Changeset: a6d9d12a58e6 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=a6d9d12a58e6
Modified Files:
sql/backends/monet5/sql.c
sql/backends/monet5/sql_rdf.h
sql/backends/monet5/sql_rdf_jgraph.c
sql/backends/monet5/sql_rdf_rel.c
sql/backends/monet5/sql_rdf_rel.h
Branch: rdf
Log Message:
Optimize for the case with Optional keyword.
Adding ifthenelse projection instead of using left outer join
diffs (truncated from 647 to 300 lines):
diff --git a/sql/backends/monet5/sql.c b/sql/backends/monet5/sql.c
--- a/sql/backends/monet5/sql.c
+++ b/sql/backends/monet5/sql.c
@@ -133,6 +133,10 @@ sql_symbol2relation(mvc *c, symbol *sym)
if (1) c->emod = mod_explain;
}
}
+ else {
+
+ _rel_print(c,r);
+ }
if (1){
r = rel_optimizer(c, r);
diff --git a/sql/backends/monet5/sql_rdf.h b/sql/backends/monet5/sql_rdf.h
--- a/sql/backends/monet5/sql_rdf.h
+++ b/sql/backends/monet5/sql_rdf.h
@@ -51,6 +51,9 @@ extern PropStat *global_c_propstat;
extern BAT *global_mbat;
extern BATiter global_mapi;
-#define USINGRDFSCAN 1
+#define USINGRDFSCAN 0
+
+#define APPLY_OPTIMIZATION_FOR_OPTIONAL 1 /* Instead of using
left join, we use a project with ifthenelse */
+ /* on the set of optional
columns */
#endif /*_SQL_RDF_H */
diff --git a/sql/backends/monet5/sql_rdf_jgraph.c
b/sql/backends/monet5/sql_rdf_jgraph.c
--- a/sql/backends/monet5/sql_rdf_jgraph.c
+++ b/sql/backends/monet5/sql_rdf_jgraph.c
@@ -1197,7 +1197,7 @@ void extract_prop_and_subj_from_exps(mvc
static
-void tranforms_exps(mvc *c, sql_rel *r, list *trans_select_exps, list
*trans_tbl_exps, str tblname, int colIdx, oid tmpPropId, str *atblname, str
*asubjcolname, list *sp_prj_exps){
+void tranforms_exps(mvc *c, sql_rel *r, list *trans_select_exps, list
*trans_tbl_exps, str tblname, int colIdx, oid tmpPropId, str *atblname, str
*asubjcolname, list *sp_prj_exps, list *base_column_exps){
list *tmpexps = NULL;
list *tmp_tbl_exps = NULL;
@@ -1279,10 +1279,12 @@ void tranforms_exps(mvc *c, sql_rel *r,
sql_column *tmpcol = get_rdf_column(c, tblname,
origcolname);
sql_exp *e = exp_alias(sa, tmpexp->rname,
tmpexp->name, origtblname, origcolname, &tmpcol->type, CARD_MULTI,
tmpcol->null, 0);
sql_exp *proj_e = exp_alias(sa, tmpexp->rname,
tmpexp->name, tmpexp->rname, tmpexp->name, &tmpcol->type, CARD_MULTI,
tmpcol->null, 0);
+ sql_exp *base_col_e = exp_copy(sa, proj_e);
printf("tmpcolname in rdf basetable is %s\n",
tmpcolname);
append(trans_tbl_exps, e);
if (sp_prj_exps) append(sp_prj_exps, proj_e);
+ if (base_column_exps) append(base_column_exps,
base_col_e);
}
if (strcmp(tmpexp->name, "s") == 0){
@@ -1293,7 +1295,7 @@ void tranforms_exps(mvc *c, sql_rel *r,
sql_column *tmpcol = get_rdf_column(c,
origtblname, origcolname);
sql_exp *e = exp_alias(sa, tmpexp->rname,
tmpexp->name, origtblname, origcolname, &tmpcol->type, CARD_MULTI,
tmpcol->null, 0);
sql_exp *proj_e = exp_alias(sa, tmpexp->rname,
tmpexp->name, tmpexp->rname, tmpexp->name, &tmpcol->type, CARD_MULTI,
tmpcol->null, 0);
-
+ sql_exp *base_col_e = exp_copy(sa, proj_e);
if (*atblname == NULL){
*atblname = GDKstrdup(tmpexp->rname);
@@ -1301,6 +1303,7 @@ void tranforms_exps(mvc *c, sql_rel *r,
}
append(trans_tbl_exps, e);
if (sp_prj_exps) append(sp_prj_exps, proj_e);
+ if (base_column_exps) append(base_column_exps,
base_col_e);
}
}
@@ -1309,7 +1312,7 @@ void tranforms_exps(mvc *c, sql_rel *r,
static
-void tranforms_mvprop_exps(mvc *c, sql_rel *r, mvPropRel *mvproprel, int
tblId, oid tblnameoid, int colIdx, oid tmpPropId, int isMVcol, list
*sp_prj_exps){
+void tranforms_mvprop_exps(mvc *c, sql_rel *r, mvPropRel *mvproprel, int
tblId, oid tblnameoid, int colIdx, oid tmpPropId, int isMVcol, list
*sp_prj_exps, list *base_column_exps){
list *tmpexps = NULL;
list *trans_select_exps = NULL;
@@ -1403,10 +1406,12 @@ void tranforms_mvprop_exps(mvc *c, sql_r
sql_column *tmpcol = get_rdf_column(c,
mvtblname, origcolname);
sql_exp *e = exp_alias(sa, tmpexp->rname,
tmpexp->name, origtblname, origcolname, &tmpcol->type, CARD_MULTI,
tmpcol->null, 0);
sql_exp *proj_e = exp_alias(sa, tmpexp->rname,
tmpexp->name, tmpexp->rname, tmpexp->name, &tmpcol->type, CARD_MULTI,
tmpcol->null, 0);
+ sql_exp *base_col_e = exp_copy(sa, proj_e);
printf("tmpmvcolname in rdf basetable is %s\n",
tmpmvcolname);
append(trans_tbl_exps, e);
append(sp_prj_exps, proj_e);
+ if (base_column_exps) append(base_column_exps,
base_col_e);
}
if (strcmp(tmpexp->name, "s") == 0){
@@ -1417,9 +1422,11 @@ void tranforms_mvprop_exps(mvc *c, sql_r
sql_column *tmpcol = get_rdf_column(c,
origtblname, origcolname);
sql_exp *e = exp_alias(sa, tmpexp->rname,
tmpexp->name, origtblname, origcolname, &tmpcol->type, CARD_MULTI,
tmpcol->null, 0);
sql_exp *proj_e = exp_alias(sa, tmpexp->rname,
tmpexp->name, tmpexp->rname, tmpexp->name, &tmpcol->type, CARD_MULTI,
tmpcol->null, 0);
+ sql_exp *base_col_e = exp_copy(sa, proj_e);
append(trans_tbl_exps, e);
append(sp_prj_exps, proj_e);
+ if (base_column_exps) append(base_column_exps,
base_col_e);
mvproprel->atblname = GDKstrdup(tmpexp->rname);
mvproprel->asubjcolname =
GDKstrdup(tmpexp->name);
@@ -1643,6 +1650,191 @@ sql_rel *connect_sp_select_and_mv_prop(m
}
+static
+list *single_exp_list(sql_allocator *sa, sql_exp *e){
+ list *lst = NULL;
+ lst = new_exp_list(sa);
+
+ append(lst, e);
+
+ return lst;
+}
+
+/*
+ * Return sys.isnull(e)
+ * Right now, it is more like e = null
+ * */
+static
+sql_exp* exp_isnull(sql_allocator *sa, sql_exp *e){
+ sql_exp *l = NULL;
+ sql_exp *r = NULL;
+ sql_exp *isnull_exp = NULL;
+
+ l = e;
+ r = exp_atom(sa, atom_general(sa, exp_subtype(l), NULL));
+
+ isnull_exp = exp_compare(sa, l, r, cmp_equal);
+
+ return isnull_exp;
+}
+
+static
+sql_exp* exp_isnotnull(sql_allocator *sa, sql_exp *e){
+ sql_exp *l = NULL;
+ sql_exp *r = NULL;
+ sql_exp *isnotnull_exp = NULL;
+
+ l = e;
+ r = exp_atom(sa, atom_general(sa, exp_subtype(l), NULL));
+
+ isnotnull_exp = exp_compare(sa, l, r, cmp_notequal);
+
+ return isnotnull_exp;
+}
+
+/*
+ * Create exps for optional set of columns
+ * e.g., base_column_exps [s1.p1, s1.p2, s1.p3]
+ * Then, the opt_exps will be
+ * sys.ifthenelse(
+ * sys.ifthenelse(
+ * sys.isnull(
+ * sys.or(sys.isnull(s1.p1),
+ * sys.or(sys.isnull(s1.p2), sys.isnull(s1.p3))
+ * )),
+ * boolean "false",
+ * sys.or(sys.isnull(s1.p1),
+ * sys.or(sys.isnull(s1.p2), sys.isnull(s1.p3))
+ * )
+ * ),
+ * NULL,
+ * s1.p1
+ * )
+ *
+ * Look complicated :)
+ *
+ * If it is a set of required columns, then put sys.isnotnull()
+ *
+ * NOTE THAT as set of base columns can look like s1.s, s1.o, s2.s, s2.o,...
+ * we only put the condition on o, while keeping s as original
+ *
+ * */
+
+static
+list *create_optional_exps(sql_allocator *sa, list *base_column_exps, int
isOptionalGroup, int contain_mv_col){
+ list *opt_exps = NULL ;
+ list *req_exps = NULL;
+ sql_exp *or_exp = NULL;
+ node *en = NULL;
+
+ list *only_o_exps = NULL; //keeping only o
+ only_o_exps = new_exp_list(sa);
+
+ for (en = base_column_exps->h; en; en = en->next){
+ sql_exp *o_exp = (sql_exp *) en->data;
+ assert(o_exp->type == e_column);
+ if (strcmp(o_exp->name, "o") == 0){
+ sql_exp *tmp_o_exp = exp_copy(sa, o_exp);
+ append(only_o_exps, tmp_o_exp);
+ }
+ }
+
+ if (isOptionalGroup){
+ opt_exps = new_exp_list(sa);
+ if (contain_mv_col){
+ printf("Do nothing for group of optional prop
containing mv col \n");
+ return base_column_exps;
+ } else {
+ node *first_node = only_o_exps->h;
+ sql_exp *first_exp = (sql_exp *) first_node->data;
+ sql_exp *first_isnull_exp = exp_isnull(sa, first_exp);
+
+ if (first_node->next){
+ for (en = first_node->next; en; en = en->next){
+ sql_exp *tmpexp = (sql_exp *) en->data;
+ sql_exp *tmp_isnull_exp = NULL;
+ assert(tmpexp->type == e_column);
+ tmp_isnull_exp = exp_isnull(sa,
tmpexp);
+
+ if (or_exp == NULL){
+ list *lst1 =
single_exp_list(sa, first_isnull_exp);
+ list *lst2 =
single_exp_list(sa, tmp_isnull_exp);
+ or_exp = exp_or(sa, lst1,
lst2);
+ } else {
+ list *lst1 =
single_exp_list(sa, or_exp);
+ list *lst2 =
single_exp_list(sa, tmp_isnull_exp);
+
+ or_exp = exp_or(sa, lst1,
lst2);
+ }
+ }
+ }
+
+ if (or_exp == NULL){
+ or_exp = exp_copy(sa, first_isnull_exp);
+ }
+
+ //Replace each column by ifthenelse
+ //e.g.: col1 = ifthenelse ( or (col1 == null, col2 ==
null, col3 ==null), null, col1)
+ for (en =base_column_exps->h; en; en = en->next){
+ sql_exp *tmpexp = (sql_exp *) en->data;
+ sql_exp *if_exp = NULL;
+ sql_exp *exp_null = NULL;
+ list *lst_ifthen;
+ list *lst_else;
+
+ assert(tmpexp->type == e_column);
+ assert(or_exp != NULL);
+
+ if (strcmp(tmpexp->name, "o") == 0){
+
+ exp_null = exp_atom(sa,
atom_general(sa, exp_subtype(tmpexp), NULL));
+
+ lst_ifthen = single_exp_list(sa,
exp_null);
+ lst_else = single_exp_list(sa, tmpexp);
+
+ if_exp = exp_if(sa, or_exp, lst_ifthen,
lst_else);
+
+ append(opt_exps, if_exp);
+ } else {
+ sql_exp *s_exp = NULL;
+ assert (strcmp(tmpexp->name, "s") ==
0);
+
+ s_exp = exp_copy(sa, tmpexp);
+
+ append(opt_exps, s_exp);
+ }
+
+ }
+
+ }
+
+ return opt_exps;
+ } else {
+ req_exps = new_exp_list(sa);
+ for (en = base_column_exps->h; en; en = en->next){
+ sql_exp *tmpexp = (sql_exp *) en->data;
+ sql_exp *exp_notnull = NULL;
+ assert(tmpexp->type == e_column);
+
+ exp_notnull = exp_isnotnull(sa, tmpexp);
+
+ append(req_exps, exp_notnull);
+ }
+
+ return req_exps;
+ }
+
+}
+
+static
+void append_sp_rdfscan_proj_exps(list *opt_col_exps, list
*sp_rdfscan_proj_exps){
+ node *en;
+ for (en = opt_col_exps->h; en; en = en->next){
+ sql_exp *tmpexp = (sql_exp *) en->data;
+ append(sp_rdfscan_proj_exps, tmpexp);
+ }
+}
+
/*
* Input:
* - A sub-join graph (jsg) that all nodes are connected by using inner join
@@ -1655,7 +1847,7 @@ sql_rel *connect_sp_select_and_mv_prop(m
* - sp_prj_exps stores all the columns should be selected in the "original
order"
* */
static
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
