Changeset: 0f6db86a9e4b for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/0f6db86a9e4b
Modified Files:
sql/server/rel_statistics.c
sql/server/sql_mvc.c
sql/server/sql_mvc.h
sql/storage/bat/bat_storage.c
sql/storage/sql_storage.h
sql/storage/store.c
Branch: properties
Log Message:
Propagate unique flag
diffs (203 lines):
diff --git a/sql/server/rel_statistics.c b/sql/server/rel_statistics.c
--- a/sql/server/rel_statistics.c
+++ b/sql/server/rel_statistics.c
@@ -41,7 +41,7 @@ rel_propagate_column_ref_statistics(mvc
case op_select:
case op_anti:
case op_semi: {
- bool found_without_semantics = false, found_left =
false, found_right = false;
+ bool found_without_semantics = false, found_left =
false, found_right = false, still_unique = true;
if ((is_innerjoin(rel->op) || is_select(rel->op)) &&
list_length(rel->exps) == 1 && exp_is_false(rel->exps->h->data))
return NULL; /* nothing will pass, skip */
@@ -54,17 +54,18 @@ rel_propagate_column_ref_statistics(mvc
if (!found_left && !found_right)
return NULL;
+ still_unique = !list_empty(rel->exps); /* cartesian
products */
if (!list_empty(rel->exps) && rel->op != op_anti) { /*
if there's an or, the MIN and MAX get difficult to propagate */
for (node *n = rel->exps->h ; n ; n = n->next) {
sql_exp *comp = n->data, *le = comp->l,
*lne = NULL, *re = comp->r, *rne = NULL, *fe = comp->f, *fne = NULL;
if (comp->type == e_cmp) {
-
if (is_theta_exp(comp->flag) &&
((lne = comparison_find_column(le, e)) || (rne = comparison_find_column(re, e))
|| (fe && (fne = comparison_find_column(fe, e))))) {
atom *lval_min =
find_prop_and_get(le->p, PROP_MIN), *lval_max = find_prop_and_get(le->p,
PROP_MAX),
*rval_min =
find_prop_and_get(re->p, PROP_MIN), *rval_max = find_prop_and_get(re->p,
PROP_MAX);
found_without_semantics
|= !comp->semantics;
+ still_unique &=
comp->flag == cmp_equal && is_unique(le) && is_unique(re); /* unique if only
equi-joins on unique columns are there */
if (is_full(rel->op) ||
(is_left(rel->op) && found_left) || (is_right(rel->op) && found_right)) /* on
outer joins, min and max cannot be propagated on some cases */
continue;
/* if (end2 >= start1
&& start2 <= end1) then the 2 intervals are intersected */
@@ -145,7 +146,11 @@ rel_propagate_column_ref_statistics(mvc
break;
}
}
+ } else {
+ still_unique = false;
}
+ } else {
+ still_unique = false;
}
}
}
@@ -153,6 +158,8 @@ rel_propagate_column_ref_statistics(mvc
set_has_nil(e);
if (!is_outerjoin(rel->op) && found_without_semantics)
/* at an outer join, null values pass */
set_has_no_nil(e);
+ if (is_unique(e) && is_join(rel->op) && !still_unique)
+ set_not_unique(e);
return e;
}
case op_table:
@@ -171,6 +178,9 @@ rel_propagate_column_ref_statistics(mvc
set_property(sql, e, PROP_MIN, fval);
if (!has_nil(found))
set_has_no_nil(e);
+ if (is_unique(found) || (need_distinct(rel) &&
list_length(rel->exps) == 1) ||
+ (is_groupby(rel->op) &&
list_length(rel->r) == 1 && exps_find_exp(rel->r, e)))
+ set_unique(e);
return e;
}
return NULL;
@@ -204,12 +214,14 @@ rel_basetable_get_statistics(visitor *v,
(void)depth;
if ((c = name_find_column(rel, exp_relname(e), exp_name(e), -2, NULL)))
{
- bool nonil = false;
+ bool nonil = false, unique = false;
ValRecord min, max;
- int ok = mvc_col_stats(sql, c, &nonil, &min, &max);
+ int ok = mvc_col_stats(sql, c, &nonil, &unique, &min, &max);
if (has_nil(e) && nonil)
set_has_no_nil(e);
+ if (!is_unique(e) && unique)
+ set_unique(e);
if ((ok & 1) == 1) {
if (!VALisnil(&min)) {
prop *p = e->p = prop_create(sql->sa, PROP_MIN,
e->p);
@@ -260,13 +272,19 @@ rel_setop_get_statistics(mvc *sql, sql_r
if (is_union(rel->op)) {
if (!has_nil(le) && !has_nil(re))
set_has_no_nil(e);
+ if (need_distinct(rel) && list_length(rel->exps) == 1)
+ set_unique(e);
} else if (is_inter(rel->op)) {
if (!has_nil(le) || !has_nil(re))
set_has_no_nil(e);
+ if (is_unique(le) || (need_distinct(rel) &&
list_length(rel->exps) == 1))
+ set_unique(e);
} else {
assert(is_except(rel->op));
if (!has_nil(le))
set_has_no_nil(e);
+ if (is_unique(le) || (need_distinct(rel) &&
list_length(rel->exps) == 1))
+ set_unique(e);
}
return false;
}
@@ -389,7 +407,7 @@ rel_propagate_statistics(visitor *v, sql
}
} break;
case e_cmp:
- /* propagating min and max of booleans is not very worth it */
+ /* TODO? propagating min/max/unique of booleans is not very
worth it */
if (e->flag == cmp_or || e->flag == cmp_filter) {
if (!have_nil(e->l) && !have_nil(e->r))
set_has_no_nil(e);
diff --git a/sql/server/sql_mvc.c b/sql/server/sql_mvc.c
--- a/sql/server/sql_mvc.c
+++ b/sql/server/sql_mvc.c
@@ -1541,10 +1541,10 @@ mvc_is_duplicate_eliminated(mvc *m, sql_
}
int
-mvc_col_stats(mvc *m, sql_column *col, bool *nonil, ValPtr min, ValPtr max)
+mvc_col_stats(mvc *m, sql_column *col, bool *nonil, bool *unique, ValPtr min,
ValPtr max)
{
TRC_DEBUG(SQL_TRANS, "Retrieving column stats for: %s\n",
col->base.name);
- return sql_trans_col_stats(m->session->tr, col, nonil, min, max);
+ return sql_trans_col_stats(m->session->tr, col, nonil, unique, min,
max);
}
int
diff --git a/sql/server/sql_mvc.h b/sql/server/sql_mvc.h
--- a/sql/server/sql_mvc.h
+++ b/sql/server/sql_mvc.h
@@ -218,7 +218,7 @@ extern int mvc_access(mvc *m, sql_table
extern int mvc_is_sorted(mvc *c, sql_column *col);
extern int mvc_is_unique(mvc *m, sql_column *col);
extern int mvc_is_duplicate_eliminated(mvc *c, sql_column *col);
-extern int mvc_col_stats(mvc *m, sql_column *col, bool *nonil, ValPtr min,
ValPtr max);
+extern int mvc_col_stats(mvc *m, sql_column *col, bool *nonil, bool *unique,
ValPtr min, ValPtr max);
extern int mvc_create_ukey(sql_key **kres, mvc *m, sql_table *t, const char
*name, key_type kt);
extern int mvc_create_fkey(sql_fkey **kres, mvc *m, sql_table *t, const char
*name, key_type kt, sql_key *rkey, int on_delete, int on_update);
diff --git a/sql/storage/bat/bat_storage.c b/sql/storage/bat/bat_storage.c
--- a/sql/storage/bat/bat_storage.c
+++ b/sql/storage/bat/bat_storage.c
@@ -2705,7 +2705,7 @@ double_elim_col(sql_trans *tr, sql_colum
}
static int
-col_stats(sql_trans *tr, sql_column *c, bool *nonil, ValPtr min, ValPtr max)
+col_stats(sql_trans *tr, sql_column *c, bool *nonil, bool *unique, ValPtr min,
ValPtr max)
{
int ok = 0;
BAT *b = NULL;
@@ -2713,6 +2713,7 @@ col_stats(sql_trans *tr, sql_column *c,
assert(tr->active);
*nonil = false;
+ *unique = false;
if (!c || !isTable(c->t) || !c->t->s)
return ok;
@@ -2736,6 +2737,7 @@ col_stats(sql_trans *tr, sql_column *c,
bat_iterator_end(&bi);
bat_destroy(b);
}
+ *unique = d->cs.ucnt == 0 && (d->cs.st == ST_DICT ?
(((b = quick_descriptor(d->cs.bid)) != NULL) && b->tkey) : b->tkey);
if (*nonil && d->cs.ucnt > 0)
*nonil &= ((b = quick_descriptor(d->cs.uvbid))
!= NULL) && b->tnonil && !b->tnil;
}
diff --git a/sql/storage/sql_storage.h b/sql/storage/sql_storage.h
--- a/sql/storage/sql_storage.h
+++ b/sql/storage/sql_storage.h
@@ -154,7 +154,7 @@ typedef size_t (*count_idx_fptr) (sql_tr
typedef size_t (*dcount_col_fptr) (sql_trans *tr, sql_column *c);
typedef int (*min_max_col_fptr) (sql_trans *tr, sql_column *c);
typedef int (*prop_col_fptr) (sql_trans *tr, sql_column *c);
-typedef int (*proprec_col_fptr) (sql_trans *tr, sql_column *c, bool *nonil,
ValPtr min, ValPtr max);
+typedef int (*proprec_col_fptr) (sql_trans *tr, sql_column *c, bool *nonil,
bool *unique, ValPtr min, ValPtr max);
/*
-- create the necessary storage resources for columns, indices and tables
@@ -396,7 +396,7 @@ extern int sql_trans_alter_storage(sql_t
extern int sql_trans_is_sorted(sql_trans *tr, sql_column *col);
extern int sql_trans_is_unique(sql_trans *tr, sql_column *col);
extern int sql_trans_is_duplicate_eliminated(sql_trans *tr, sql_column *col);
-extern int sql_trans_col_stats(sql_trans *tr, sql_column *col, bool *nonil,
ValPtr min, ValPtr max);
+extern int sql_trans_col_stats(sql_trans *tr, sql_column *col, bool *nonil,
bool *unique, ValPtr min, ValPtr max);
extern size_t sql_trans_dist_count(sql_trans *tr, sql_column *col);
extern int sql_trans_ranges(sql_trans *tr, sql_column *col, void **min, void
**max);
diff --git a/sql/storage/store.c b/sql/storage/store.c
--- a/sql/storage/store.c
+++ b/sql/storage/store.c
@@ -6173,11 +6173,11 @@ sql_trans_is_duplicate_eliminated( sql_t
}
int
-sql_trans_col_stats( sql_trans *tr, sql_column *col, bool *nonil, ValPtr min,
ValPtr max )
+sql_trans_col_stats( sql_trans *tr, sql_column *col, bool *nonil, bool
*unique, ValPtr min, ValPtr max )
{
sqlstore *store = tr->store;
if (col && isTable(col->t) && store->storage_api.col_stats)
- return store->storage_api.col_stats(tr, col, nonil, min, max);
+ return store->storage_api.col_stats(tr, col, nonil, unique,
min, max);
return 0;
}
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
