misc changes

* sysvars should be REQUIRED_ARG
* fix a mix of US and UK spelling (use US)
* use consistent naming
* work if VEC_DISTANCE arguments are in the swapped order (const, col)
* work if VEC_DISTANCE argument is NULL/invalid or wrong length
* abort INSERT if the value is invalid or wrong length
* store the "number of neighbors" in a blob in endianness-independent way
* use field->store(longlong, bool) not field->store(double)
* a lot more error checking everywhere
* cleanup after errors
* simplify calling conventions, remove reinterpret_cast's
* todo/XXX comments
* whitespaces
* use float consistently

memory management is still totally PoC quality

Initial HNSW implementation

mysql-test/main/mysqld--help.result

@@ -402,10 +402,10 @@ The following specify which files/extra groups are read (specified before remain
  height-balanced, DOUBLE_PREC_HB - double precision
  height-balanced, JSON_HB - height-balanced, stored as
  JSON
- --hnsw-ef-constructor 
+ --hnsw-ef-constructor=# 
  hnsw_ef_constructor
- --hnsw-ef-search    hnsw_ef_search
- --hnsw-max-connection-per-layer 
+ --hnsw-ef-search=#  hnsw_ef_search
+ --hnsw-max-connection-per-layer=# 
  hnsw_max_connection_per_layer
  --host-cache-size=# How many host names should be cached to avoid resolving
  (Automatically configured unless set explicitly)

mysql-test/main/vector.result

@@ -80,6 +80,21 @@ id	d
 9	0.4719976290006591
 10	0.5069011044450041
 3	0.5865673124650332
+select id,vec_distance(x'b047263c9f87233fcfd27e3eae493e3f0329f43e', v) d from t1 order by d limit 3;
+id	d
+9	0.4719976290006591
+10	0.5069011044450041
+3	0.5865673124650332
+select id>0,vec_distance(v, NULL) d from t1 order by d limit 3;
+id>0	d
+1	NULL
+1	NULL
+1	NULL
+select id>0,vec_distance(v, x'123456') d from t1 order by d limit 3;
+id>0	d
+1	NULL
+1	NULL
+1	NULL
 select t1.id as id1, t2.id as id2, vec_distance(t1.v, t2.v) from t1, t1 as t2 order by 3,1,2;
 id1	id2	vec_distance(t1.v, t2.v)
 1	1	0
@@ -182,5 +197,11 @@ id1	id2	vec_distance(t1.v, t2.v)
 9	8	1.2575258643523053
 7	8	1.288239696195716
 8	7	1.288239696195716
+insert t1 (v) values ('');
+ERROR 22007: Incorrect vector value: '...' for column `test`.`t1`.`v` at row 1
+insert t1 (v) values (x'1234');
+ERROR 22007: Incorrect vector value: '...' for column `test`.`t1`.`v` at row 1
+insert t1 (v) values (x'12345678');
+ERROR 22007: Incorrect vector value: '...' for column `test`.`t1`.`v` at row 1
 drop table t1;
 db.opt

mysql-test/main/vector.test

@@ -10,7 +10,7 @@ create table t1 (id int auto_increment primary key, v blob not null, vector inde
 show create table t1;
 show keys from t1;
 query_vertical select * from information_schema.statistics where table_name='t1';
-# print unpack(H40,pack(f5,map{rand}1..5))
+# print unpack("H*",pack("f*",map{rand}1..5))
 insert t1 (v) values (x'e360d63ebe554f3fcdbc523f4522193f5236083d'),
                      (x'f511303f72224a3fdd05fe3eb22a133ffae86a3f'),
                      (x'f09baa3ea172763f123def3e0c7fe53e288bf33e'),
@@ -24,8 +24,23 @@ insert t1 (v) values (x'e360d63ebe554f3fcdbc523f4522193f5236083d'),
 
 select id, hex(v) from t1;
 flush tables;
+# test with a valid query vector
 select id,vec_distance(v, x'b047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
+# swapped arguments
+select id,vec_distance(x'b047263c9f87233fcfd27e3eae493e3f0329f43e', v) d from t1 order by d limit 3;
+# test with NULL (id is unpredictable)
+select id>0,vec_distance(v, NULL) d from t1 order by d limit 3;
+# test with invalid query vector (id is unpredictable)
+select id>0,vec_distance(v, x'123456') d from t1 order by d limit 3;
 select t1.id as id1, t2.id as id2, vec_distance(t1.v, t2.v) from t1, t1 as t2 order by 3,1,2;
+
+--error ER_TRUNCATED_WRONG_VALUE_FOR_FIELD
+insert t1 (v) values ('');
+--error ER_TRUNCATED_WRONG_VALUE_FOR_FIELD
+insert t1 (v) values (x'1234');
+--error ER_TRUNCATED_WRONG_VALUE_FOR_FIELD
+insert t1 (v) values (x'12345678');
+
 drop table t1;
 let $datadir=`select @@datadir`;
 list_files $datadir/test;

mysql-test/suite/sys_vars/r/sysvars_server_notembedded.result

@@ -1431,7 +1431,7 @@ NUMERIC_MAX_VALUE	4294967295
 NUMERIC_BLOCK_SIZE	1
 ENUM_VALUE_LIST	NULL
 READ_ONLY	NO
-COMMAND_LINE_ARGUMENT	NONE
+COMMAND_LINE_ARGUMENT	REQUIRED
 VARIABLE_NAME	HNSW_EF_SEARCH
 VARIABLE_SCOPE	SESSION
 VARIABLE_TYPE	INT UNSIGNED
@@ -1441,7 +1441,7 @@ NUMERIC_MAX_VALUE	4294967295
 NUMERIC_BLOCK_SIZE	1
 ENUM_VALUE_LIST	NULL
 READ_ONLY	NO
-COMMAND_LINE_ARGUMENT	NONE
+COMMAND_LINE_ARGUMENT	REQUIRED
 VARIABLE_NAME	HNSW_MAX_CONNECTION_PER_LAYER
 VARIABLE_SCOPE	SESSION
 VARIABLE_TYPE	INT UNSIGNED
@@ -1451,7 +1451,7 @@ NUMERIC_MAX_VALUE	4294967295
 NUMERIC_BLOCK_SIZE	1
 ENUM_VALUE_LIST	NULL
 READ_ONLY	NO
-COMMAND_LINE_ARGUMENT	NONE
+COMMAND_LINE_ARGUMENT	REQUIRED
 VARIABLE_NAME	HOSTNAME
 VARIABLE_SCOPE	GLOBAL
 VARIABLE_TYPE	VARCHAR

sql/field.h

@@ -965,6 +965,10 @@ class Field: public Value_source
   {
     return store(to, length, &my_charset_bin);
   }
+  int store_binary(const uchar *to, size_t length)
+  {
+    return store_binary((const char*)(to), length);
+  }
   virtual int  store_hex_hybrid(const char *str, size_t length);
   virtual int  store(double nr)=0;
   virtual int  store(longlong nr, bool unsigned_val)=0;

sql/item_vectorfunc.h

@@ -51,10 +51,17 @@ class Item_func_vec_distance: public Item_real_func
     static LEX_CSTRING name= {STRING_WITH_LEN("vec_distance") };
     return name;
   }
+  Item *get_const_arg() const
+  {
+    if (args[0]->type() == Item::FIELD_ITEM && args[1]->const_item())
+      return args[1];
+    if (args[1]->type() == Item::FIELD_ITEM && args[0]->const_item())
+      return args[0];
+    return NULL;
+  }
   key_map part_of_sortkey() const override;
   Item *do_get_copy(THD *thd) const override
   { return get_item_copy<Item_func_vec_distance>(thd, this); }
-  virtual ~Item_func_vec_distance() {};
 };
 
 

sql/sys_vars.cc

@@ -7388,18 +7388,18 @@ static Sys_var_enum Sys_block_encryption_mode(
 static Sys_var_uint Sys_hnsw_ef_search(
        "hnsw_ef_search",
        "hnsw_ef_search",
-       SESSION_VAR(hnsw_ef_search), CMD_LINE(NO_ARG),
+       SESSION_VAR(hnsw_ef_search), CMD_LINE(REQUIRED_ARG),
        VALID_RANGE(0, UINT_MAX), DEFAULT(10),
        BLOCK_SIZE(1));
 static Sys_var_uint Sys_hnsw_ef_constructor(
        "hnsw_ef_constructor",
        "hnsw_ef_constructor",
-       SESSION_VAR(hnsw_ef_constructor), CMD_LINE(NO_ARG),
+       SESSION_VAR(hnsw_ef_constructor), CMD_LINE(REQUIRED_ARG),
        VALID_RANGE(0, UINT_MAX), DEFAULT(10),
        BLOCK_SIZE(1));
 static Sys_var_uint Sys_hnsw_max_connection_per_layer(
        "hnsw_max_connection_per_layer",
        "hnsw_max_connection_per_layer",
-       SESSION_VAR(hnsw_max_connection_per_layer), CMD_LINE(NO_ARG),
+       SESSION_VAR(hnsw_max_connection_per_layer), CMD_LINE(REQUIRED_ARG),
        VALID_RANGE(0, UINT_MAX), DEFAULT(50),
        BLOCK_SIZE(1));