* allpaths.c
* Routines to find possible search paths for processing a query
*
- * Portions Copyright (c) 1996-2006, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1996-2011, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
*
* IDENTIFICATION
- * $PostgreSQL: pgsql/src/backend/optimizer/path/allpaths.c,v 1.151 2006/08/10 02:36:28 tgl Exp $
+ * src/backend/optimizer/path/allpaths.c
*
*-------------------------------------------------------------------------
*/
#include "postgres.h"
+#include <math.h>
+
+#include "catalog/pg_class.h"
+#include "nodes/nodeFuncs.h"
#ifdef OPTIMIZER_DEBUG
#include "nodes/print.h"
#endif
#include "optimizer/plancat.h"
#include "optimizer/planner.h"
#include "optimizer/prep.h"
+#include "optimizer/restrictinfo.h"
#include "optimizer/var.h"
#include "parser/parse_clause.h"
-#include "parser/parse_expr.h"
#include "parser/parsetree.h"
#include "rewrite/rewriteManip.h"
+#include "utils/lsyscache.h"
/* These parameters are set by GUC */
bool enable_geqo = false; /* just in case GUC doesn't set it */
int geqo_threshold;
+/* Hook for plugins to replace standard_join_search() */
+join_search_hook_type join_search_hook = NULL;
+
static void set_base_rel_pathlists(PlannerInfo *root);
-static void set_rel_pathlist(PlannerInfo *root, RelOptInfo *rel, Index rti);
+static void set_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
+ Index rti, RangeTblEntry *rte);
static void set_plain_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
RangeTblEntry *rte);
static void set_append_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
- Index rti, RangeTblEntry *rte);
+ Index rti, RangeTblEntry *rte);
+static List *accumulate_append_subpath(List *subpaths, Path *path);
+static void set_dummy_rel_pathlist(RelOptInfo *rel);
static void set_subquery_pathlist(PlannerInfo *root, RelOptInfo *rel,
Index rti, RangeTblEntry *rte);
static void set_function_pathlist(PlannerInfo *root, RelOptInfo *rel,
RangeTblEntry *rte);
static void set_values_pathlist(PlannerInfo *root, RelOptInfo *rel,
- RangeTblEntry *rte);
+ RangeTblEntry *rte);
+static void set_cte_pathlist(PlannerInfo *root, RelOptInfo *rel,
+ RangeTblEntry *rte);
+static void set_worktable_pathlist(PlannerInfo *root, RelOptInfo *rel,
+ RangeTblEntry *rte);
+static void set_foreign_pathlist(PlannerInfo *root, RelOptInfo *rel,
+ RangeTblEntry *rte);
static RelOptInfo *make_rel_from_joinlist(PlannerInfo *root, List *joinlist);
-static RelOptInfo *make_one_rel_by_joins(PlannerInfo *root, int levels_needed,
- List *initial_rels);
static bool subquery_is_pushdown_safe(Query *subquery, Query *topquery,
bool *differentTypes);
static bool recurse_pushdown_safe(Node *setOp, Query *topquery,
if (rel->reloptkind != RELOPT_BASEREL)
continue;
- set_rel_pathlist(root, rel, rti);
+ set_rel_pathlist(root, rel, rti, root->simple_rte_array[rti]);
}
}
* Build access paths for a base relation
*/
static void
-set_rel_pathlist(PlannerInfo *root, RelOptInfo *rel, Index rti)
+set_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
+ Index rti, RangeTblEntry *rte)
{
- RangeTblEntry *rte = rt_fetch(rti, root->parse->rtable);
-
if (rte->inh)
{
/* It's an "append relation", process accordingly */
}
else if (rel->rtekind == RTE_FUNCTION)
{
- /* RangeFunction --- generate a separate plan for it */
+ /* RangeFunction --- generate a suitable path for it */
set_function_pathlist(root, rel, rte);
}
else if (rel->rtekind == RTE_VALUES)
{
- /* Values list --- generate a separate plan for it */
+ /* Values list --- generate a suitable path for it */
set_values_pathlist(root, rel, rte);
}
+ else if (rel->rtekind == RTE_CTE)
+ {
+ /* CTE reference --- generate a suitable path for it */
+ if (rte->self_reference)
+ set_worktable_pathlist(root, rel, rte);
+ else
+ set_cte_pathlist(root, rel, rte);
+ }
else
{
- /* Plain relation */
Assert(rel->rtekind == RTE_RELATION);
- set_plain_rel_pathlist(root, rel, rte);
+ if (get_rel_relkind(rte->relid) == RELKIND_FOREIGN_TABLE)
+ {
+ /* Foreign table */
+ set_foreign_pathlist(root, rel, rte);
+ }
+ else
+ {
+ /* Plain relation */
+ set_plain_rel_pathlist(root, rel, rte);
+ }
}
#ifdef OPTIMIZER_DEBUG
static void
set_plain_rel_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
{
- /* Mark rel with estimated output rows, width, etc */
- set_baserel_size_estimates(root, rel);
+ /*
+ * If we can prove we don't need to scan the rel via constraint exclusion,
+ * set up a single dummy path for it. We only need to check for regular
+ * baserels; if it's an otherrel, CE was already checked in
+ * set_append_rel_pathlist().
+ */
+ if (rel->reloptkind == RELOPT_BASEREL &&
+ relation_excluded_by_constraints(root, rel, rte))
+ {
+ set_dummy_rel_pathlist(rel);
+ return;
+ }
- /* Test any partial indexes of rel for applicability */
+ /*
+ * Test any partial indexes of rel for applicability. We must do this
+ * first since partial unique indexes can affect size estimates.
+ */
check_partial_indexes(root, rel);
+ /* Mark rel with estimated output rows, width, etc */
+ set_baserel_size_estimates(root, rel);
+
/*
* Check to see if we can extract any restriction conditions from join
* quals that are OR-of-AND structures. If so, add them to the rel's
- * restriction list, and recompute the size estimates.
+ * restriction list, and redo the above steps.
*/
if (create_or_index_quals(root, rel))
- set_baserel_size_estimates(root, rel);
-
- /*
- * If we can prove we don't need to scan the rel via constraint exclusion,
- * set up a single dummy path for it. (Rather than inventing a special
- * "dummy" path type, we represent this as an AppendPath with no members.)
- */
- if (relation_excluded_by_constraints(rel, rte))
{
- /* Reset output-rows estimate to 0 */
- rel->rows = 0;
-
- add_path(rel, (Path *) create_append_path(rel, NIL));
-
- /* Select cheapest path (pretty easy in this case...) */
- set_cheapest(rel);
-
- return;
+ check_partial_indexes(root, rel);
+ set_baserel_size_estimates(root, rel);
}
/*
* set_append_rel_pathlist
* Build access paths for an "append relation"
*
- * The passed-in rel and RTE represent the entire append relation. The
+ * The passed-in rel and RTE represent the entire append relation. The
* relation's contents are computed by appending together the output of
* the individual member relations. Note that in the inheritance case,
* the first member relation is actually the same table as is mentioned in
Index rti, RangeTblEntry *rte)
{
int parentRTindex = rti;
+ List *live_childrels = NIL;
List *subpaths = NIL;
+ List *all_child_pathkeys = NIL;
+ double parent_rows;
+ double parent_size;
+ double *parent_attrsizes;
+ int nattrs;
ListCell *l;
/*
- * XXX for now, can't handle inherited expansion of FOR UPDATE/SHARE; can
- * we do better? (This will take some redesign because the executor
- * currently supposes that every rowMark relation is involved in every
- * row returned by the query.)
- */
- if (get_rowmark(root->parse, parentRTindex))
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("SELECT FOR UPDATE/SHARE is not supported for inheritance queries")));
-
- /*
- * We might have looked up indexes for the parent rel, but they're
- * really not relevant to the appendrel. Reset the pointer to avoid
- * any confusion.
- */
- rel->indexlist = NIL;
-
- /*
- * Initialize to compute size estimates for whole append relation
+ * Initialize to compute size estimates for whole append relation.
+ *
+ * We handle width estimates by weighting the widths of different child
+ * rels proportionally to their number of rows. This is sensible because
+ * the use of width estimates is mainly to compute the total relation
+ * "footprint" if we have to sort or hash it. To do this, we sum the
+ * total equivalent size (in "double" arithmetic) and then divide by the
+ * total rowcount estimate. This is done separately for the total rel
+ * width and each attribute.
+ *
+ * Note: if you consider changing this logic, beware that child rels could
+ * have zero rows and/or width, if they were excluded by constraints.
*/
- rel->rows = 0;
- rel->width = 0;
+ parent_rows = 0;
+ parent_size = 0;
+ nattrs = rel->max_attr - rel->min_attr + 1;
+ parent_attrsizes = (double *) palloc0(nattrs * sizeof(double));
/*
* Generate access paths for each member relation, and pick the cheapest
{
AppendRelInfo *appinfo = (AppendRelInfo *) lfirst(l);
int childRTindex;
+ RangeTblEntry *childRTE;
RelOptInfo *childrel;
- Path *childpath;
+ List *childquals;
+ Node *childqual;
+ ListCell *lcp;
ListCell *parentvars;
ListCell *childvars;
continue;
childRTindex = appinfo->child_relid;
+ childRTE = root->simple_rte_array[childRTindex];
/*
- * Make a RelOptInfo for the child so we can do planning. Mark it as
- * an "other rel" since it will not be part of the main join tree.
+ * The child rel's RelOptInfo was already created during
+ * add_base_rels_to_query.
*/
- childrel = build_simple_rel(root, childRTindex,
- RELOPT_OTHER_MEMBER_REL);
+ childrel = find_base_rel(root, childRTindex);
+ Assert(childrel->reloptkind == RELOPT_OTHER_MEMBER_REL);
/*
- * Copy the parent's targetlist and quals to the child, with
- * appropriate substitution of variables.
+ * We have to copy the parent's targetlist and quals to the child,
+ * with appropriate substitution of variables. However, only the
+ * baserestrictinfo quals are needed before we can check for
+ * constraint exclusion; so do that first and then check to see if we
+ * can disregard this child.
+ *
+ * As of 8.4, the child rel's targetlist might contain non-Var
+ * expressions, which means that substitution into the quals could
+ * produce opportunities for const-simplification, and perhaps even
+ * pseudoconstant quals. To deal with this, we strip the RestrictInfo
+ * nodes, do the substitution, do const-simplification, and then
+ * reconstitute the RestrictInfo layer.
*/
- childrel->reltargetlist = (List *)
- adjust_appendrel_attrs((Node *) rel->reltargetlist,
- appinfo);
- childrel->baserestrictinfo = (List *)
- adjust_appendrel_attrs((Node *) rel->baserestrictinfo,
- appinfo);
+ childquals = get_all_actual_clauses(rel->baserestrictinfo);
+ childquals = (List *) adjust_appendrel_attrs((Node *) childquals,
+ appinfo);
+ childqual = eval_const_expressions(root, (Node *)
+ make_ands_explicit(childquals));
+ if (childqual && IsA(childqual, Const) &&
+ (((Const *) childqual)->constisnull ||
+ !DatumGetBool(((Const *) childqual)->constvalue)))
+ {
+ /*
+ * Restriction reduces to constant FALSE or constant NULL after
+ * substitution, so this child need not be scanned.
+ */
+ set_dummy_rel_pathlist(childrel);
+ continue;
+ }
+ childquals = make_ands_implicit((Expr *) childqual);
+ childquals = make_restrictinfos_from_actual_clauses(root,
+ childquals);
+ childrel->baserestrictinfo = childquals;
+
+ if (relation_excluded_by_constraints(root, childrel, childRTE))
+ {
+ /*
+ * This child need not be scanned, so we can omit it from the
+ * appendrel. Mark it with a dummy cheapest-path though, in case
+ * best_appendrel_indexscan() looks at it later.
+ */
+ set_dummy_rel_pathlist(childrel);
+ continue;
+ }
+
+ /* CE failed, so finish copying targetlist and join quals */
childrel->joininfo = (List *)
adjust_appendrel_attrs((Node *) rel->joininfo,
appinfo);
+ childrel->reltargetlist = (List *)
+ adjust_appendrel_attrs((Node *) rel->reltargetlist,
+ appinfo);
/*
- * Copy the parent's attr_needed data as well, with appropriate
- * adjustment of relids and attribute numbers.
+ * We have to make child entries in the EquivalenceClass data
+ * structures as well. This is needed either if the parent
+ * participates in some eclass joins (because we will want to
+ * consider inner-indexscan joins on the individual children)
+ * or if the parent has useful pathkeys (because we should try
+ * to build MergeAppend paths that produce those sort orderings).
*/
- pfree(childrel->attr_needed);
- childrel->attr_needed =
- adjust_appendrel_attr_needed(rel, appinfo,
- childrel->min_attr,
- childrel->max_attr);
+ if (rel->has_eclass_joins || has_useful_pathkeys(root, rel))
+ add_child_rel_equivalences(root, appinfo, rel, childrel);
+ childrel->has_eclass_joins = rel->has_eclass_joins;
/*
- * Compute the child's access paths, and add the cheapest one
- * to the Append path we are constructing for the parent.
- *
- * It's possible that the child is itself an appendrel, in which
- * case we can "cut out the middleman" and just add its child
- * paths to our own list. (We don't try to do this earlier because
- * we need to apply both levels of transformation to the quals.)
- * This test also handles the case where the child rel need not
- * be scanned because of constraint exclusion: it'll have an
- * Append path with no subpaths, and will vanish from our list.
+ * Note: we could compute appropriate attr_needed data for the child's
+ * variables, by transforming the parent's attr_needed through the
+ * translated_vars mapping. However, currently there's no need
+ * because attr_needed is only examined for base relations not
+ * otherrels. So we just leave the child's attr_needed empty.
*/
- set_rel_pathlist(root, childrel, childRTindex);
- childpath = childrel->cheapest_total_path;
- if (IsA(childpath, AppendPath))
- subpaths = list_concat(subpaths,
- ((AppendPath *) childpath)->subpaths);
- else
- subpaths = lappend(subpaths, childpath);
+ /* Remember which childrels are live, for MergeAppend logic below */
+ live_childrels = lappend(live_childrels, childrel);
/*
- * Propagate size information from the child back to the parent. For
- * simplicity, we use the largest widths from any child as the parent
- * estimates.
+ * Compute the child's access paths, and add the cheapest one to the
+ * Append path we are constructing for the parent.
*/
- rel->rows += childrel->rows;
- if (childrel->width > rel->width)
- rel->width = childrel->width;
+ set_rel_pathlist(root, childrel, childRTindex, childRTE);
- forboth(parentvars, rel->reltargetlist,
- childvars, childrel->reltargetlist)
+ subpaths = accumulate_append_subpath(subpaths,
+ childrel->cheapest_total_path);
+
+ /*
+ * Collect a list of all the available path orderings for all the
+ * children. We use this as a heuristic to indicate which sort
+ * orderings we should build MergeAppend paths for.
+ */
+ foreach(lcp, childrel->pathlist)
{
- Var *parentvar = (Var *) lfirst(parentvars);
- Var *childvar = (Var *) lfirst(childvars);
+ Path *childpath = (Path *) lfirst(lcp);
+ List *childkeys = childpath->pathkeys;
+ ListCell *lpk;
+ bool found = false;
- if (IsA(parentvar, Var) &&
- IsA(childvar, Var))
+ /* Ignore unsorted paths */
+ if (childkeys == NIL)
+ continue;
+
+ /* Have we already seen this ordering? */
+ foreach(lpk, all_child_pathkeys)
+ {
+ List *existing_pathkeys = (List *) lfirst(lpk);
+
+ if (compare_pathkeys(existing_pathkeys,
+ childkeys) == PATHKEYS_EQUAL)
+ {
+ found = true;
+ break;
+ }
+ }
+ if (!found)
{
- int pndx = parentvar->varattno - rel->min_attr;
- int cndx = childvar->varattno - childrel->min_attr;
+ /* No, so add it to all_child_pathkeys */
+ all_child_pathkeys = lappend(all_child_pathkeys, childkeys);
+ }
+ }
- if (childrel->attr_widths[cndx] > rel->attr_widths[pndx])
- rel->attr_widths[pndx] = childrel->attr_widths[cndx];
+ /*
+ * Accumulate size information from each child.
+ */
+ if (childrel->rows > 0)
+ {
+ parent_rows += childrel->rows;
+ parent_size += childrel->width * childrel->rows;
+
+ forboth(parentvars, rel->reltargetlist,
+ childvars, childrel->reltargetlist)
+ {
+ Var *parentvar = (Var *) lfirst(parentvars);
+ Var *childvar = (Var *) lfirst(childvars);
+
+ /*
+ * Accumulate per-column estimates too. Whole-row Vars and
+ * PlaceHolderVars can be ignored here.
+ */
+ if (IsA(parentvar, Var) &&
+ IsA(childvar, Var))
+ {
+ int pndx = parentvar->varattno - rel->min_attr;
+ int cndx = childvar->varattno - childrel->min_attr;
+
+ parent_attrsizes[pndx] += childrel->attr_widths[cndx] * childrel->rows;
+ }
}
}
}
/*
- * Finally, build Append path and install it as the only access path for
- * the parent rel. (Note: this is correct even if we have zero or one
- * live subpath due to constraint exclusion.)
+ * Save the finished size estimates.
+ */
+ rel->rows = parent_rows;
+ if (parent_rows > 0)
+ {
+ int i;
+
+ rel->width = rint(parent_size / parent_rows);
+ for (i = 0; i < nattrs; i++)
+ rel->attr_widths[i] = rint(parent_attrsizes[i] / parent_rows);
+ }
+ else
+ rel->width = 0; /* attr_widths should be zero already */
+
+ /*
+ * Set "raw tuples" count equal to "rows" for the appendrel; needed
+ * because some places assume rel->tuples is valid for any baserel.
+ */
+ rel->tuples = parent_rows;
+
+ pfree(parent_attrsizes);
+
+ /*
+ * Next, build an unordered Append path for the rel. (Note: this is
+ * correct even if we have zero or one live subpath due to constraint
+ * exclusion.)
*/
add_path(rel, (Path *) create_append_path(rel, subpaths));
+ /*
+ * Next, build MergeAppend paths based on the collected list of child
+ * pathkeys. We consider both cheapest-startup and cheapest-total
+ * cases, ie, for each interesting ordering, collect all the cheapest
+ * startup subpaths and all the cheapest total paths, and build a
+ * MergeAppend path for each list.
+ */
+ foreach(l, all_child_pathkeys)
+ {
+ List *pathkeys = (List *) lfirst(l);
+ List *startup_subpaths = NIL;
+ List *total_subpaths = NIL;
+ bool startup_neq_total = false;
+ ListCell *lcr;
+
+ /* Select the child paths for this ordering... */
+ foreach(lcr, live_childrels)
+ {
+ RelOptInfo *childrel = (RelOptInfo *) lfirst(lcr);
+ Path *cheapest_startup,
+ *cheapest_total;
+
+ /* Locate the right paths, if they are available. */
+ cheapest_startup =
+ get_cheapest_path_for_pathkeys(childrel->pathlist,
+ pathkeys,
+ STARTUP_COST);
+ cheapest_total =
+ get_cheapest_path_for_pathkeys(childrel->pathlist,
+ pathkeys,
+ TOTAL_COST);
+
+ /*
+ * If we can't find any paths with the right order just add the
+ * cheapest-total path; we'll have to sort it.
+ */
+ if (cheapest_startup == NULL)
+ cheapest_startup = childrel->cheapest_total_path;
+ if (cheapest_total == NULL)
+ cheapest_total = childrel->cheapest_total_path;
+
+ /*
+ * Notice whether we actually have different paths for the
+ * "cheapest" and "total" cases; frequently there will be no
+ * point in two create_merge_append_path() calls.
+ */
+ if (cheapest_startup != cheapest_total)
+ startup_neq_total = true;
+
+ startup_subpaths =
+ accumulate_append_subpath(startup_subpaths, cheapest_startup);
+ total_subpaths =
+ accumulate_append_subpath(total_subpaths, cheapest_total);
+ }
+
+ /* ... and build the MergeAppend paths */
+ add_path(rel, (Path *) create_merge_append_path(root,
+ rel,
+ startup_subpaths,
+ pathkeys));
+ if (startup_neq_total)
+ add_path(rel, (Path *) create_merge_append_path(root,
+ rel,
+ total_subpaths,
+ pathkeys));
+ }
+
+ /* Select cheapest path */
+ set_cheapest(rel);
+}
+
+/*
+ * accumulate_append_subpath
+ * Add a subpath to the list being built for an Append or MergeAppend
+ *
+ * It's possible that the child is itself an Append path, in which case
+ * we can "cut out the middleman" and just add its child paths to our
+ * own list. (We don't try to do this earlier because we need to
+ * apply both levels of transformation to the quals.)
+ */
+static List *
+accumulate_append_subpath(List *subpaths, Path *path)
+{
+ if (IsA(path, AppendPath))
+ {
+ AppendPath *apath = (AppendPath *) path;
+
+ /* list_copy is important here to avoid sharing list substructure */
+ return list_concat(subpaths, list_copy(apath->subpaths));
+ }
+ else
+ return lappend(subpaths, path);
+}
+
+/*
+ * set_dummy_rel_pathlist
+ * Build a dummy path for a relation that's been excluded by constraints
+ *
+ * Rather than inventing a special "dummy" path type, we represent this as an
+ * AppendPath with no members (see also IS_DUMMY_PATH macro).
+ */
+static void
+set_dummy_rel_pathlist(RelOptInfo *rel)
+{
+ /* Set dummy size estimates --- we leave attr_widths[] as zeroes */
+ rel->rows = 0;
+ rel->width = 0;
+
+ add_path(rel, (Path *) create_append_path(rel, NIL));
+
/* Select cheapest path (pretty easy in this case...) */
set_cheapest(rel);
}
Query *subquery = rte->subquery;
bool *differentTypes;
double tuple_fraction;
+ PlannerInfo *subroot;
List *pathkeys;
- List *subquery_pathkeys;
+
+ /*
+ * Must copy the Query so that planning doesn't mess up the RTE contents
+ * (really really need to fix the planner to not scribble on its input,
+ * someday).
+ */
+ subquery = copyObject(subquery);
/* We need a workspace for keeping track of set-op type coercions */
differentTypes = (bool *)
tuple_fraction = root->tuple_fraction;
/* Generate the plan for the subquery */
- rel->subplan = subquery_planner(subquery, tuple_fraction,
- &subquery_pathkeys);
-
- /* Copy number of output rows from subplan */
- rel->tuples = rel->subplan->plan_rows;
+ rel->subplan = subquery_planner(root->glob, subquery,
+ root,
+ false, tuple_fraction,
+ &subroot);
+ rel->subrtable = subroot->parse->rtable;
+ rel->subrowmark = subroot->rowMarks;
/* Mark rel with estimated output rows, width, etc */
- set_baserel_size_estimates(root, rel);
+ set_subquery_size_estimates(root, rel, subroot);
/* Convert subquery pathkeys to outer representation */
- pathkeys = convert_subquery_pathkeys(root, rel, subquery_pathkeys);
+ pathkeys = convert_subquery_pathkeys(root, rel, subroot->query_pathkeys);
/* Generate appropriate path */
add_path(rel, create_subqueryscan_path(rel, pathkeys));
set_cheapest(rel);
}
+/*
+ * set_cte_pathlist
+ * Build the (single) access path for a non-self-reference CTE RTE
+ */
+static void
+set_cte_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
+{
+ Plan *cteplan;
+ PlannerInfo *cteroot;
+ Index levelsup;
+ int ndx;
+ ListCell *lc;
+ int plan_id;
+
+ /*
+ * Find the referenced CTE, and locate the plan previously made for it.
+ */
+ levelsup = rte->ctelevelsup;
+ cteroot = root;
+ while (levelsup-- > 0)
+ {
+ cteroot = cteroot->parent_root;
+ if (!cteroot) /* shouldn't happen */
+ elog(ERROR, "bad levelsup for CTE \"%s\"", rte->ctename);
+ }
+
+ /*
+ * Note: cte_plan_ids can be shorter than cteList, if we are still working
+ * on planning the CTEs (ie, this is a side-reference from another CTE).
+ * So we mustn't use forboth here.
+ */
+ ndx = 0;
+ foreach(lc, cteroot->parse->cteList)
+ {
+ CommonTableExpr *cte = (CommonTableExpr *) lfirst(lc);
+
+ if (strcmp(cte->ctename, rte->ctename) == 0)
+ break;
+ ndx++;
+ }
+ if (lc == NULL) /* shouldn't happen */
+ elog(ERROR, "could not find CTE \"%s\"", rte->ctename);
+ if (ndx >= list_length(cteroot->cte_plan_ids))
+ elog(ERROR, "could not find plan for CTE \"%s\"", rte->ctename);
+ plan_id = list_nth_int(cteroot->cte_plan_ids, ndx);
+ Assert(plan_id > 0);
+ cteplan = (Plan *) list_nth(root->glob->subplans, plan_id - 1);
+
+ /* Mark rel with estimated output rows, width, etc */
+ set_cte_size_estimates(root, rel, cteplan);
+
+ /* Generate appropriate path */
+ add_path(rel, create_ctescan_path(root, rel));
+
+ /* Select cheapest path (pretty easy in this case...) */
+ set_cheapest(rel);
+}
+
+/*
+ * set_worktable_pathlist
+ * Build the (single) access path for a self-reference CTE RTE
+ */
+static void
+set_worktable_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
+{
+ Plan *cteplan;
+ PlannerInfo *cteroot;
+ Index levelsup;
+
+ /*
+ * We need to find the non-recursive term's plan, which is in the plan
+ * level that's processing the recursive UNION, which is one level *below*
+ * where the CTE comes from.
+ */
+ levelsup = rte->ctelevelsup;
+ if (levelsup == 0) /* shouldn't happen */
+ elog(ERROR, "bad levelsup for CTE \"%s\"", rte->ctename);
+ levelsup--;
+ cteroot = root;
+ while (levelsup-- > 0)
+ {
+ cteroot = cteroot->parent_root;
+ if (!cteroot) /* shouldn't happen */
+ elog(ERROR, "bad levelsup for CTE \"%s\"", rte->ctename);
+ }
+ cteplan = cteroot->non_recursive_plan;
+ if (!cteplan) /* shouldn't happen */
+ elog(ERROR, "could not find plan for CTE \"%s\"", rte->ctename);
+
+ /* Mark rel with estimated output rows, width, etc */
+ set_cte_size_estimates(root, rel, cteplan);
+
+ /* Generate appropriate path */
+ add_path(rel, create_worktablescan_path(root, rel));
+
+ /* Select cheapest path (pretty easy in this case...) */
+ set_cheapest(rel);
+}
+
+/*
+ * set_foreign_pathlist
+ * Build the (single) access path for a foreign table RTE
+ */
+static void
+set_foreign_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
+{
+ /* Mark rel with estimated output rows, width, etc */
+ set_foreign_size_estimates(root, rel);
+
+ /* Generate appropriate path */
+ add_path(rel, (Path *) create_foreignscan_path(root, rel));
+
+ /* Select cheapest path (pretty easy in this case...) */
+ set_cheapest(rel);
+}
+
/*
* make_rel_from_joinlist
* Build access paths using a "joinlist" to guide the join path search.
{
/*
* Consider the different orders in which we could join the rels,
- * using either GEQO or regular optimizer.
+ * using a plugin, GEQO, or the regular join search code.
+ *
+ * We put the initial_rels list into a PlannerInfo field because
+ * has_legal_joinclause() needs to look at it (ugly :-().
*/
- if (enable_geqo && levels_needed >= geqo_threshold)
+ root->initial_rels = initial_rels;
+
+ if (join_search_hook)
+ return (*join_search_hook) (root, levels_needed, initial_rels);
+ else if (enable_geqo && levels_needed >= geqo_threshold)
return geqo(root, levels_needed, initial_rels);
else
- return make_one_rel_by_joins(root, levels_needed, initial_rels);
+ return standard_join_search(root, levels_needed, initial_rels);
}
}
/*
- * make_one_rel_by_joins
- * Find all possible joinpaths for a query by successively finding ways
+ * standard_join_search
+ * Find possible joinpaths for a query by successively finding ways
* to join component relations into join relations.
*
* 'levels_needed' is the number of iterations needed, ie, the number of
*
* 'initial_rels' is a list of RelOptInfo nodes for each independent
* jointree item. These are the components to be joined together.
+ * Note that levels_needed == list_length(initial_rels).
*
* Returns the final level of join relations, i.e., the relation that is
* the result of joining all the original relations together.
+ * At least one implementation path must be provided for this relation and
+ * all required sub-relations.
+ *
+ * To support loadable plugins that modify planner behavior by changing the
+ * join searching algorithm, we provide a hook variable that lets a plugin
+ * replace or supplement this function. Any such hook must return the same
+ * final join relation as the standard code would, but it might have a
+ * different set of implementation paths attached, and only the sub-joinrels
+ * needed for these paths need have been instantiated.
+ *
+ * Note to plugin authors: the functions invoked during standard_join_search()
+ * modify root->join_rel_list and root->join_rel_hash. If you want to do more
+ * than one join-order search, you'll probably need to save and restore the
+ * original states of those data structures. See geqo_eval() for an example.
*/
-static RelOptInfo *
-make_one_rel_by_joins(PlannerInfo *root, int levels_needed, List *initial_rels)
+RelOptInfo *
+standard_join_search(PlannerInfo *root, int levels_needed, List *initial_rels)
{
- List **joinitems;
int lev;
RelOptInfo *rel;
+ /*
+ * This function cannot be invoked recursively within any one planning
+ * problem, so join_rel_level[] can't be in use already.
+ */
+ Assert(root->join_rel_level == NULL);
+
/*
* We employ a simple "dynamic programming" algorithm: we first find all
* ways to build joins of two jointree items, then all ways to build joins
* joins, and so on until we have considered all ways to join all the
* items into one rel.
*
- * joinitems[j] is a list of all the j-item rels. Initially we set
- * joinitems[1] to represent all the single-jointree-item relations.
+ * root->join_rel_level[j] is a list of all the j-item rels. Initially we
+ * set root->join_rel_level[1] to represent all the single-jointree-item
+ * relations.
*/
- joinitems = (List **) palloc0((levels_needed + 1) * sizeof(List *));
+ root->join_rel_level = (List **) palloc0((levels_needed + 1) * sizeof(List *));
- joinitems[1] = initial_rels;
+ root->join_rel_level[1] = initial_rels;
for (lev = 2; lev <= levels_needed; lev++)
{
- ListCell *x;
+ ListCell *lc;
/*
* Determine all possible pairs of relations to be joined at this
* level, and build paths for making each one from every available
* pair of lower-level relations.
*/
- joinitems[lev] = make_rels_by_joins(root, lev, joinitems);
+ join_search_one_level(root, lev);
/*
* Do cleanup work on each just-processed rel.
*/
- foreach(x, joinitems[lev])
+ foreach(lc, root->join_rel_level[lev])
{
- rel = (RelOptInfo *) lfirst(x);
+ rel = (RelOptInfo *) lfirst(lc);
/* Find and save the cheapest paths for this rel */
set_cheapest(rel);
/*
* We should have a single rel at the final level.
*/
- if (joinitems[levels_needed] == NIL)
+ if (root->join_rel_level[levels_needed] == NIL)
elog(ERROR, "failed to build any %d-way joins", levels_needed);
- Assert(list_length(joinitems[levels_needed]) == 1);
+ Assert(list_length(root->join_rel_level[levels_needed]) == 1);
- rel = (RelOptInfo *) linitial(joinitems[levels_needed]);
+ rel = (RelOptInfo *) linitial(root->join_rel_level[levels_needed]);
+
+ root->join_rel_level = NULL;
return rel;
}
* 1. If the subquery has a LIMIT clause, we must not push down any quals,
* since that could change the set of rows returned.
*
- * 2. If the subquery contains EXCEPT or EXCEPT ALL set ops we cannot push
- * quals into it, because that would change the results.
+ * 2. If the subquery contains any window functions, we can't push quals
+ * into it, because that could change the results.
+ *
+ * 3. If the subquery contains EXCEPT or EXCEPT ALL set ops we cannot push
+ * quals into it, because that could change the results.
*
- * 3. For subqueries using UNION/UNION ALL/INTERSECT/INTERSECT ALL, we can
+ * 4. For subqueries using UNION/UNION ALL/INTERSECT/INTERSECT ALL, we can
* push quals into each component query, but the quals can only reference
* subquery columns that suffer no type coercions in the set operation.
* Otherwise there are possible semantic gotchas. So, we check the
if (subquery->limitOffset != NULL || subquery->limitCount != NULL)
return false;
+ /* Check point 2 */
+ if (subquery->hasWindowFuncs)
+ return false;
+
/* Are we at top level, or looking at a setop component? */
if (subquery == topquery)
{
*
* 4. If the subquery uses DISTINCT ON, we must not push down any quals that
* refer to non-DISTINCT output columns, because that could change the set
- * of rows returned. This condition is vacuous for DISTINCT, because then
- * there are no non-DISTINCT output columns, but unfortunately it's fairly
- * expensive to tell the difference between DISTINCT and DISTINCT ON in the
- * parsetree representation. It's cheaper to just make sure all the Vars
- * in the qual refer to DISTINCT columns.
+ * of rows returned. (This condition is vacuous for DISTINCT, because then
+ * there are no non-DISTINCT output columns, so we needn't check. But note
+ * we are assuming that the qual can't distinguish values that the DISTINCT
+ * operator sees as equal. This is a bit shaky but we have no way to test
+ * for the case, and it's unlikely enough that we shouldn't refuse the
+ * optimization just because it could theoretically happen.)
*
* 5. We must not push down any quals that refer to subselect outputs that
* return sets, else we'd introduce functions-returning-sets into the
* subquery's WHERE/HAVING quals.
+ *
+ * 6. We must not push down any quals that refer to subselect outputs that
+ * contain volatile functions, for fear of introducing strange results due
+ * to multiple evaluation of a volatile function.
*/
static bool
qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
if (contain_subplans(qual))
return false;
+ /*
+ * It would be unsafe to push down window function calls, but at least for
+ * the moment we could never see any in a qual anyhow.
+ */
+ Assert(!contain_window_function(qual));
+
/*
* Examine all Vars used in clause; since it's a restriction clause, all
* such Vars must refer to subselect output columns.
*/
- vars = pull_var_clause(qual, false);
+ vars = pull_var_clause(qual, PVC_INCLUDE_PLACEHOLDERS);
foreach(vl, vars)
{
Var *var = (Var *) lfirst(vl);
TargetEntry *tle;
+ /*
+ * XXX Punt if we find any PlaceHolderVars in the restriction clause.
+ * It's not clear whether a PHV could safely be pushed down, and even
+ * less clear whether such a situation could arise in any cases of
+ * practical interest anyway. So for the moment, just refuse to push
+ * down.
+ */
+ if (!IsA(var, Var))
+ {
+ safe = false;
+ break;
+ }
+
Assert(var->varno == rti);
/* Check point 2 */
Assert(tle != NULL);
Assert(!tle->resjunk);
- /* If subquery uses DISTINCT or DISTINCT ON, check point 4 */
- if (subquery->distinctClause != NIL &&
- !targetIsInSortList(tle, subquery->distinctClause))
+ /* If subquery uses DISTINCT ON, check point 4 */
+ if (subquery->hasDistinctOn &&
+ !targetIsInSortList(tle, InvalidOid, subquery->distinctClause))
{
/* non-DISTINCT column, so fail */
safe = false;
safe = false;
break;
}
+
+ /* Refuse volatile functions (point 6) */
+ if (contain_volatile_functions((Node *) tle->expr))
+ {
+ safe = false;
+ break;
+ }
}
list_free(vars);
*/
qual = ResolveNew(qual, rti, 0, rte,
subquery->targetList,
- CMD_SELECT, 0);
+ CMD_SELECT, 0,
+ &subquery->hasSubLinks);
/*
* Now attach the qual to the proper place: normally WHERE, but if the
case T_TidPath:
ptype = "TidScan";
break;
+ case T_ForeignPath:
+ ptype = "ForeignScan";
+ break;
case T_AppendPath:
ptype = "Append";
break;
+ case T_MergeAppendPath:
+ ptype = "MergeAppend";
+ break;
case T_ResultPath:
ptype = "Result";
break;
{
MergePath *mp = (MergePath *) path;
- if (mp->outersortkeys || mp->innersortkeys)
- {
- for (i = 0; i < indent; i++)
- printf("\t");
- printf(" sortouter=%d sortinner=%d\n",
- ((mp->outersortkeys) ? 1 : 0),
- ((mp->innersortkeys) ? 1 : 0));
- }
+ for (i = 0; i < indent; i++)
+ printf("\t");
+ printf(" sortouter=%d sortinner=%d materializeinner=%d\n",
+ ((mp->outersortkeys) ? 1 : 0),
+ ((mp->innersortkeys) ? 1 : 0),
+ ((mp->materialize_inner) ? 1 : 0));
}
print_path(root, jp->outerjoinpath, indent + 1);
projects / postgresql / blobdiff