Merge tulin@bk-internal.mysql.com:/home/bk/mysql-4.1

into poseidon.ndb.mysql.com:/home/tomas/mysql-4.1-ndb-merge

client/client_priv.h

@@ -45,5 +45,5 @@ enum options_client
   OPT_COMPATIBLE, OPT_RECONNECT, OPT_DELIMITER, OPT_SECURE_AUTH,
   OPT_OPEN_FILES_LIMIT, OPT_SET_CHARSET, OPT_CREATE_OPTIONS,
   OPT_START_POSITION, OPT_STOP_POSITION, OPT_START_DATETIME, OPT_STOP_DATETIME,
-  OPT_SIGINT_IGNORE
+  OPT_SIGINT_IGNORE, OPT_HEXBLOB
 };

client/mysqldump.c

@@ -81,7 +81,8 @@ static my_bool  verbose=0,tFlag=0,cFlag=0,dFlag=0,quick= 1, extended_insert= 1,
                 opt_alldbs=0,opt_create_db=0,opt_first_slave=0,opt_set_charset,
 		opt_autocommit=0,opt_master_data,opt_disable_keys=1,opt_xml=0,
 		opt_delete_master_logs=0, tty_password=0,
-		opt_single_transaction=0, opt_comments= 0, opt_compact= 0;
+		opt_single_transaction=0, opt_comments= 0, opt_compact= 0,
+		opt_hex_blob=0;
 static ulong opt_max_allowed_packet, opt_net_buffer_length;
 static MYSQL mysql_connection,*sock=0;
 static char  insert_pat[12 * 1024],*opt_password=0,*current_user=0,
@@ -316,6 +317,8 @@ static struct my_option my_long_options[] =
   {"comments", 'i', "Write additional information.",
    (gptr*) &opt_comments, (gptr*) &opt_comments, 0, GET_BOOL, NO_ARG,
    1, 0, 0, 0, 0, 0},
+  {"hex-blob", OPT_HEXBLOB, "Dump BLOBs in HEX. this mode does not work with extended-insert",
+   (gptr*) &opt_hex_blob, (gptr*) &opt_hex_blob, 0, GET_BOOL, NO_ARG, 0, 0, 0, 0, 0, 0},
   {0, 0, 0, 0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0}
 };
 
@@ -1507,6 +1510,7 @@ static void dumpTable(uint numFields, char *table)
 
       for (i = 0; i < mysql_num_fields(res); i++)
       {
+        int is_blob;
 	if (!(field = mysql_fetch_field(res)))
 	{
 	  sprintf(query,"%s: Not enough fields from table %s! Aborting.\n",
@@ -1515,6 +1519,17 @@ static void dumpTable(uint numFields, char *table)
 	  error= EX_CONSCHECK;
 	  goto err;
 	}
+	
+	/*
+	   63 is my_charset_bin. If charsetnr is not 63,
+	   we have not a BLOB but a TEXT column. 
+	   we'll dump it in hex only BLOB columns.
+	*/
+        is_blob= (opt_hex_blob && field->charsetnr == 63 &&
+                  (field->type == FIELD_TYPE_BLOB ||
+                   field->type == FIELD_TYPE_LONG_BLOB ||
+                   field->type == FIELD_TYPE_MEDIUM_BLOB ||
+                   field->type == FIELD_TYPE_TINY_BLOB)) ? 1 : 0;
 	if (extended_insert)
 	{
 	  ulong length = lengths[i];
@@ -1535,12 +1550,28 @@ static void dumpTable(uint numFields, char *table)
 		  error= EX_EOM;
 		  goto err;
 		}
-		dynstr_append(&extended_row,"'");
-		extended_row.length +=
-		  mysql_real_escape_string(&mysql_connection,
-					   &extended_row.str[extended_row.length],row[i],length);
-		extended_row.str[extended_row.length]='\0';
-		dynstr_append(&extended_row,"'");
+                if (opt_hex_blob && is_blob)
+                {
+                  ulong counter;
+                  unsigned char *ptr= row[i];
+                  dynstr_append(&extended_row, "0x");
+                  for (counter = 0; counter < lengths[i]; counter++)
+                  {
+                    char xx[3];
+                    sprintf(xx, "%02X", ptr[counter]);
+                    dynstr_append(&extended_row, xx);
+                  }
+                }
+                else
+                {
+                  dynstr_append(&extended_row,"'");
+                  extended_row.length +=
+                  mysql_real_escape_string(&mysql_connection,
+                                           &extended_row.str[extended_row.length],
+                                           row[i],length);
+                  extended_row.str[extended_row.length]='\0';
+                  dynstr_append(&extended_row,"'");
+                }
 	      }
 	      else
 	      {
@@ -1591,8 +1622,20 @@ static void dumpTable(uint numFields, char *table)
 		print_quoted_xml(md_result_file, row[i], lengths[i]);
 		fputs("</field>\n", md_result_file);
 	      }
-	      else
-		unescape(md_result_file, row[i], lengths[i]);
+	      else if (opt_hex_blob && is_blob)
+              { /* sakaik got this idea. */
+                ulong counter;
+                char xx[4];
+                unsigned char *ptr= row[i];
+                fputs("0x", md_result_file);
+                for (counter = 0; counter < lengths[i]; counter++)
+                {
+                  sprintf(xx, "%02X", ptr[counter]);
+                  fputs(xx, md_result_file);
+                }
+              }
+              else
+                unescape(md_result_file, row[i], lengths[i]);
 	    }
 	    else
 	    {

innobase/dict/dict0dict.c

@@ -53,6 +53,30 @@ rw_lock_t	dict_operation_lock;	/* table create, drop, etc. reserve
 /* Identifies generated InnoDB foreign key names */
 static char	dict_ibfk[] = "_ibfk_";
 
+/**********************************************************************
+Compares NUL-terminated UTF-8 strings case insensitively.
+
+NOTE: the prototype of this function is copied from ha_innodb.cc! If you change
+this function, you MUST change also the prototype here! */
+extern
+int
+innobase_strcasecmp(
+/*================*/
+				/* out: 0 if a=b, <0 if a<b, >1 if a>b */
+	const char*	a,	/* in: first string to compare */
+	const char*	b);	/* in: second string to compare */
+
+/**********************************************************************
+Makes all characters in a NUL-terminated UTF-8 string lower case.
+
+NOTE: the prototype of this function is copied from ha_innodb.cc! If you change
+this function, you MUST change also the prototype here! */
+extern
+void
+innobase_casedn_str(
+/*================*/
+	char*	a);	/* in/out: string to put in lower case */
+
 /**************************************************************************
 Adds a column to the data dictionary hash table. */
 static
@@ -2066,7 +2090,7 @@ dict_foreign_find_index(
 					break;
 				}
 
-				if (0 != ut_cmp_in_lower_case(columns[i],
+				if (0 != innobase_strcasecmp(columns[i],
 								col_name)) {
 				  	break;
 				}
@@ -2436,7 +2460,7 @@ dict_scan_col(
 
 			col = dict_table_get_nth_col(table, i);
 
-			if (0 == ut_cmp_in_lower_case(col->name, *name)) {
+			if (0 == innobase_strcasecmp(col->name, *name)) {
 		    		/* Found */
 
 		    		*success = TRUE;
@@ -2528,30 +2552,19 @@ dict_scan_table_name(
 
 	table_name_len = strlen(table_name);
 
+	/* Copy database_name, '/', table_name, '\0' */
 	ref = mem_heap_alloc(heap, database_name_len + table_name_len + 2);
-
-#ifdef __WIN__
-	ut_cpy_in_lower_case(ref, database_name, database_name_len);
-#else
-	if (srv_lower_case_table_names) {
-		ut_cpy_in_lower_case(ref, database_name, database_name_len);
-	} else {
-		memcpy(ref, database_name, database_name_len);
-	}
-#endif
-	(ref)[database_name_len] = '/';
-
-#ifdef __WIN__
-	ut_cpy_in_lower_case(ref + database_name_len + 1,
-					table_name, table_name_len + 1);
-#else
+	memcpy(ref, database_name, database_name_len);
+	ref[database_name_len] = '/';
+	memcpy(ref + database_name_len + 1, table_name, table_name_len + 1);
+#ifndef __WIN__
 	if (srv_lower_case_table_names) {
-		ut_cpy_in_lower_case(ref + database_name_len + 1,
-					table_name, table_name_len + 1);
-	} else {
-		strcpy(ref + database_name_len + 1, table_name);
+#endif /* !__WIN__ */
+		/* The table name is always put to lower case on Windows. */
+		innobase_casedn_str(ref);
+#ifndef __WIN__
 	}
-#endif
+#endif /* !__WIN__ */
 
 	*success = TRUE;
 	*ref_name = ref;

innobase/include/ut0byte.h

@@ -229,25 +229,6 @@ ut_bit_set_nth(
 	ulint	a,	/* in: ulint */
 	ulint	n,	/* in: nth bit requested */
 	ibool	val);	/* in: value for the bit to set */
-/****************************************************************
-Copies a string to a memory location, setting characters to lower case. */
-
-void
-ut_cpy_in_lower_case(
-/*=================*/
-	char*		dest,	/* in: destination */
-	const char*	source,	/* in: source */
-	ulint		len);	/* in: string length */
-/****************************************************************
-Compares two strings when converted to lower case. */
-
-int
-ut_cmp_in_lower_case(
-/*=================*/
-				/* out: -1, 0, 1 if str1 < str2, str1 == str2,
-				str1 > str2, respectively */
-	const char*	str1,	/* in: string1 */
-	const char*	str2);	/* in: string2 */
 
 #ifndef UNIV_NONINL
 #include "ut0byte.ic"

innobase/os/os0file.c

@@ -2010,6 +2010,11 @@ try_again:
 
 		return(TRUE);
 	}
+
+	fprintf(stderr,
+"InnoDB: Error: tried to read %lu bytes at offset %lu %lu.\n"
+"InnoDB: Was only able to read %ld.\n", (ulong)n, (ulong)offset_high,
+					(ulong)offset, (long)ret);
 #endif	
 #ifdef __WIN__
 error_handling:

innobase/srv/srv0srv.c

@@ -951,7 +951,13 @@ retry:
 
 		trx->op_info = "sleeping before joining InnoDB queue";
 
-		os_thread_sleep(50000);
+		/* Peter Zaitsev suggested that we take the sleep away
+		altogether. But the sleep may be good in pathological
+		situations of lots of thread switches. Simply put some
+		threads aside for a while to reduce the number of thread
+		switches. */
+
+		os_thread_sleep(10000);
 
 		trx->op_info = "";
 

innobase/ut/ut0byte.c

@@ -29,51 +29,3 @@ ut_dulint_sort(dulint* arr, dulint* aux_arr, ulint low, ulint high)
 	UT_SORT_FUNCTION_BODY(ut_dulint_sort, arr, aux_arr, low, high,
 				ut_dulint_cmp);
 }
-
-/****************************************************************
-Copies a string to a memory location, setting characters to lower case. */
-
-void
-ut_cpy_in_lower_case(
-/*=================*/
-	char*		dest,	/* in: destination */
-	const char*	source,	/* in: source */
-	ulint		len)	/* in: string length */
-{
-        ulint i;
-
-	for (i = 0; i < len; i++) {
-	        dest[i] = tolower(source[i]);
-	}
-}
-
-/****************************************************************
-Compares two strings when converted to lower case. */
-
-int
-ut_cmp_in_lower_case(
-/*=================*/
-				/* out: -1, 0, 1 if str1 < str2, str1 == str2,
-				str1 > str2, respectively */
-	const char*	str1,	/* in: string1 */
-	const char*	str2)	/* in: string2 */
-{
-	for (;;) {
-		int c1, c2;
-		if (!*str1) {
-			return(*str2 ? -1 : 0);
-		} else if (!*str2) {
-			return 1;
-		}
-		c1 = tolower(*str1++);
-		c2 = tolower(*str2++);
-		if (c1 < c2) {
-			return(-1);
-		}
-		if (c1 > c2) {
-			return(1);
-		}
-	}
-
-	return(0);
-}

mysql-test/r/union.result

@@ -1033,3 +1033,81 @@ a
 No
 aaa,bbb
 drop table t1,t2,t3,t4;
+create table t1 as
+(select _latin1'test') union
+(select _latin1'TEST') union
+(select _latin1'TeST');
+show create table t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `test` char(4) NOT NULL default ''
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+select count(*) from t1;
+count(*)
+1
+drop table t1;
+create table t1 as
+(select _latin1'test' collate latin1_bin) union
+(select _latin1'TEST') union
+(select _latin1'TeST');
+show create table t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `_latin1'test' collate latin1_bin` char(4) character set latin1 collate latin1_bin NOT NULL default ''
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+select count(*) from t1;
+count(*)
+3
+drop table t1;
+create table t1 as
+(select _latin1'test') union
+(select _latin1'TEST' collate latin1_bin) union
+(select _latin1'TeST');
+show create table t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `test` char(4) character set latin1 collate latin1_bin NOT NULL default ''
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+select count(*) from t1;
+count(*)
+3
+drop table t1;
+create table t1 as
+(select _latin1'test') union
+(select _latin1'TEST') union
+(select _latin1'TeST' collate latin1_bin);
+show create table t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `test` char(4) character set latin1 collate latin1_bin NOT NULL default ''
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+select count(*) from t1;
+count(*)
+3
+drop table t1;
+create table t2 (
+a char character set latin1 collate latin1_swedish_ci,
+b char character set latin1 collate latin1_bin);
+create table t1 as
+(select a from t2) union
+(select b from t2);
+ERROR HY000: Illegal mix of collations for operation 'UNION'
+create table t1 as
+(select a collate latin1_german1_ci from t2) union
+(select b from t2);
+show create table t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a collate latin1_german1_ci` char(1) character set latin1 collate latin1_german1_ci default NULL
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+drop table t1;
+create table t1 as
+(select a from t2) union
+(select b collate latin1_german1_ci from t2);
+show create table t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` char(1) character set latin1 collate latin1_german1_ci default NULL
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+drop table t1;
+drop table t2;

mysql-test/t/union.test

@@ -595,3 +595,58 @@ select a as a from t3 union select "1";
 select a as a from t4 union select a from t3;
 select a as a from t1 union select a from t4;
 drop table t1,t2,t3,t4;
+
+#
+# Bug #6139 UNION doesn't understand collate in the column of second select
+#
+create table t1 as
+(select _latin1'test') union
+(select _latin1'TEST') union
+(select _latin1'TeST');
+show create table t1;
+select count(*) from t1;
+drop table t1;
+
+create table t1 as
+(select _latin1'test' collate latin1_bin) union
+(select _latin1'TEST') union
+(select _latin1'TeST');
+show create table t1;
+select count(*) from t1;
+drop table t1;
+
+create table t1 as
+(select _latin1'test') union
+(select _latin1'TEST' collate latin1_bin) union
+(select _latin1'TeST');
+show create table t1;
+select count(*) from t1;
+drop table t1;
+
+create table t1 as
+(select _latin1'test') union
+(select _latin1'TEST') union
+(select _latin1'TeST' collate latin1_bin);
+show create table t1;
+select count(*) from t1;
+drop table t1;
+
+create table t2 (
+a char character set latin1 collate latin1_swedish_ci,
+b char character set latin1 collate latin1_bin);
+--error 1271
+create table t1 as
+(select a from t2) union
+(select b from t2);
+create table t1 as
+(select a collate latin1_german1_ci from t2) union
+(select b from t2);
+show create table t1;
+drop table t1;
+create table t1 as
+(select a from t2) union
+(select b collate latin1_german1_ci from t2);
+show create table t1;
+drop table t1;
+drop table t2;
+

sql/ha_innodb.cc

@@ -429,6 +429,36 @@ innobase_mysql_print_thd(
 	putc('\n', f);
 }
 
+/**********************************************************************
+Compares NUL-terminated UTF-8 strings case insensitively.
+
+NOTE that the exact prototype of this function has to be in
+/innobase/dict/dict0dict.c! */
+extern "C"
+int
+innobase_strcasecmp(
+/*================*/
+				/* out: 0 if a=b, <0 if a<b, >1 if a>b */
+	const char*	a,	/* in: first string to compare */
+	const char*	b)	/* in: second string to compare */
+{
+	return(my_strcasecmp(system_charset_info, a, b));
+}
+
+/**********************************************************************
+Makes all characters in a NUL-terminated UTF-8 string lower case.
+
+NOTE that the exact prototype of this function has to be in
+/innobase/dict/dict0dict.c! */
+extern "C"
+void
+innobase_casedn_str(
+/*================*/
+	char*	a)	/* in/out: string to put in lower case */
+{
+	my_casedn_str(system_charset_info, a);
+}
+
 /*************************************************************************
 Creates a temporary file. */
 extern "C"
@@ -687,14 +717,7 @@ innobase_query_caching_of_table_permitted(
 					    separator between db and table */
 	norm_name[full_name_len] = '\0';
 #ifdef __WIN__
-	/* Put to lower case */
-
-	char*	ptr = norm_name;
-
-	while (*ptr != '\0') {
-	        *ptr = tolower(*ptr);
-	        ptr++;
-	}
+	innobase_casedn_str(norm_name);
 #endif
 	/* The call of row_search_.. will start a new transaction if it is
 	not yet started */
@@ -1539,14 +1562,7 @@ normalize_table_name(
 	norm_name[name_ptr - db_ptr - 1] = '/';
 
 #ifdef __WIN__
-	/* Put to lower case */
-
-	ptr = norm_name;
-
-	while (*ptr != '\0') {
-	        *ptr = tolower(*ptr);
-	        ptr++;
-	}
+	innobase_casedn_str(norm_name);
 #endif
 }
 
@@ -3559,9 +3575,9 @@ create_index(
 
 			field = form->field[j];
 
-			if (0 == ut_cmp_in_lower_case(
-					(char*)field->field_name,
-					(char*)key_part->field->field_name)) {
+			if (0 == innobase_strcasecmp(
+					field->field_name,
+					key_part->field->field_name)) {
 				/* Found the corresponding column */
 
 				break;
@@ -4003,7 +4019,7 @@ innobase_drop_database(
 	namebuf[len] = '/';
 	namebuf[len + 1] = '\0';
 #ifdef  __WIN__
-	my_casedn_str(system_charset_info, namebuf);
+	innobase_casedn_str(namebuf);
 #endif
 	trx = trx_allocate_for_mysql();
 	trx->mysql_thd = current_thd;

sql/item.cc

@@ -2558,8 +2558,8 @@ bool Item_type_holder::join_types(THD *thd, Item *item)
   if (use_new_field || use_expression_type ||
       (new_result_type != item_type) || (new_length > max_length) ||
       (!maybe_null && item->maybe_null) ||
-      (item_type == STRING_RESULT &&
-       !my_charset_same(collation.collation, item->collation.collation)))
+      (item_type == STRING_RESULT && 
+       collation.collation != item->collation.collation))
   {
     if (use_expression_type || item->type() != Item::FIELD_ITEM)
       field_example= 0;

sql/sql_union.cc

@@ -264,9 +264,27 @@ int st_select_lex_unit::prepare(THD *thd_arg, select_result *sel_result,
     }
   }
 
-  // it is not single select
   if (first_select->next_select())
   {
+
+  // it is not single select
+
+    /*
+      Check that it was possible to aggregate
+      all collations together for UNION.
+    */
+    List_iterator_fast<Item> tp(types);
+    Item *type;
+    while ((type= tp++))
+    {
+      if (type->result_type() == STRING_RESULT &&
+          type->collation.derivation == DERIVATION_NONE)
+      {
+        my_error(ER_CANT_AGGREGATE_NCOLLATIONS, MYF(0), "UNION");
+        goto err;
+      }
+    }
+
     union_result->tmp_table_param.field_count= types.elements;
     if (!(table= create_tmp_table(thd_arg,
 				  &union_result->tmp_table_param, types,

strings/CHARSET_INFO.txt

@@ -58,8 +58,7 @@ using its non-default counterpart for the given character set.
 
 binary_numner - ID of a charset+collation pair, which consists
 of the same character set and the binary collation of this
-character set. Not really used now. Intended to optimize
-"SELECT BINARY x" in the future.
+character set. Not really used now. 
 
 Names
 -----
@@ -121,7 +120,7 @@ Misc fields
                      e.g. Letter 'A' with two dots above is
                      substituted with 'AE'. 
   mbminlen         - mininum multibyte sequence length.
-                     Now always 1 accept ucs2. For ucs2
+                     Now always 1 except ucs2. For ucs2
                      it is 2.
   mbmaxlen         - maximum multibyte sequence length.
                      1 for 8bit charsets. Can be also 2 or 3.