp/ErrorMessage.txt
View File

@ -108,6 +108,8 @@
2060 ERR_UNION_DECIMAL_OVERFLOW Union operation exceeds maximum DECIMAL precision of 38.
2061 ERR_NOT_SUPPORTED_GROUPBY_ORDERBY_EXPRESSION %1% is not in GROUP BY clause, not a column or an expression that contains function.
# Sub-query errors
3001 ERR_NON_SUPPORT_SUB_QUERY_TYPE This subquery type is not supported yet.
3002 ERR_MORE_THAN_1_ROW Subquery returns more than 1 row.

131
utils/rowgroup/rowaggregation.cpp
View File

@ -1254,6 +1254,108 @@ void RowAggregation::doMinMax(const Row& rowIn, int64_t colIn, int64_t colOut, i
}
}
}
//------------------------------------------------------------------------------
// Update the fields with anything that goes in.
// rowIn(in) - Row to be included in aggregation.
// colIn(in) - column in the input row group
// colOut(in) - column in the output row group
//------------------------------------------------------------------------------
void RowAggregation::doSelectSome(const Row& rowIn, int64_t colIn, int64_t colOut)
{
int colDataType = (fRowGroupIn.getColTypes())[colIn];
switch (colDataType)
{
case execplan::CalpontSystemCatalog::UTINYINT:
case execplan::CalpontSystemCatalog::USMALLINT:
case execplan::CalpontSystemCatalog::UMEDINT:
case execplan::CalpontSystemCatalog::UINT:
case execplan::CalpontSystemCatalog::UBIGINT:
case execplan::CalpontSystemCatalog::TINYINT:
case execplan::CalpontSystemCatalog::SMALLINT:
case execplan::CalpontSystemCatalog::MEDINT:
case execplan::CalpontSystemCatalog::INT:
case execplan::CalpontSystemCatalog::BIGINT:
{
fRow.setIntField(rowIn.getIntField(colIn), colOut);
break;
}
case execplan::CalpontSystemCatalog::DECIMAL:
case execplan::CalpontSystemCatalog::UDECIMAL:
{
if (LIKELY(rowIn.getColumnWidth(colIn) == datatypes::MAXDECIMALWIDTH))
{
fRow.setInt128Field(rowIn.getTSInt128Field(colIn).getValue(), colOut);
}
else if (rowIn.getColumnWidth(colIn) <= datatypes::MAXLEGACYWIDTH)
{
fRow.setIntField(rowIn.getIntField(colIn), colOut);
}
else
{
idbassert(0);
throw std::logic_error("RowAggregation::doMinMax(): DECIMAL bad length.");
}
break;
}
case execplan::CalpontSystemCatalog::CHAR:
case execplan::CalpontSystemCatalog::VARCHAR:
case execplan::CalpontSystemCatalog::TEXT:
{
auto valIn = rowIn.getStringField(colIn);
fRow.setStringField(valIn, colOut);
break;
}
case execplan::CalpontSystemCatalog::DOUBLE:
case execplan::CalpontSystemCatalog::UDOUBLE:
{
double valIn = rowIn.getDoubleField(colIn);
fRow.setDoubleField(valIn, colOut);
break;
}
case execplan::CalpontSystemCatalog::FLOAT:
case execplan::CalpontSystemCatalog::UFLOAT:
{
float valIn = rowIn.getFloatField(colIn);
fRow.setFloatField(valIn, colOut);
break;
}
case execplan::CalpontSystemCatalog::DATE:
case execplan::CalpontSystemCatalog::DATETIME:
case execplan::CalpontSystemCatalog::TIMESTAMP:
case execplan::CalpontSystemCatalog::TIME:
{
uint64_t valIn = rowIn.getUintField(colIn);
fRow.setUintField(valIn, colOut);
break;
}
case execplan::CalpontSystemCatalog::LONGDOUBLE:
{
long double valIn = rowIn.getLongDoubleField(colIn);
fRow.setLongDoubleField(valIn, colOut);
break;
}
case execplan::CalpontSystemCatalog::CLOB:
case execplan::CalpontSystemCatalog::BLOB:
{
fRow.setVarBinaryField(rowIn.getVarBinaryField(colIn), rowIn.getVarBinaryLength(colIn), colOut);
break;
}
default:
{
idbassert_s(0, "unknown data type in doSelectSome()");
break;
}
}
}
//------------------------------------------------------------------------------
// Update the sum fields if input is not null.
@ -1723,6 +1825,11 @@ void RowAggregation::updateEntry(const Row& rowIn, std::vector<mcsv1sdk::mcsv1Co
doUDAF(rowIn, colIn, colOut, colOut + 1, i, rgContextColl);
break;
}
case ROWAGG_SELECT_SOME:
{
doSelectSome(rowIn, colIn, colOut);
break;
}
default:
{
@ -1783,6 +1890,12 @@ void RowAggregation::mergeEntries(const Row& rowIn)
case ROWAGG_UDAF: doUDAF(rowIn, colOut, colOut, colOut + 1, i); break;
case ROWAGG_SELECT_SOME:
{
doSelectSome(rowIn, colOut, colOut);
break;
}
default:
std::ostringstream errmsg;
errmsg << "RowAggregation: function (id = " << (uint64_t)fFunctionCols[i]->fAggFunction
@ -2617,6 +2730,12 @@ void RowAggregationUM::updateEntry(const Row& rowIn, std::vector<mcsv1sdk::mcsv1
break;
}
case ROWAGG_SELECT_SOME:
{
doSelectSome(rowIn, colIn, colOut);
break;
}
default:
{
// need a exception to show the value
@ -4211,6 +4330,12 @@ void RowAggregationUMP2::updateEntry(const Row& rowIn, std::vector<mcsv1sdk::mcs
break;
}
case ROWAGG_SELECT_SOME:
{
doSelectSome(rowIn, colIn, colOut);
break;
}
default:
{
std::ostringstream errmsg;
@ -4703,6 +4828,12 @@ void RowAggregationDistinct::updateEntry(const Row& rowIn, std::vector<mcsv1sdk:
break;
}
case ROWAGG_SELECT_SOME:
{
doSelectSome(rowIn, colIn, colOut);
break;
}
default:
{
std::ostringstream errmsg;

6
utils/rowgroup/rowaggregation.h
View File

@ -113,7 +113,10 @@ enum RowAggFunctionType
ROWAGG_DUP_FUNCT, // duplicate aggregate Function(), except AVG and UDAF, in select
ROWAGG_DUP_AVG, // duplicate AVG(column_name) in select
ROWAGG_DUP_STATS, // duplicate statistics functions in select
ROWAGG_DUP_UDAF // duplicate UDAF function in select
ROWAGG_DUP_UDAF, // duplicate UDAF function in select
// a dummy "select some" aggregate needed for non-group-by values in SELECT's with GROUP BY's
ROWAGG_SELECT_SOME
};
//------------------------------------------------------------------------------
@ -538,6 +541,7 @@ class RowAggregation : public messageqcpp::Serializeable
virtual void updateEntry(const Row& row, std::vector<mcsv1sdk::mcsv1Context>* rgContextColl = nullptr);
void mergeEntries(const Row& row);
virtual void doMinMax(const Row&, int64_t, int64_t, int);
virtual void doSelectSome(const Row& rowIn, int64_t colIn, int64_t colOut);
virtual void doSum(const Row&, int64_t, int64_t, int);
virtual void doAvg(const Row&, int64_t, int64_t, int64_t, bool merge = false);
virtual void doStatistics(const Row&, int64_t, int64_t, int64_t);