1996-08-28 03:59:28 +02:00
|
|
|
/*-------------------------------------------------------------------------
|
|
|
|
*
|
1999-02-14 00:22:53 +01:00
|
|
|
* relation.h
|
2002-11-06 01:00:45 +01:00
|
|
|
* Definitions for planner's internal data structures.
|
1996-08-28 03:59:28 +02:00
|
|
|
*
|
|
|
|
*
|
2002-06-20 22:29:54 +02:00
|
|
|
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
|
2000-01-26 06:58:53 +01:00
|
|
|
* Portions Copyright (c) 1994, Regents of the University of California
|
1996-08-28 03:59:28 +02:00
|
|
|
*
|
2002-12-12 16:49:42 +01:00
|
|
|
* $Id: relation.h,v 1.74 2002/12/12 15:49:40 tgl Exp $
|
1996-08-28 03:59:28 +02:00
|
|
|
*
|
|
|
|
*-------------------------------------------------------------------------
|
|
|
|
*/
|
|
|
|
#ifndef RELATION_H
|
|
|
|
#define RELATION_H
|
|
|
|
|
2000-02-15 21:49:31 +01:00
|
|
|
#include "access/sdir.h"
|
1999-07-16 01:04:24 +02:00
|
|
|
#include "nodes/parsenodes.h"
|
1996-08-28 03:59:28 +02:00
|
|
|
|
|
|
|
/*
|
1999-02-18 01:49:48 +01:00
|
|
|
* Relids
|
1997-09-07 07:04:48 +02:00
|
|
|
* List of relation identifiers (indexes into the rangetable).
|
1999-08-16 04:17:58 +02:00
|
|
|
*
|
|
|
|
* Note: these are lists of integers, not Nodes.
|
1996-08-28 03:59:28 +02:00
|
|
|
*/
|
|
|
|
|
1999-02-18 01:49:48 +01:00
|
|
|
typedef List *Relids;
|
1996-08-28 03:59:28 +02:00
|
|
|
|
2000-02-15 21:49:31 +01:00
|
|
|
/*
|
|
|
|
* When looking for a "cheapest path", this enum specifies whether we want
|
|
|
|
* cheapest startup cost or cheapest total cost.
|
|
|
|
*/
|
2000-04-12 19:17:23 +02:00
|
|
|
typedef enum CostSelector
|
|
|
|
{
|
2001-10-28 07:26:15 +01:00
|
|
|
STARTUP_COST, TOTAL_COST
|
2000-04-12 19:17:23 +02:00
|
|
|
} CostSelector;
|
2000-02-15 21:49:31 +01:00
|
|
|
|
2000-09-29 20:21:41 +02:00
|
|
|
/*----------
|
1998-07-18 06:22:52 +02:00
|
|
|
* RelOptInfo
|
1999-08-16 04:17:58 +02:00
|
|
|
* Per-relation information for planning/optimization
|
1996-08-28 03:59:28 +02:00
|
|
|
*
|
2002-03-12 01:52:10 +01:00
|
|
|
* For planning purposes, a "base rel" is either a plain relation (a table)
|
|
|
|
* or the output of a sub-SELECT that appears in the range table.
|
|
|
|
* In either case it is uniquely identified by an RT index. A "joinrel"
|
|
|
|
* is the joining of two or more base rels. A joinrel is identified by
|
|
|
|
* the set of RT indexes for its component baserels. We create RelOptInfo
|
|
|
|
* nodes for each baserel and joinrel, and store them in the Query's
|
|
|
|
* base_rel_list and join_rel_list respectively.
|
2000-09-29 20:21:41 +02:00
|
|
|
*
|
2002-03-12 01:52:10 +01:00
|
|
|
* Note that there is only one joinrel for any given set of component
|
|
|
|
* baserels, no matter what order we assemble them in; so an unordered
|
|
|
|
* set is the right datatype to identify it with.
|
2000-09-29 20:21:41 +02:00
|
|
|
*
|
2002-03-12 01:52:10 +01:00
|
|
|
* We also have "other rels", which are like base rels in that they refer to
|
|
|
|
* single RT indexes; but they are not part of the join tree, and are stored
|
|
|
|
* in other_rel_list not base_rel_list. An otherrel is created for each
|
|
|
|
* join RTE as an aid in processing Vars that refer to the join's outputs,
|
2002-09-04 22:31:48 +02:00
|
|
|
* but it serves no other purpose in planning. It is important not to
|
2002-03-12 01:52:10 +01:00
|
|
|
* confuse this otherrel with the joinrel that represents the matching set
|
|
|
|
* of base relations.
|
|
|
|
*
|
|
|
|
* A second category of otherrels are those made for child relations of an
|
|
|
|
* inheritance scan (SELECT FROM foo*). The parent table's RTE and
|
|
|
|
* corresponding baserel represent the whole result of the inheritance scan.
|
|
|
|
* The planner creates separate RTEs and associated RelOptInfos for each child
|
|
|
|
* table (including the parent table, in its capacity as a member of the
|
|
|
|
* inheritance set). These RelOptInfos are physically identical to baserels,
|
|
|
|
* but are otherrels because they are not in the main join tree. These added
|
|
|
|
* RTEs and otherrels are used to plan the scans of the individual tables in
|
|
|
|
* the inheritance set; then the parent baserel is given an Append plan
|
|
|
|
* comprising the best plans for the individual child tables.
|
|
|
|
*
|
|
|
|
* Parts of this data structure are specific to various scan and join
|
2002-09-04 22:31:48 +02:00
|
|
|
* mechanisms. It didn't seem worth creating new node types for them.
|
1996-08-28 03:59:28 +02:00
|
|
|
*
|
1999-08-16 04:17:58 +02:00
|
|
|
* relids - List of base-relation identifiers; it is a base relation
|
|
|
|
* if there is just one, a join relation if more than one
|
2000-01-09 01:26:47 +01:00
|
|
|
* rows - estimated number of tuples in the relation after restriction
|
|
|
|
* clauses have been applied (ie, output rows of a plan for it)
|
1999-08-16 04:17:58 +02:00
|
|
|
* width - avg. number of bytes per tuple in the relation after the
|
2000-01-09 01:26:47 +01:00
|
|
|
* appropriate projections have been done (ie, output width)
|
2000-02-15 21:49:31 +01:00
|
|
|
* targetlist - List of TargetEntry nodes for the attributes we need
|
|
|
|
* to output from this relation
|
1999-08-16 04:17:58 +02:00
|
|
|
* pathlist - List of Path nodes, one for each potentially useful
|
|
|
|
* method of generating the relation
|
2000-02-15 21:49:31 +01:00
|
|
|
* cheapest_startup_path - the pathlist member with lowest startup cost
|
|
|
|
* (regardless of its ordering)
|
|
|
|
* cheapest_total_path - the pathlist member with lowest total cost
|
|
|
|
* (regardless of its ordering)
|
1999-08-16 04:17:58 +02:00
|
|
|
* pruneable - flag to let the planner know whether it can prune the
|
|
|
|
* pathlist of this RelOptInfo or not.
|
1996-08-28 03:59:28 +02:00
|
|
|
*
|
2002-03-12 01:52:10 +01:00
|
|
|
* If the relation is a base relation it will have these fields set:
|
1999-08-16 04:17:58 +02:00
|
|
|
*
|
2002-05-12 22:10:05 +02:00
|
|
|
* rtekind - distinguishes plain relation, subquery, or function RTE
|
2001-05-20 22:28:20 +02:00
|
|
|
* indexlist - list of IndexOptInfo nodes for relation's indexes
|
2002-05-12 22:10:05 +02:00
|
|
|
* (always NIL if it's not a table)
|
|
|
|
* pages - number of disk pages in relation (zero if not a table)
|
2000-01-09 01:26:47 +01:00
|
|
|
* tuples - number of tuples in relation (not considering restrictions)
|
2002-05-12 22:10:05 +02:00
|
|
|
* subplan - plan for subquery (NULL if it's not a subquery)
|
2000-09-29 20:21:41 +02:00
|
|
|
*
|
|
|
|
* Note: for a subquery, tuples and subplan are not set immediately
|
|
|
|
* upon creation of the RelOptInfo object; they are filled in when
|
|
|
|
* set_base_rel_pathlist processes the object.
|
1999-08-16 04:17:58 +02:00
|
|
|
*
|
2002-03-12 01:52:10 +01:00
|
|
|
* For otherrels that are inheritance children, these fields are filled
|
|
|
|
* in just as for a baserel. In otherrels for join RTEs, these fields
|
|
|
|
* are empty --- the only useful field of a join otherrel is its
|
|
|
|
* outerjoinset.
|
|
|
|
*
|
|
|
|
* If the relation is a join relation it will have these fields set:
|
2000-11-12 01:37:02 +01:00
|
|
|
*
|
2002-03-12 01:52:10 +01:00
|
|
|
* joinrti - RT index of corresponding JOIN RTE, if any; 0 if none
|
|
|
|
* joinrteids - List of RT indexes of JOIN RTEs included in this join
|
|
|
|
* (including joinrti)
|
|
|
|
*
|
|
|
|
* The presence of the remaining fields depends on the restrictions
|
|
|
|
* and joins that the relation participates in:
|
1996-08-28 03:59:28 +02:00
|
|
|
*
|
2000-02-07 05:41:04 +01:00
|
|
|
* baserestrictinfo - List of RestrictInfo nodes, containing info about
|
|
|
|
* each qualification clause in which this relation
|
|
|
|
* participates (only used for base rels)
|
2000-02-15 21:49:31 +01:00
|
|
|
* baserestrictcost - Estimated cost of evaluating the baserestrictinfo
|
|
|
|
* clauses at a single tuple (only used for base rels)
|
2002-03-12 01:52:10 +01:00
|
|
|
* outerjoinset - For a base rel: if the rel appears within the nullable
|
|
|
|
* side of an outer join, the list of all relids
|
|
|
|
* participating in the highest such outer join; else NIL.
|
|
|
|
* For a join otherrel: the list of all baserel relids
|
2002-09-04 22:31:48 +02:00
|
|
|
* syntactically within the join. Otherwise, unused.
|
1997-09-07 07:04:48 +02:00
|
|
|
* joininfo - List of JoinInfo nodes, containing info about each join
|
|
|
|
* clause in which this relation participates
|
2002-11-24 22:52:15 +01:00
|
|
|
* index_outer_relids - only used for base rels; list of outer relids
|
|
|
|
* that participate in indexable joinclauses for this rel
|
|
|
|
* index_inner_paths - only used for base rels; list of InnerIndexscanInfo
|
|
|
|
* nodes showing best indexpaths for various subsets of
|
|
|
|
* index_outer_relids.
|
2000-02-07 05:41:04 +01:00
|
|
|
*
|
|
|
|
* Note: Keeping a restrictinfo list in the RelOptInfo is useful only for
|
|
|
|
* base rels, because for a join rel the set of clauses that are treated as
|
|
|
|
* restrict clauses varies depending on which sub-relations we choose to join.
|
|
|
|
* (For example, in a 3-base-rel join, a clause relating rels 1 and 2 must be
|
|
|
|
* treated as a restrictclause if we join {1} and {2 3} to make {1 2 3}; but
|
|
|
|
* if we join {1 2} and {3} then that clause will be a restrictclause in {1 2}
|
2000-04-12 19:17:23 +02:00
|
|
|
* and should not be processed again at the level of {1 2 3}.) Therefore,
|
2000-02-07 05:41:04 +01:00
|
|
|
* the restrictinfo list in the join case appears in individual JoinPaths
|
|
|
|
* (field joinrestrictinfo), not in the parent relation. But it's OK for
|
|
|
|
* the RelOptInfo to store the joininfo lists, because those are the same
|
|
|
|
* for a given rel no matter how we form it.
|
2000-02-15 21:49:31 +01:00
|
|
|
*
|
|
|
|
* We store baserestrictcost in the RelOptInfo (for base relations) because
|
|
|
|
* we know we will need it at least once (to price the sequential scan)
|
|
|
|
* and may need it multiple times to price index scans.
|
2000-09-12 23:07:18 +02:00
|
|
|
*
|
|
|
|
* outerjoinset is used to ensure correct placement of WHERE clauses that
|
|
|
|
* apply to outer-joined relations; we must not apply such WHERE clauses
|
|
|
|
* until after the outer join is performed.
|
2000-09-29 20:21:41 +02:00
|
|
|
*----------
|
1996-08-28 03:59:28 +02:00
|
|
|
*/
|
2002-03-12 01:52:10 +01:00
|
|
|
typedef enum RelOptKind
|
|
|
|
{
|
|
|
|
RELOPT_BASEREL,
|
|
|
|
RELOPT_JOINREL,
|
|
|
|
RELOPT_OTHER_JOIN_REL,
|
|
|
|
RELOPT_OTHER_CHILD_REL
|
|
|
|
} RelOptKind;
|
1996-08-28 03:59:28 +02:00
|
|
|
|
1998-07-18 06:22:52 +02:00
|
|
|
typedef struct RelOptInfo
|
1997-09-07 07:04:48 +02:00
|
|
|
{
|
1997-09-08 04:41:22 +02:00
|
|
|
NodeTag type;
|
1997-09-07 07:04:48 +02:00
|
|
|
|
2002-03-12 01:52:10 +01:00
|
|
|
RelOptKind reloptkind;
|
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
/* all relations included in this RelOptInfo */
|
2002-11-27 21:52:04 +01:00
|
|
|
Relids relids; /* integer list of base relids (rangetable
|
2000-04-12 19:17:23 +02:00
|
|
|
* indexes) */
|
1999-08-16 04:17:58 +02:00
|
|
|
|
2000-01-09 01:26:47 +01:00
|
|
|
/* size estimates generated by planner */
|
|
|
|
double rows; /* estimated number of result tuples */
|
|
|
|
int width; /* estimated avg width of result tuples */
|
1997-09-07 07:04:48 +02:00
|
|
|
|
|
|
|
/* materialization information */
|
1997-09-08 04:41:22 +02:00
|
|
|
List *targetlist;
|
1999-02-04 02:47:02 +01:00
|
|
|
List *pathlist; /* Path structures */
|
2000-02-15 21:49:31 +01:00
|
|
|
struct Path *cheapest_startup_path;
|
|
|
|
struct Path *cheapest_total_path;
|
1997-09-08 04:41:22 +02:00
|
|
|
bool pruneable;
|
1997-09-07 07:04:48 +02:00
|
|
|
|
2000-09-29 20:21:41 +02:00
|
|
|
/* information about a base rel (not set for join rels!) */
|
2002-05-12 22:10:05 +02:00
|
|
|
RTEKind rtekind; /* RELATION, SUBQUERY, or FUNCTION */
|
2001-05-20 22:28:20 +02:00
|
|
|
List *indexlist;
|
2000-01-09 01:26:47 +01:00
|
|
|
long pages;
|
|
|
|
double tuples;
|
2002-05-12 22:10:05 +02:00
|
|
|
struct Plan *subplan; /* if subquery */
|
1997-09-07 07:04:48 +02:00
|
|
|
|
2002-03-12 01:52:10 +01:00
|
|
|
/* information about a join rel (not set for base rels!) */
|
|
|
|
Index joinrti;
|
|
|
|
List *joinrteids;
|
|
|
|
|
1997-09-07 07:04:48 +02:00
|
|
|
/* used by various scans and joins: */
|
2000-04-12 19:17:23 +02:00
|
|
|
List *baserestrictinfo; /* RestrictInfo structures (if
|
|
|
|
* base rel) */
|
|
|
|
Cost baserestrictcost; /* cost of evaluating the above */
|
2001-03-22 05:01:46 +01:00
|
|
|
Relids outerjoinset; /* integer list of base relids */
|
1999-02-04 02:47:02 +01:00
|
|
|
List *joininfo; /* JoinInfo structures */
|
2000-04-12 19:17:23 +02:00
|
|
|
|
2002-11-24 22:52:15 +01:00
|
|
|
/* cached info about inner indexscan paths for relation: */
|
|
|
|
Relids index_outer_relids; /* other relids in indexable join
|
|
|
|
* clauses */
|
|
|
|
List *index_inner_paths; /* InnerIndexscanInfo nodes */
|
2000-04-12 19:17:23 +02:00
|
|
|
/*
|
2002-11-24 22:52:15 +01:00
|
|
|
* Inner indexscans are not in the main pathlist because they are
|
|
|
|
* not usable except in specific join contexts. We use the
|
|
|
|
* index_inner_paths list just to avoid recomputing the best inner
|
|
|
|
* indexscan repeatedly for similar outer relations. See comments
|
|
|
|
* for InnerIndexscanInfo.
|
1999-08-16 04:17:58 +02:00
|
|
|
*/
|
1999-05-26 00:43:53 +02:00
|
|
|
} RelOptInfo;
|
1997-09-07 07:04:48 +02:00
|
|
|
|
2000-01-09 01:26:47 +01:00
|
|
|
/*
|
|
|
|
* IndexOptInfo
|
|
|
|
* Per-index information for planning/optimization
|
|
|
|
*
|
|
|
|
* Prior to Postgres 7.0, RelOptInfo was used to describe both relations
|
|
|
|
* and indexes, but that created confusion without actually doing anything
|
|
|
|
* useful. So now we have a separate IndexOptInfo struct for indexes.
|
|
|
|
*
|
2001-05-20 22:28:20 +02:00
|
|
|
* ncolumns and nkeys are the same except for a functional index,
|
|
|
|
* wherein ncolumns is 1 (the single function output) while nkeys
|
|
|
|
* is the number of table columns passed to the function. classlist[]
|
|
|
|
* and ordering[] have ncolumns entries, while indexkeys[] has nkeys
|
|
|
|
* entries.
|
2001-10-25 07:50:21 +02:00
|
|
|
*
|
2001-05-20 22:28:20 +02:00
|
|
|
* Note: for historical reasons, the arrays classlist, indexkeys and
|
|
|
|
* ordering have an extra entry that is always zero. Some code scans
|
|
|
|
* until it sees a zero rather than looking at ncolumns or nkeys.
|
2000-01-09 01:26:47 +01:00
|
|
|
*/
|
|
|
|
|
|
|
|
typedef struct IndexOptInfo
|
|
|
|
{
|
|
|
|
NodeTag type;
|
|
|
|
|
|
|
|
Oid indexoid; /* OID of the index relation */
|
|
|
|
|
|
|
|
/* statistics from pg_class */
|
2002-11-24 22:52:15 +01:00
|
|
|
long pages; /* number of disk pages in index */
|
|
|
|
double tuples; /* number of index tuples in index */
|
2000-01-09 01:26:47 +01:00
|
|
|
|
|
|
|
/* index descriptor information */
|
2001-05-20 22:28:20 +02:00
|
|
|
int ncolumns; /* number of columns in index */
|
|
|
|
int nkeys; /* number of keys used by index */
|
2002-11-24 22:52:15 +01:00
|
|
|
Oid *classlist; /* OIDs of operator classes for columns */
|
2001-05-20 22:28:20 +02:00
|
|
|
int *indexkeys; /* column numbers of index's keys */
|
|
|
|
Oid *ordering; /* OIDs of sort operators for each column */
|
2000-01-09 01:26:47 +01:00
|
|
|
Oid relam; /* OID of the access method (in pg_am) */
|
|
|
|
|
2001-10-28 07:26:15 +01:00
|
|
|
RegProcedure amcostestimate; /* OID of the access method's cost fcn */
|
2000-01-23 00:50:30 +01:00
|
|
|
|
2002-11-24 22:52:15 +01:00
|
|
|
Oid indproc; /* OID of func if functional index, else 0 */
|
|
|
|
List *indpred; /* predicate if a partial index, else NIL */
|
|
|
|
bool unique; /* true if a unique index */
|
|
|
|
|
|
|
|
/* cached info about inner indexscan paths for index */
|
|
|
|
Relids outer_relids; /* other relids in usable join clauses */
|
|
|
|
List *inner_paths; /* List of InnerIndexscanInfo nodes */
|
2000-01-09 01:26:47 +01:00
|
|
|
} IndexOptInfo;
|
|
|
|
|
2002-03-12 01:52:10 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
* A Var is considered to belong to a relation if it's either from one
|
|
|
|
* of the actual base rels making up the relation, or it's a join alias
|
|
|
|
* var that is included in the relation.
|
|
|
|
*/
|
|
|
|
#define VARISRELMEMBER(varno,rel) (intMember((varno), (rel)->relids) || \
|
|
|
|
intMember((varno), (rel)->joinrteids))
|
|
|
|
|
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
/*
|
|
|
|
* PathKeys
|
|
|
|
*
|
|
|
|
* The sort ordering of a path is represented by a list of sublists of
|
2000-04-12 19:17:23 +02:00
|
|
|
* PathKeyItem nodes. An empty list implies no known ordering. Otherwise
|
1999-08-16 04:17:58 +02:00
|
|
|
* the first sublist represents the primary sort key, the second the
|
2000-04-12 19:17:23 +02:00
|
|
|
* first secondary sort key, etc. Each sublist contains one or more
|
1999-08-16 04:17:58 +02:00
|
|
|
* PathKeyItem nodes, each of which can be taken as the attribute that
|
2000-04-12 19:17:23 +02:00
|
|
|
* appears at that sort position. (See the top of optimizer/path/pathkeys.c
|
1999-08-16 04:17:58 +02:00
|
|
|
* for more information.)
|
|
|
|
*/
|
1997-09-07 07:04:48 +02:00
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
typedef struct PathKeyItem
|
1997-09-07 07:04:48 +02:00
|
|
|
{
|
1997-09-08 04:41:22 +02:00
|
|
|
NodeTag type;
|
1997-09-07 07:04:48 +02:00
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
Node *key; /* the item that is ordered */
|
|
|
|
Oid sortop; /* the ordering operator ('<' op) */
|
2000-04-12 19:17:23 +02:00
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
/*
|
2000-04-12 19:17:23 +02:00
|
|
|
* key typically points to a Var node, ie a relation attribute, but it
|
2002-12-12 16:49:42 +01:00
|
|
|
* can also point to a FuncExpr clause representing the value indexed by a
|
2000-04-12 19:17:23 +02:00
|
|
|
* functional index. Someday we might allow arbitrary expressions as
|
|
|
|
* path keys, so don't assume more than you must.
|
1999-08-16 04:17:58 +02:00
|
|
|
*/
|
|
|
|
} PathKeyItem;
|
1999-02-09 04:51:42 +01:00
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
/*
|
|
|
|
* Type "Path" is used as-is for sequential-scan paths. For other
|
|
|
|
* path types it is the first component of a larger struct.
|
2002-11-27 21:52:04 +01:00
|
|
|
*
|
|
|
|
* Note: "pathtype" is the NodeTag of the Plan node we could build from this
|
|
|
|
* Path. It is partially redundant with the Path's NodeTag, but allows us
|
|
|
|
* to use the same Path type for multiple Plan types where there is no need
|
|
|
|
* to distinguish the Plan type during path processing.
|
1999-08-16 04:17:58 +02:00
|
|
|
*/
|
1997-09-07 07:04:48 +02:00
|
|
|
|
|
|
|
typedef struct Path
|
|
|
|
{
|
1997-09-08 04:41:22 +02:00
|
|
|
NodeTag type;
|
1997-09-07 07:04:48 +02:00
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
RelOptInfo *parent; /* the relation this path can build */
|
1997-09-07 07:04:48 +02:00
|
|
|
|
2000-02-15 21:49:31 +01:00
|
|
|
/* estimated execution costs for path (see costsize.c for more info) */
|
2000-04-12 19:17:23 +02:00
|
|
|
Cost startup_cost; /* cost expended before fetching any
|
|
|
|
* tuples */
|
|
|
|
Cost total_cost; /* total cost (assuming all tuples
|
|
|
|
* fetched) */
|
1997-09-07 07:04:48 +02:00
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
NodeTag pathtype; /* tag identifying scan/join method */
|
1997-09-07 07:04:48 +02:00
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
List *pathkeys; /* sort ordering of path's output */
|
|
|
|
/* pathkeys is a List of Lists of PathKeyItem nodes; see above */
|
1997-09-08 23:56:23 +02:00
|
|
|
} Path;
|
1997-09-07 07:04:48 +02:00
|
|
|
|
1999-07-25 01:21:14 +02:00
|
|
|
/*----------
|
2000-04-12 19:17:23 +02:00
|
|
|
* IndexPath represents an index scan. Although an indexscan can only read
|
1999-07-25 01:21:14 +02:00
|
|
|
* a single relation, it can scan it more than once, potentially using a
|
|
|
|
* different index during each scan. The result is the union (OR) of all the
|
2000-04-12 19:17:23 +02:00
|
|
|
* tuples matched during any scan. (The executor is smart enough not to return
|
1999-07-25 01:21:14 +02:00
|
|
|
* the same tuple more than once, even if it is matched in multiple scans.)
|
1999-08-16 04:17:58 +02:00
|
|
|
*
|
2001-05-20 22:28:20 +02:00
|
|
|
* 'indexinfo' is a list of IndexOptInfo nodes, one per scan to be performed.
|
2000-02-15 21:49:31 +01:00
|
|
|
*
|
1999-07-25 01:21:14 +02:00
|
|
|
* 'indexqual' is a list of index qualifications, also one per scan.
|
|
|
|
* Each entry in 'indexqual' is a sublist of qualification expressions with
|
1999-08-16 04:17:58 +02:00
|
|
|
* implicit AND semantics across the sublist items. Only expressions that
|
|
|
|
* are usable as indexquals (as determined by indxpath.c) may appear here.
|
1999-07-25 01:21:14 +02:00
|
|
|
* NOTE that the semantics of the top-level list in 'indexqual' is OR
|
|
|
|
* combination, while the sublists are implicitly AND combinations!
|
2000-02-19 00:47:31 +01:00
|
|
|
* Also note that indexquals lists do not contain RestrictInfo nodes,
|
|
|
|
* just bare clause expressions.
|
2000-02-15 21:49:31 +01:00
|
|
|
*
|
|
|
|
* 'indexscandir' is one of:
|
|
|
|
* ForwardScanDirection: forward scan of an ordered index
|
|
|
|
* BackwardScanDirection: backward scan of an ordered index
|
|
|
|
* NoMovementScanDirection: scan of an unordered index, or don't care
|
|
|
|
* (The executor doesn't care whether it gets ForwardScanDirection or
|
|
|
|
* NoMovementScanDirection for an indexscan, but the planner wants to
|
|
|
|
* distinguish ordered from unordered indexes for building pathkeys.)
|
|
|
|
*
|
2000-03-22 23:08:35 +01:00
|
|
|
* 'rows' is the estimated result tuple count for the indexscan. This
|
|
|
|
* is the same as path.parent->rows for a simple indexscan, but it is
|
2002-11-24 22:52:15 +01:00
|
|
|
* different for a nestloop inner scan, because the additional indexquals
|
2000-03-22 23:08:35 +01:00
|
|
|
* coming from join clauses make the scan more selective than the parent
|
|
|
|
* rel's restrict clauses alone would do.
|
1999-07-25 01:21:14 +02:00
|
|
|
*----------
|
|
|
|
*/
|
1997-09-07 07:04:48 +02:00
|
|
|
typedef struct IndexPath
|
|
|
|
{
|
1997-09-08 04:41:22 +02:00
|
|
|
Path path;
|
2001-05-20 22:28:20 +02:00
|
|
|
List *indexinfo;
|
1997-09-08 04:41:22 +02:00
|
|
|
List *indexqual;
|
2000-02-15 21:49:31 +01:00
|
|
|
ScanDirection indexscandir;
|
2000-03-22 23:08:35 +01:00
|
|
|
double rows; /* estimated number of result tuples */
|
1997-09-08 23:56:23 +02:00
|
|
|
} IndexPath;
|
1997-09-07 07:04:48 +02:00
|
|
|
|
2000-11-12 01:37:02 +01:00
|
|
|
/*
|
|
|
|
* TidPath represents a scan by TID
|
|
|
|
*/
|
1999-11-23 21:07:06 +01:00
|
|
|
typedef struct TidPath
|
|
|
|
{
|
2000-04-12 19:17:23 +02:00
|
|
|
Path path;
|
2002-12-05 16:50:39 +01:00
|
|
|
List *tideval; /* qual(s) involving CTID = something */
|
2000-04-12 19:17:23 +02:00
|
|
|
} TidPath;
|
1999-11-23 21:07:06 +01:00
|
|
|
|
2000-11-12 01:37:02 +01:00
|
|
|
/*
|
|
|
|
* AppendPath represents an Append plan, ie, successive execution of
|
|
|
|
* several member plans. Currently it is only used to handle expansion
|
|
|
|
* of inheritance trees.
|
|
|
|
*/
|
|
|
|
typedef struct AppendPath
|
|
|
|
{
|
|
|
|
Path path;
|
|
|
|
List *subpaths; /* list of component Paths */
|
|
|
|
} AppendPath;
|
|
|
|
|
2002-11-06 01:00:45 +01:00
|
|
|
/*
|
|
|
|
* ResultPath represents use of a Result plan node, either to compute a
|
|
|
|
* variable-free targetlist or to gate execution of a subplan with a
|
|
|
|
* one-time (variable-free) qual condition. Note that in the former case
|
|
|
|
* path.parent will be NULL; in the latter case it is copied from the subpath.
|
|
|
|
*/
|
|
|
|
typedef struct ResultPath
|
|
|
|
{
|
|
|
|
Path path;
|
|
|
|
Path *subpath;
|
|
|
|
List *constantqual;
|
|
|
|
} ResultPath;
|
|
|
|
|
2002-11-30 06:21:03 +01:00
|
|
|
/*
|
|
|
|
* MaterialPath represents use of a Material plan node, i.e., caching of
|
|
|
|
* the output of its subpath. This is used when the subpath is expensive
|
|
|
|
* and needs to be scanned repeatedly, or when we need mark/restore ability
|
|
|
|
* and the subpath doesn't have it.
|
|
|
|
*/
|
|
|
|
typedef struct MaterialPath
|
|
|
|
{
|
|
|
|
Path path;
|
|
|
|
Path *subpath;
|
|
|
|
} MaterialPath;
|
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
/*
|
|
|
|
* All join-type paths share these fields.
|
|
|
|
*/
|
|
|
|
|
|
|
|
typedef struct JoinPath
|
1997-09-07 07:04:48 +02:00
|
|
|
{
|
1997-09-08 04:41:22 +02:00
|
|
|
Path path;
|
1997-09-07 07:04:48 +02:00
|
|
|
|
2000-09-12 23:07:18 +02:00
|
|
|
JoinType jointype;
|
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
Path *outerjoinpath; /* path for the outer side of the join */
|
|
|
|
Path *innerjoinpath; /* path for the inner side of the join */
|
2000-09-12 23:07:18 +02:00
|
|
|
|
2000-04-12 19:17:23 +02:00
|
|
|
List *joinrestrictinfo; /* RestrictInfos to apply to join */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* See the notes for RelOptInfo to understand why joinrestrictinfo is
|
2000-02-07 05:41:04 +01:00
|
|
|
* needed in JoinPath, and can't be merged into the parent RelOptInfo.
|
|
|
|
*/
|
1999-08-16 04:17:58 +02:00
|
|
|
} JoinPath;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* A nested-loop path needs no special fields.
|
|
|
|
*/
|
|
|
|
|
|
|
|
typedef JoinPath NestPath;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* A mergejoin path has these fields.
|
|
|
|
*
|
2000-02-19 00:47:31 +01:00
|
|
|
* path_mergeclauses lists the clauses (in the form of RestrictInfos)
|
2000-12-13 00:33:34 +01:00
|
|
|
* that will be used in the merge. (Before 7.0, this was a list of bare
|
|
|
|
* clause expressions, but we can save on list memory and cost_qual_eval
|
|
|
|
* work by leaving it in the form of a RestrictInfo list.)
|
2000-02-19 00:47:31 +01:00
|
|
|
*
|
1999-08-16 04:17:58 +02:00
|
|
|
* Note that the mergeclauses are a subset of the parent relation's
|
|
|
|
* restriction-clause list. Any join clauses that are not mergejoinable
|
|
|
|
* appear only in the parent's restrict list, and must be checked by a
|
|
|
|
* qpqual at execution time.
|
2000-02-19 00:47:31 +01:00
|
|
|
*
|
|
|
|
* outersortkeys (resp. innersortkeys) is NIL if the outer path
|
|
|
|
* (resp. inner path) is already ordered appropriately for the
|
|
|
|
* mergejoin. If it is not NIL then it is a PathKeys list describing
|
|
|
|
* the ordering that must be created by an explicit sort step.
|
1999-08-16 04:17:58 +02:00
|
|
|
*/
|
1999-02-12 18:25:05 +01:00
|
|
|
|
1997-09-07 07:04:48 +02:00
|
|
|
typedef struct MergePath
|
|
|
|
{
|
1999-02-12 18:25:05 +01:00
|
|
|
JoinPath jpath;
|
2000-04-12 19:17:23 +02:00
|
|
|
List *path_mergeclauses; /* join clauses to be used for
|
|
|
|
* merge */
|
2000-02-19 00:47:31 +01:00
|
|
|
List *outersortkeys; /* keys for explicit sort, if any */
|
|
|
|
List *innersortkeys; /* keys for explicit sort, if any */
|
1997-09-08 23:56:23 +02:00
|
|
|
} MergePath;
|
1997-09-07 07:04:48 +02:00
|
|
|
|
1999-02-22 20:55:44 +01:00
|
|
|
/*
|
1999-08-16 04:17:58 +02:00
|
|
|
* A hashjoin path has these fields.
|
|
|
|
*
|
|
|
|
* The remarks above for mergeclauses apply for hashclauses as well.
|
2000-02-19 00:47:31 +01:00
|
|
|
*
|
|
|
|
* Hashjoin does not care what order its inputs appear in, so we have
|
|
|
|
* no need for sortkeys.
|
1999-02-22 20:55:44 +01:00
|
|
|
*/
|
1996-08-28 03:59:28 +02:00
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
typedef struct HashPath
|
1997-09-07 07:04:48 +02:00
|
|
|
{
|
1999-08-16 04:17:58 +02:00
|
|
|
JoinPath jpath;
|
2000-04-12 19:17:23 +02:00
|
|
|
List *path_hashclauses; /* join clauses used for hashing */
|
1999-08-16 04:17:58 +02:00
|
|
|
} HashPath;
|
1996-08-28 03:59:28 +02:00
|
|
|
|
1999-02-22 20:55:44 +01:00
|
|
|
/*
|
1999-07-25 01:21:14 +02:00
|
|
|
* Restriction clause info.
|
1999-08-16 04:17:58 +02:00
|
|
|
*
|
1999-07-25 01:21:14 +02:00
|
|
|
* We create one of these for each AND sub-clause of a restriction condition
|
2000-09-29 20:21:41 +02:00
|
|
|
* (WHERE or JOIN/ON clause). Since the restriction clauses are logically
|
|
|
|
* ANDed, we can use any one of them or any subset of them to filter out
|
|
|
|
* tuples, without having to evaluate the rest. The RestrictInfo node itself
|
|
|
|
* stores data used by the optimizer while choosing the best query plan.
|
1999-08-16 04:17:58 +02:00
|
|
|
*
|
2000-02-07 05:41:04 +01:00
|
|
|
* If a restriction clause references a single base relation, it will appear
|
|
|
|
* in the baserestrictinfo list of the RelOptInfo for that base rel.
|
1999-08-16 04:17:58 +02:00
|
|
|
*
|
2000-02-07 05:41:04 +01:00
|
|
|
* If a restriction clause references more than one base rel, it will
|
|
|
|
* appear in the JoinInfo lists of every RelOptInfo that describes a strict
|
1999-08-16 04:17:58 +02:00
|
|
|
* subset of the base rels mentioned in the clause. The JoinInfo lists are
|
|
|
|
* used to drive join tree building by selecting plausible join candidates.
|
2000-02-07 05:41:04 +01:00
|
|
|
* The clause cannot actually be applied until we have built a join rel
|
|
|
|
* containing all the base rels it references, however.
|
|
|
|
*
|
2000-09-12 23:07:18 +02:00
|
|
|
* When we construct a join rel that includes all the base rels referenced
|
|
|
|
* in a multi-relation restriction clause, we place that clause into the
|
|
|
|
* joinrestrictinfo lists of paths for the join rel, if neither left nor
|
2001-03-22 05:01:46 +01:00
|
|
|
* right sub-path includes all base rels referenced in the clause. The clause
|
2000-09-12 23:07:18 +02:00
|
|
|
* will be applied at that join level, and will not propagate any further up
|
|
|
|
* the join tree. (Note: the "predicate migration" code was once intended to
|
2000-02-07 05:41:04 +01:00
|
|
|
* push restriction clauses up and down the plan tree based on evaluation
|
|
|
|
* costs, but it's dead code and is unlikely to be resurrected in the
|
|
|
|
* foreseeable future.)
|
|
|
|
*
|
|
|
|
* Note that in the presence of more than two rels, a multi-rel restriction
|
|
|
|
* might reach different heights in the join tree depending on the join
|
|
|
|
* sequence we use. So, these clauses cannot be associated directly with
|
|
|
|
* the join RelOptInfo, but must be kept track of on a per-join-path basis.
|
1999-08-16 04:17:58 +02:00
|
|
|
*
|
2000-09-29 20:21:41 +02:00
|
|
|
* When dealing with outer joins we have to be very careful about pushing qual
|
|
|
|
* clauses up and down the tree. An outer join's own JOIN/ON conditions must
|
|
|
|
* be evaluated exactly at that join node, and any quals appearing in WHERE or
|
|
|
|
* in a JOIN above the outer join cannot be pushed down below the outer join.
|
|
|
|
* Otherwise the outer join will produce wrong results because it will see the
|
|
|
|
* wrong sets of input rows. All quals are stored as RestrictInfo nodes
|
|
|
|
* during planning, but there's a flag to indicate whether a qual has been
|
|
|
|
* pushed down to a lower level than its original syntactic placement in the
|
|
|
|
* join tree would suggest. If an outer join prevents us from pushing a qual
|
|
|
|
* down to its "natural" semantic level (the level associated with just the
|
|
|
|
* base rels used in the qual) then the qual will appear in JoinInfo lists
|
|
|
|
* that reference more than just the base rels it actually uses. By
|
|
|
|
* pretending that the qual references all the rels appearing in the outer
|
|
|
|
* join, we prevent it from being evaluated below the outer join's joinrel.
|
|
|
|
* When we do form the outer join's joinrel, we still need to distinguish
|
|
|
|
* those quals that are actually in that join's JOIN/ON condition from those
|
|
|
|
* that appeared higher in the tree and were pushed down to the join rel
|
|
|
|
* because they used no other rels. That's what the ispusheddown flag is for;
|
|
|
|
* it tells us that a qual came from a point above the join of the specific
|
|
|
|
* set of base rels that it uses (or that the JoinInfo structures claim it
|
|
|
|
* uses). A clause that originally came from WHERE will *always* have its
|
|
|
|
* ispusheddown flag set; a clause that came from an INNER JOIN condition,
|
|
|
|
* but doesn't use all the rels being joined, will also have ispusheddown set
|
|
|
|
* because it will get attached to some lower joinrel.
|
|
|
|
*
|
2000-04-12 19:17:23 +02:00
|
|
|
* In general, the referenced clause might be arbitrarily complex. The
|
1999-08-16 04:17:58 +02:00
|
|
|
* kinds of clauses we can handle as indexscan quals, mergejoin clauses,
|
|
|
|
* or hashjoin clauses are fairly limited --- the code for each kind of
|
|
|
|
* path is responsible for identifying the restrict clauses it can use
|
|
|
|
* and ignoring the rest. Clauses not implemented by an indexscan,
|
2000-09-12 23:07:18 +02:00
|
|
|
* mergejoin, or hashjoin will be placed in the plan qual or joinqual field
|
2001-06-05 07:26:05 +02:00
|
|
|
* of the finished Plan node, where they will be enforced by general-purpose
|
1999-08-16 04:17:58 +02:00
|
|
|
* qual-expression-evaluation code. (But we are still entitled to count
|
|
|
|
* their selectivity when estimating the result tuple count, if we
|
|
|
|
* can guess what it is...)
|
1999-02-22 20:55:44 +01:00
|
|
|
*/
|
1996-08-28 03:59:28 +02:00
|
|
|
|
1999-02-03 21:15:53 +01:00
|
|
|
typedef struct RestrictInfo
|
1997-09-07 07:04:48 +02:00
|
|
|
{
|
1997-09-08 04:41:22 +02:00
|
|
|
NodeTag type;
|
1997-09-07 07:04:48 +02:00
|
|
|
|
2000-09-12 23:07:18 +02:00
|
|
|
Expr *clause; /* the represented clause of WHERE or JOIN */
|
|
|
|
|
2000-09-29 20:21:41 +02:00
|
|
|
bool ispusheddown; /* TRUE if clause was pushed down in level */
|
1997-09-07 07:04:48 +02:00
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
/* only used if clause is an OR clause: */
|
2000-04-12 19:17:23 +02:00
|
|
|
List *subclauseindices; /* indexes matching subclauses */
|
2000-01-09 01:26:47 +01:00
|
|
|
/* subclauseindices is a List of Lists of IndexOptInfos */
|
1997-09-07 07:04:48 +02:00
|
|
|
|
2001-06-05 07:26:05 +02:00
|
|
|
/* cache space for costs (currently only used for join clauses) */
|
|
|
|
Cost eval_cost; /* eval cost of clause; -1 if not yet set */
|
2001-10-25 07:50:21 +02:00
|
|
|
Selectivity this_selec; /* selectivity; -1 if not yet set */
|
2001-06-05 07:26:05 +02:00
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
/* valid if clause is mergejoinable, else InvalidOid: */
|
2000-04-12 19:17:23 +02:00
|
|
|
Oid mergejoinoperator; /* copy of clause operator */
|
1999-08-16 04:17:58 +02:00
|
|
|
Oid left_sortop; /* leftside sortop needed for mergejoin */
|
|
|
|
Oid right_sortop; /* rightside sortop needed for mergejoin */
|
1997-09-07 07:04:48 +02:00
|
|
|
|
2000-12-14 23:30:45 +01:00
|
|
|
/* cache space for mergeclause processing; NIL if not yet set */
|
|
|
|
List *left_pathkey; /* canonical pathkey for left side */
|
|
|
|
List *right_pathkey; /* canonical pathkey for right side */
|
|
|
|
|
2002-03-01 07:01:20 +01:00
|
|
|
/* cache space for mergeclause processing; -1 if not yet set */
|
2002-09-04 22:31:48 +02:00
|
|
|
Selectivity left_mergescansel; /* fraction of left side to scan */
|
|
|
|
Selectivity right_mergescansel; /* fraction of right side to scan */
|
2002-03-01 07:01:20 +01:00
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
/* valid if clause is hashjoinable, else InvalidOid: */
|
2000-04-12 19:17:23 +02:00
|
|
|
Oid hashjoinoperator; /* copy of clause operator */
|
2000-12-14 23:30:45 +01:00
|
|
|
|
|
|
|
/* cache space for hashclause processing; -1 if not yet set */
|
2001-10-28 07:26:15 +01:00
|
|
|
Selectivity left_bucketsize; /* avg bucketsize of left side */
|
2001-05-07 02:43:27 +02:00
|
|
|
Selectivity right_bucketsize; /* avg bucketsize of right side */
|
1999-08-16 04:17:58 +02:00
|
|
|
} RestrictInfo;
|
1997-09-07 07:04:48 +02:00
|
|
|
|
1999-08-16 04:17:58 +02:00
|
|
|
/*
|
|
|
|
* Join clause info.
|
|
|
|
*
|
|
|
|
* We make a list of these for each RelOptInfo, containing info about
|
|
|
|
* all the join clauses this RelOptInfo participates in. (For this
|
|
|
|
* purpose, a "join clause" is a WHERE clause that mentions both vars
|
|
|
|
* belonging to this relation and vars belonging to relations not yet
|
|
|
|
* joined to it.) We group these clauses according to the set of
|
|
|
|
* other base relations (unjoined relations) mentioned in them.
|
|
|
|
* There is one JoinInfo for each distinct set of unjoined_relids,
|
|
|
|
* and its jinfo_restrictinfo lists the clause(s) that use that set
|
|
|
|
* of other relations.
|
|
|
|
*/
|
1997-09-07 07:04:48 +02:00
|
|
|
|
1998-09-01 05:29:17 +02:00
|
|
|
typedef struct JoinInfo
|
1997-09-07 07:04:48 +02:00
|
|
|
{
|
1997-09-08 04:41:22 +02:00
|
|
|
NodeTag type;
|
2001-10-28 07:26:15 +01:00
|
|
|
Relids unjoined_relids; /* some rels not yet part of my RelOptInfo */
|
2000-04-12 19:17:23 +02:00
|
|
|
List *jinfo_restrictinfo; /* relevant RestrictInfos */
|
1999-05-26 00:43:53 +02:00
|
|
|
} JoinInfo;
|
1997-09-07 07:04:48 +02:00
|
|
|
|
2002-11-24 22:52:15 +01:00
|
|
|
/*
|
|
|
|
* Inner indexscan info.
|
|
|
|
*
|
|
|
|
* An inner indexscan is one that uses one or more joinclauses as index
|
|
|
|
* conditions (perhaps in addition to plain restriction clauses). So it
|
|
|
|
* can only be used as the inner path of a nestloop join where the outer
|
|
|
|
* relation includes all other relids appearing in those joinclauses.
|
|
|
|
* The set of usable joinclauses, and thus the best inner indexscan,
|
|
|
|
* thus varies depending on which outer relation we consider; so we have
|
|
|
|
* to recompute the best such path for every join. To avoid lots of
|
|
|
|
* redundant computation, we cache the results of such searches. For
|
|
|
|
* each index we compute the set of possible otherrelids (all relids
|
|
|
|
* appearing in joinquals that could become indexquals for this index).
|
|
|
|
* Two outer relations whose relids have the same intersection with this
|
|
|
|
* set will have the same set of available joinclauses and thus the same
|
|
|
|
* best inner indexscan for that index. Similarly, for each base relation,
|
|
|
|
* we form the union of the per-index otherrelids sets. Two outer relations
|
|
|
|
* with the same intersection with that set will have the same best overall
|
|
|
|
* inner indexscan for the base relation. We use lists of InnerIndexscanInfo
|
|
|
|
* nodes to cache the results of these searches at both the index and
|
|
|
|
* relation level.
|
|
|
|
*
|
|
|
|
* The search key also includes a bool showing whether the join being
|
|
|
|
* considered is an outer join. Since we constrain the join order for
|
|
|
|
* outer joins, I believe that this bool can only have one possible value
|
|
|
|
* for any particular base relation; but store it anyway to avoid confusion.
|
|
|
|
*/
|
|
|
|
|
|
|
|
typedef struct InnerIndexscanInfo
|
|
|
|
{
|
|
|
|
NodeTag type;
|
|
|
|
/* The lookup key: */
|
|
|
|
Relids other_relids; /* a set of relevant other relids */
|
|
|
|
bool isouterjoin; /* true if join is outer */
|
|
|
|
/* Best path for this lookup key: */
|
|
|
|
Path *best_innerpath; /* best inner indexscan, or NULL if none */
|
|
|
|
} InnerIndexscanInfo;
|
|
|
|
|
2001-11-05 18:46:40 +01:00
|
|
|
#endif /* RELATION_H */
|