Commit 50888477 authored by unknown's avatar unknown

MWL#89: Cost-based choice between Materialization and IN->EXISTS transformation

* Fixed a "crack" between semijoin analysis and materialization analysis
  where semijoin didn't set the correct strategy for the IN predicate.

* Cosmetic changes in the code/comments.
parent 2c811505
...@@ -74,8 +74,6 @@ int check_and_do_in_subquery_rewrites(JOIN *join) ...@@ -74,8 +74,6 @@ int check_and_do_in_subquery_rewrites(JOIN *join)
1) this join is inside a subquery (of any type except FROM-clause 1) this join is inside a subquery (of any type except FROM-clause
subquery) and subquery) and
2) we aren't just normalizing a VIEW 2) we aren't just normalizing a VIEW
3) The join and its select_lex object do not represent the 'fake'
select used to compute the result of a UNION.
Then perform early unconditional subquery transformations: Then perform early unconditional subquery transformations:
- Convert subquery predicate into semi-join, or - Convert subquery predicate into semi-join, or
...@@ -87,9 +85,8 @@ int check_and_do_in_subquery_rewrites(JOIN *join) ...@@ -87,9 +85,8 @@ int check_and_do_in_subquery_rewrites(JOIN *join)
TODO: for PS, make the whole block execute only on the first execution TODO: for PS, make the whole block execute only on the first execution
*/ */
Item_subselect *subselect; Item_subselect *subselect;
if (!thd->lex->view_prepare_mode && // (1) if (!thd->lex->view_prepare_mode && // (1)
(subselect= parent_unit->item))// && // (2) (subselect= parent_unit->item)) // (2)
// select_lex == parent_unit->fake_select_lex) // (3)
{ {
Item_in_subselect *in_subs= NULL; Item_in_subselect *in_subs= NULL;
if (subselect->substype() == Item_subselect::IN_SUBS) if (subselect->substype() == Item_subselect::IN_SUBS)
...@@ -134,7 +131,13 @@ int check_and_do_in_subquery_rewrites(JOIN *join) ...@@ -134,7 +131,13 @@ int check_and_do_in_subquery_rewrites(JOIN *join)
DBUG_RETURN(-1); /* purecov: deadcode */ DBUG_RETURN(-1); /* purecov: deadcode */
} }
if (select_lex == parent_unit->fake_select_lex) if (select_lex == parent_unit->fake_select_lex)
{
/*
The join and its select_lex object represent the 'fake' select used
to compute the result of a UNION.
*/
DBUG_RETURN(0); DBUG_RETURN(0);
}
DBUG_PRINT("info", ("Checking if subq can be converted to semi-join")); DBUG_PRINT("info", ("Checking if subq can be converted to semi-join"));
/* /*
...@@ -528,6 +531,15 @@ skip_conversion: ...@@ -528,6 +531,15 @@ skip_conversion:
FALSE)) FALSE))
DBUG_RETURN(TRUE); DBUG_RETURN(TRUE);
} }
/*
Revert to the IN->EXISTS strategy in the rare case when the subquery could
be flattened.
TODO: This is a limitation done for simplicity. Such subqueries could also
be executed via materialization. In order to determine this, we should
re-run the test for materialization that was done in
check_and_do_in_subquery_rewrites.
*/
(*in_subq)->exec_method= Item_in_subselect::IN_TO_EXISTS;
} }
if (arena) if (arena)
...@@ -3524,6 +3536,34 @@ static void remove_subq_pushed_predicates(JOIN *join, Item **where) ...@@ -3524,6 +3536,34 @@ static void remove_subq_pushed_predicates(JOIN *join, Item **where)
} }
/**
Setup for execution all subqueries of a query, for which the optimizer
chose hash semi-join.
@details Iterate over all immediate child subqueries of the query, and if
they are under an IN predicate, and the optimizer chose to compute it via
materialization:
- optimize each subquery,
- choose an optimial execution strategy for the IN predicate - either
materialization, or an IN=>EXISTS transformation with an approriate
engine.
This phase must be called after substitute_for_best_equal_field() because
that function may replace items with other items from a multiple equality,
and we need to reference the correct items in the index access method of the
IN predicate.
@return Operation status
@retval FALSE success.
@retval TRUE error occurred.
*/
bool JOIN::optimize_unflattened_subqueries()
{
return select_lex->optimize_unflattened_subqueries();
}
bool JOIN::choose_subquery_plan() bool JOIN::choose_subquery_plan()
{ {
double mat_strategy_cost; /* The cost to compute IN via materialization. */ double mat_strategy_cost; /* The cost to compute IN via materialization. */
...@@ -3564,7 +3604,13 @@ bool JOIN::choose_subquery_plan() ...@@ -3564,7 +3604,13 @@ bool JOIN::choose_subquery_plan()
} }
} }
else else
in_subs->exec_method= Item_in_subselect::IN_TO_EXISTS; {
/*
Previous optimizer phases should have chosen either a materialization
or IN->EXISTS strategy.
*/
DBUG_ASSERT(in_subs->exec_method == Item_in_subselect::IN_TO_EXISTS);
}
if (in_subs->exec_method == Item_in_subselect::MATERIALIZATION) if (in_subs->exec_method == Item_in_subselect::MATERIALIZATION)
{ {
......
...@@ -2618,34 +2618,6 @@ err: ...@@ -2618,34 +2618,6 @@ err:
} }
/**
Setup for execution all subqueries of a query, for which the optimizer
chose hash semi-join.
@details Iterate over all immediate child subqueries of the query, and if
they are under an IN predicate, and the optimizer chose to compute it via
materialization:
- optimize each subquery,
- choose an optimial execution strategy for the IN predicate - either
materialization, or an IN=>EXISTS transformation with an approriate
engine.
This phase must be called after substitute_for_best_equal_field() because
that function may replace items with other items from a multiple equality,
and we need to reference the correct items in the index access method of the
IN predicate.
@return Operation status
@retval FALSE success.
@retval TRUE error occurred.
*/
bool JOIN::optimize_unflattened_subqueries()
{
return select_lex->optimize_unflattened_subqueries();
}
/***************************************************************************** /*****************************************************************************
Create JOIN_TABS, make a guess about the table types, Create JOIN_TABS, make a guess about the table types,
Approximate how many records will be used in each table Approximate how many records will be used in each table
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment