Add small bugfixes and refactoring.

Reviewed by: @Alena0704

Author: Daniil Anisimov

aqo.c

@@ -61,7 +61,6 @@ static const struct config_enum_entry format_options[] = {
 };
 
 /* Parameters of autotuning */
-int			aqo_stat_size = STAT_SAMPLE_SIZE;
 int			auto_tuning_window_size = 5;
 double		auto_tuning_exploration = 0.1;
 int			auto_tuning_max_iterations = 50;

aqo.h

@@ -211,7 +211,6 @@ extern double predicted_ppi_rows;
 extern double fss_ppi_hash;
 
 /* Parameters of autotuning */
-extern int	aqo_stat_size;
 extern int	auto_tuning_window_size;
 extern double auto_tuning_exploration;
 extern int	auto_tuning_max_iterations;

hash.c

@@ -344,7 +344,7 @@ get_clause_hash(Expr *clause, int nargs, int *args_hash, int *eclass_hash)
 /*
  * Computes hash for given string.
  */
-int
+static int
 get_str_hash(const char *str)
 {
 	return DatumGetInt32(hash_any((const unsigned char *) str,
@@ -381,7 +381,7 @@ get_int_array_hash(int *arr, int len)
  * Sorts given array in-place to compute hash.
  * The hash is order-insensitive.
  */
-int
+static int
 get_unsorted_unsafe_int_array_hash(int *arr, int len)
 {
 	qsort(arr, len, sizeof(*arr), int_cmp);
@@ -396,7 +396,7 @@ get_unsorted_unsafe_int_array_hash(int *arr, int len)
  * using 'hash_any'.
  * Frees allocated memory before returning hash.
  */
-int
+static int
 get_unordered_int_list_hash(List *lst)
 {
 	int			i = 0;
@@ -448,7 +448,7 @@ replace_patterns(const char *str, const char *start_pattern,
  * Computes hash for given feature subspace.
  * Hash is supposed to be clause-order-insensitive.
  */
-int
+static int
 get_fss_hash(int clauses_hash, int eclasses_hash, int relidslist_hash)
 {
 	int			hashes[3];
@@ -517,7 +517,7 @@ remove_locations(const char *str)
  * Returns index of given value in given sorted integer array
  * or -1 if not found.
  */
-int
+static int
 get_id_in_sorted_int_array(int val, int n, int *arr)
 {
 	int		   *i;
@@ -536,7 +536,7 @@ get_id_in_sorted_int_array(int val, int n, int *arr)
  * Returns class of equivalence for given argument hash or 0 if such hash
  * does not belong to any equivalence class.
  */
-int
+static int
 get_arg_eclass(int arg_hash, int nargs, int *args_hash, int *eclass_hash)
 {
 	int			di = get_id_in_sorted_int_array(arg_hash, nargs, args_hash);
@@ -551,7 +551,7 @@ get_arg_eclass(int arg_hash, int nargs, int *args_hash, int *eclass_hash)
  * Builds list of non-constant arguments of equivalence clauses
  * of given clauselist.
  */
-void
+static void
 get_clauselist_args(List *clauselist, int *nargs, int **args_hash)
 {
 	RestrictInfo *rinfo;
@@ -597,7 +597,7 @@ get_clauselist_args(List *clauselist, int *nargs, int **args_hash)
 /*
  * Returns class of an object in disjoint set.
  */
-int
+static int
 disjoint_set_get_parent(int *p, int v)
 {
 	if (p[v] == -1)
@@ -609,7 +609,7 @@ disjoint_set_get_parent(int *p, int v)
 /*
  * Merges two equivalence classes in disjoint set.
  */
-void
+static void
 disjoint_set_merge_eclasses(int *p, int v1, int v2)
 {
 	int			p1,
@@ -629,7 +629,7 @@ disjoint_set_merge_eclasses(int *p, int v1, int v2)
 /*
  * Constructs disjoint set on arguments.
  */
-int *
+static int *
 perform_eclasses_join(List *clauselist, int nargs, int *args_hash)
 {
 	RestrictInfo *rinfo;
@@ -706,7 +706,7 @@ get_eclasses(List *clauselist, int *nargs, int **args_hash, int **eclass_hash)
 /*
  * Checks whether the given char is brace, i. e. '{' or '}'.
  */
-bool
+static bool
 is_brace(char ch)
 {
 	return ch == '{' || ch == '}';
@@ -715,7 +715,7 @@ is_brace(char ch)
 /*
  * Returns whether arguments list contain constants.
  */
-bool
+static bool
 has_consts(List *lst)
 {
 	ListCell   *l;
@@ -729,7 +729,7 @@ has_consts(List *lst)
 /*
  * Returns pointer on the args list in clause or NULL.
  */
-List **
+static List **
 get_clause_args_ptr(Expr *clause)
 {
 	switch (clause->type)
@@ -755,7 +755,7 @@ get_clause_args_ptr(Expr *clause)
 /*
  * Returns whether the clause is an equivalence clause.
  */
-bool
+static bool
 clause_is_eq_clause(Expr *clause)
 {
 	/* TODO: fix this horrible mess */

postprocessing.c

@@ -173,7 +173,7 @@ learn_sample(aqo_obj_stat *ctx, RelSortOut *rels,
  * For given node specified by clauselist, relidslist and join_type restores
  * the same selectivities of clauses as were used at query optimization stage.
  */
-List *
+static List *
 restore_selectivities(List *clauselist, List *relidslist, JoinType join_type,
 					  bool was_parametrized)
 {
@@ -336,7 +336,7 @@ should_learn(PlanState *ps, AQOPlanNode *node, aqo_obj_stat *ctx,
 					 "predicted rows: %.0lf, updated prediction: %.0lf",
 					 query_context.query_hash, node->fss, predicted, nrows);
 
-			*rfactor = 0.9 * (RELIABILITY_MAX - RELIABILITY_MIN);
+			*rfactor = RELIABILITY_MIN + 0.9 * (RELIABILITY_MAX - RELIABILITY_MIN);
 			return true;
 		}
 	}

preprocessing.c

@@ -69,7 +69,7 @@
 /* List of feature spaces, that are processing in this backend. */
 List *cur_classes = NIL;
 
-int aqo_join_threshold = 0;
+int aqo_join_threshold = 3;
 
 static planner_hook_type	aqo_planner_next = NULL;
 

storage.c

@@ -100,7 +100,7 @@ static bool _aqo_stat_remove(uint64 queryid);
 static bool _aqo_queries_remove(uint64 queryid);
 static bool _aqo_qtexts_remove(uint64 queryid);
 static bool _aqo_data_remove(data_key *key);
-static bool neirest_neighbor(double **matrix, int old_rows, double *neighbor, int cols);
+static bool nearest_neighbor(double **matrix, int old_rows, double *neighbor, int cols);
 static double fs_distance(double *a, double *b, int len);
 
 PG_FUNCTION_INFO_V1(aqo_query_stat);
@@ -143,7 +143,7 @@ update_fss_ext(uint64 fs, int fss, OkNNrdata *data, List *reloids)
 /*
  * Forms ArrayType object for storage from simple C-array matrix.
  */
-ArrayType *
+static ArrayType *
 form_matrix(double *matrix, int nrows, int ncols)
 {
 	Datum	   *elems;
@@ -375,8 +375,8 @@ aqo_query_stat(PG_FUNCTION_ARGS)
 	MemoryContext		per_query_ctx;
 	MemoryContext		oldcontext;
 	Tuplestorestate	   *tupstore;
-	Datum				values[TOTAL_NCOLS + 1];
-	bool				nulls[TOTAL_NCOLS + 1];
+	Datum				values[TOTAL_NCOLS];
+	bool				nulls[TOTAL_NCOLS];
 	HASH_SEQ_STATUS		hash_seq;
 	StatEntry		   *entry;
 
@@ -408,13 +408,11 @@ aqo_query_stat(PG_FUNCTION_ARGS)
 
 	MemoryContextSwitchTo(oldcontext);
 
-	memset(nulls, 0, TOTAL_NCOLS + 1);
+	memset(nulls, 0, TOTAL_NCOLS);
 	LWLockAcquire(&aqo_state->stat_lock, LW_SHARED);
 	hash_seq_init(&hash_seq, stat_htab);
 	while ((entry = hash_seq_search(&hash_seq)) != NULL)
 	{
-		memset(nulls, 0, TOTAL_NCOLS + 1);
-
 		values[QUERYID] = Int64GetDatum(entry->queryid);
 		values[NEXECS] = Int64GetDatum(entry->execs_without_aqo);
 		values[NEXECS_AQO] = Int64GetDatum(entry->execs_with_aqo);
@@ -1507,8 +1505,8 @@ fs_distance(double *a, double *b, int len)
 	return res;
 }
 
-bool
-neirest_neighbor(double **matrix, int old_rows, double *neibour, int cols)
+static bool
+nearest_neighbor(double **matrix, int old_rows, double *neibour, int cols)
 {
 	int i;
 	for (i=0; i<old_rows; i++)
@@ -1538,7 +1536,9 @@ build_knn_matrix(OkNNrdata *data, const OkNNrdata *temp_data, double *features)
 
 			for (i = 0; i < temp_data->rows; i++)
 			{
-				if (k < aqo_K && !neirest_neighbor(data->matrix, old_rows, data->matrix[i], data->cols))
+				if (k < aqo_K && !nearest_neighbor(data->matrix, old_rows,
+												   temp_data->matrix[i],
+												   data->cols))
 				{
 					memcpy(data->matrix[k], temp_data->matrix[i], data->cols * sizeof(double));
 					data->rfactors[k] = temp_data->rfactors[i];
@@ -1902,8 +1902,8 @@ aqo_queries(PG_FUNCTION_ARGS)
 	MemoryContext		per_query_ctx;
 	MemoryContext		oldcontext;
 	Tuplestorestate	   *tupstore;
-	Datum				values[AQ_TOTAL_NCOLS + 1];
-	bool				nulls[AQ_TOTAL_NCOLS + 1];
+	Datum				values[AQ_TOTAL_NCOLS];
+	bool				nulls[AQ_TOTAL_NCOLS];
 	HASH_SEQ_STATUS		hash_seq;
 	QueriesEntry	   *entry;
 
@@ -1935,12 +1935,12 @@ aqo_queries(PG_FUNCTION_ARGS)
 
 	MemoryContextSwitchTo(oldcontext);
 
+	memset(nulls, 0, AQ_TOTAL_NCOLS);
+
 	LWLockAcquire(&aqo_state->queries_lock, LW_SHARED);
 	hash_seq_init(&hash_seq, queries_htab);
 	while ((entry = hash_seq_search(&hash_seq)) != NULL)
 	{
-		memset(nulls, 0, AQ_TOTAL_NCOLS + 1);
-
 		values[AQ_QUERYID] = Int64GetDatum(entry->queryid);
 		values[AQ_FS] = Int64GetDatum(entry->fs);
 		values[AQ_LEARN_AQO] = BoolGetDatum(entry->learn_aqo);
@@ -2142,7 +2142,7 @@ aqo_queries_find(uint64 queryid, QueryContextData *ctx)
 
 /*
  * Function for update and save value of smart statement timeout
- * for query in aqu_queries table
+ * for query in aqo_queries table
  */
 bool
 update_query_timeout(uint64 queryid, int64 smart_timeout)
@@ -2515,6 +2515,8 @@ aqo_cardinality_error(PG_FUNCTION_ARGS)
 	LWLockAcquire(&aqo_state->queries_lock, LW_SHARED);
 	LWLockAcquire(&aqo_state->stat_lock, LW_SHARED);
 
+	memset(nulls, 0, AQE_TOTAL_NCOLS * sizeof(nulls[0]));
+
 	hash_seq_init(&hash_seq, queries_htab);
 	while ((qentry = hash_seq_search(&hash_seq)) != NULL)
 	{
@@ -2523,8 +2525,6 @@ aqo_cardinality_error(PG_FUNCTION_ARGS)
 		int64	nexecs;
 		int		nvals;
 
-		memset(nulls, 0, AQE_TOTAL_NCOLS * sizeof(nulls[0]));
-
 		sentry = (StatEntry *) hash_search(stat_htab, &qentry->queryid,
 										   HASH_FIND, &found);
 		if (!found)
@@ -2609,6 +2609,8 @@ aqo_execution_time(PG_FUNCTION_ARGS)
 	LWLockAcquire(&aqo_state->queries_lock, LW_SHARED);
 	LWLockAcquire(&aqo_state->stat_lock, LW_SHARED);
 
+	memset(nulls, 0, ET_TOTAL_NCOLS * sizeof(nulls[0]));
+
 	hash_seq_init(&hash_seq, queries_htab);
 	while ((qentry = hash_seq_search(&hash_seq)) != NULL)
 	{
@@ -2618,8 +2620,6 @@ aqo_execution_time(PG_FUNCTION_ARGS)
 		int		nvals;
 		double	tm = 0;
 
-		memset(nulls, 0, ET_TOTAL_NCOLS * sizeof(nulls[0]));
-
 		sentry = (StatEntry *) hash_search(stat_htab, &qentry->queryid,
 										   HASH_FIND, &found);
 		if (!found)