Fixed a bug with SELECT DISTINCT and HAVING (91c4407e) · Commits · Software / OSDI20 Artifacts / mariadb

Docs/manual.texi

+61 −15

Original line number	Diff line number	Diff line
		@@ -5939,12 +5939,15 @@ A reasonable @code{tar} to unpack the distribution. GNU @code{tar} is
		known to work. Sun @code{tar} is known to have problems.

		@item
		A working ANSI C++ compiler. @code{gcc} >= 2.8.1, @code{egcs} >=
		1.0.2, SGI C++, and SunPro C++ are some of the compilers that are known to
		work. @code{libg++} is not needed when using @code{gcc}. @code{gcc}
		2.7.x has a bug that makes it impossible to compile some perfectly legal
		C++ files, such as @file{sql/sql_base.cc}. If you only have @code{gcc} 2.7.x,
		you must upgrade your @code{gcc} to be able to compile @strong{MySQL}.
		A working ANSI C++ compiler. @code{gcc} >= 2.95.2, @code{egcs} >= 1.0.2
		or @code{egcs 2.91.66}, SGI C++, and SunPro C++ are some of the
		compilers that are known to work. @code{libg++} is not needed when
		using @code{gcc}. @code{gcc} 2.7.x has a bug that makes it impossible
		to compile some perfectly legal C++ files, such as
		@file{sql/sql_base.cc}. If you only have @code{gcc} 2.7.x, you must
		upgrade your @code{gcc} to be able to compile @strong{MySQL}. @code{gcc}
		2.8.1 is also known to have problems on some platforms so it should be
		avoided if there exists a new compiler for the platform..

		@code{gcc} >= 2.95.2 is recommended when compiling @strong{MySQL}
		Version 3.23.x.
		@@ -8536,8 +8539,8 @@ We recommend the following @code{configure} line with @code{egcs} and
		@code{gcc 2.95} on AIX:

		@example
		CC="gcc -pipe -mcpu=power2 -Wa,-many" \
		CXX="gcc -pipe -mcpu=power2 -Wa,-many" \
		CC="gcc -pipe -mcpu=power -Wa,-many" \
		CXX="gcc -pipe -mcpu=power -Wa,-many" \
		CXXFLAGS="-felide-constructors -fno-exceptions -fno-rtti" \
		./configure --prefix=/usr/local/mysql --with-low-memory
		@end example
		@@ -8549,6 +8552,21 @@ available. We don't know if the @code{-fno-exceptions} is required with
		option generates faster code, we recommend that you should always use this
		option with @code{egcs / gcc}.

		If you get a problem with assembler code try changing the -mcpu=xxx to
		match your cpu. Typically power2, power, or powerpc may need to be used,
		alternatively you might need to use 604 or 604e. I'm not positive but I
		would think using "power" would likely be safe most of the time, even on
		a power2 machine.

		If you don't know what your cpu is then do a "uname -m", this will give
		you back a string that looks like "000514676700", with a format of
		xxyyyyyymmss where xx and ss are always 0's, yyyyyy is a unique system
		id and mm is the id of the CPU Planar. A chart of these values can be
		found at
		@uref{http://www.rs6000.ibm.com/doc_link/en_US/a_doc_lib/cmds/aixcmds5/uname.htm}.
		This will give you a machine type and a machine model you can use to
		determine what type of cpu you have.

		If you have problems with signals (@strong{MySQL} dies unexpectedly
		under high load) you may have found an OS bug with threads and
		signals. In this case you can tell @strong{MySQL} not to use signals by
		@@ -8569,6 +8587,29 @@ On some versions of AIX, linking with @code{libbind.a} makes
		@code{getservbyname} core dump. This is an AIX bug and should be reported
		to IBM.

		For AIX 4.2.1 and gcc you have to do the following changes.

		After configuring, edit @file{config.h} and @file{include/my_config.h}
		and change the line that says

		@example
		#define HAVE_SNPRINTF 1
		@end example

		to

		@example
		#undef HAVE_SNPRINTF
		@end example

		And finally, in @file{mysqld.cc} you need to add a prototype for initgoups.

		@example
		#ifdef _AIX41
		extern "C" int initgroups(const char *,int);
		#endif
		@end example

		@node HP-UX 10.20, HP-UX 11.x, IBM-AIX, Source install system issues
		@subsection HP-UX Version 10.20 Notes

		@@ -23777,7 +23818,7 @@ is not signaled to the other servers.
		@section MERGE Tables

		@code{MERGE} tables are new in @strong{MySQL} Version 3.23.25. The code
		is still in beta, but should stabilize soon!
		is still in gamma, but should be resonable stable.

		A @code{MERGE} table is a collection of identical @code{MyISAM} tables
		that can be used as one. You can only @code{SELECT}, @code{DELETE}, and
		@@ -23790,8 +23831,8 @@ will only clear the mapping for the table, not delete everything in the
		mapped tables. (We plan to fix this in 4.0).

		With identical tables we mean that all tables are created with identical
		column information. You can't put a MERGE over tables where the columns
		are packed differently or doesn't have exactly the same columns.
		column and key information. You can't put a MERGE over tables where the
		columns are packed differently or doesn't have exactly the same columns.
		Some of the tables can however be compressed with @code{myisampack}.
		@xref{myisampack}.

		@@ -23826,8 +23867,10 @@ More efficient repairs. It's easier to repair the individual files that
		are mapped to a @code{MERGE} file than trying to repair a real big file.
		@item
		Instant mapping of many files as one. A @code{MERGE} table uses the
		index of the individual tables. It doesn't need an index of its one.
		This makes @code{MERGE} table collections VERY fast to make or remap.
		index of the individual tables. It doesn't need to maintain an index of
		its one. This makes @code{MERGE} table collections VERY fast to make or
		remap. Note that you must specify the key definitions when you create
		a @code{MERGE} table!.
		@item
		If you have a set of tables that you join to a big table on demand or
		batch, you should instead create a @code{MERGE} table on them on demand.
		@@ -43032,8 +43075,8 @@ An open source client for exploring databases and executing SQL. Supports
		A query tool for @strong{MySQL} and PostgreSQL.
		@item @uref{http://dbman.linux.cz/,dbMan}
		A query tool written in Perl. Uses DBI and Tk.
		@item @uref{http://www.mysql.com/Downloads/Win32/Msc201.EXE, Mascon 2.1.15}
		@item @uref{http://www.mysql.com/Downloads/Win32/FrMsc201.EXE, Free Mascon 2.1.14}
		@item @uref{http://www.mysql.com/Downloads/Win32/Msc201.EXE, Mascon 202}
		@item @uref{http://www.mysql.com/Downloads/Win32/FrMsc202.EXE, Free Mascon 202}
		Mascon is a powerful Win32 GUI for the administering @strong{MySQL} server
		databases. Mascon's features include visual table design, connections to
		multiple servers, data and blob editing of tables, security setting, SQL
		@@ -44050,6 +44093,9 @@ not yet 100% confident in this code.
		@appendixsubsec Changes in release 3.23.38
		@itemize @bullet
		@item
		Fixed bug when too many rows where removed when using
		@code{SELECT DISTINCT ... HAVING}.
		@item
		@code{SHOW CREATE TABLE} now returns @code{TEMPORARY} for temporary tables.
		@item
		Added @code{Rows_examined} to slow query log.

sql/sql_select.cc

+83 −53

Original line number	Diff line number	Diff line
		@@ -36,7 +36,8 @@ const char *join_type_str[]={ "UNKNOWN","system","const","eq_ref","ref",

		static bool make_join_statistics(JOIN join,TABLE_LIST tables,COND *conds,
		DYNAMIC_ARRAY *keyuse,List<Item_func_match> &ftfuncs);
		static bool update_ref_and_keys(DYNAMIC_ARRAY keyuse,JOIN_TAB join_tab,
		static bool update_ref_and_keys(THD thd, DYNAMIC_ARRAY keyuse,
		JOIN_TAB *join_tab,
		uint tables,COND *conds,table_map table_map,
		List<Item_func_match> &ftfuncs);
		static int sort_keyuse(KEYUSE a,KEYUSE b);
		@@ -106,12 +107,14 @@ static uint find_shortest_key(TABLE *table, key_map usable_keys);
		static bool test_if_skip_sort_order(JOIN_TAB tab,ORDER order,
		ha_rows select_limit);
		static int create_sort_index(JOIN_TAB tab,ORDER order,ha_rows select_limit);
		static int remove_duplicates(JOIN join,TABLE entry,List<Item> &fields);
		static bool fix_having(JOIN join, Item *having);
		static int remove_duplicates(JOIN join,TABLE entry,List<Item> &fields,
		Item *having);
		static int remove_dup_with_compare(THD thd, TABLE entry, Field **field,
		ulong offset);
		ulong offset,Item *having);
		static int remove_dup_with_hash_index(THD thd, TABLE table,
		uint field_count, Field **first_field,
		ulong key_length);
		ulong key_length,Item *having);
		static int join_init_cache(THD thd,JOIN_TAB tables,uint table_count);
		static ulong used_blob_length(CACHE_FIELD **ptr);
		static bool store_record_in_cache(JOIN_CACHE *cache);
		@@ -717,8 +720,11 @@ mysql_select(THD thd,TABLE_LIST tables,List<Item> &fields,COND *conds,
		if (select_distinct && ! group)
		{
		thd->proc_info="Removing duplicates";
		if (remove_duplicates(&join,tmp_table,fields))
		if (having)
		having->update_used_tables();
		if (remove_duplicates(&join,tmp_table,fields, having))
		goto err; /* purecov: inspected */
		having=0;
		select_distinct=0;
		}
		tmp_table->reginfo.lock_type=TL_UNLOCK;
		@@ -749,28 +755,8 @@ mysql_select(THD thd,TABLE_LIST tables,List<Item> &fields,COND *conds,
		/* If we have already done the group, add HAVING to sorted table */
		if (having && ! group && ! join.sort_and_group)
		{
		having->update_used_tables(); // Some tables may have been const
		JOIN_TAB *table=&join.join_tab[join.const_tables];
		table_map used_tables= join.const_table_map \| table->table->map;

		Item* sort_table_cond=make_cond_for_table(having,used_tables,used_tables);
		if (sort_table_cond)
		{
		if (!table->select)
		if (!(table->select=new SQL_SELECT))
		if (fix_having(&join,&having))
		goto err;
		if (!table->select->cond)
		table->select->cond=sort_table_cond;
		else // This should never happen
		if (!(table->select->cond=new Item_cond_and(table->select->cond,
		sort_table_cond)))
		goto err;
		table->select_cond=table->select->cond;
		DBUG_EXECUTE("where",print_where(table->select->cond,
		"select and having"););
		having=make_cond_for_table(having,~ (table_map) 0,~used_tables);
		DBUG_EXECUTE("where",print_where(conds,"having after sort"););
		}
		}
		if (create_sort_index(&join.join_tab[join.const_tables],
		group ? group : order,
		@@ -941,7 +927,7 @@ make_join_statistics(JOIN join,TABLE_LIST tables,COND *conds,
		}

		if (conds \|\| outer_join)
		if (update_ref_and_keys(keyuse_array,stat,join->tables,
		if (update_ref_and_keys(join->thd,keyuse_array,stat,join->tables,
		conds,~outer_join,ftfuncs))
		DBUG_RETURN(1);

		@@ -1442,8 +1428,9 @@ sort_keyuse(KEYUSE a,KEYUSE b)
		*/

		static bool
		update_ref_and_keys(DYNAMIC_ARRAY keyuse,JOIN_TAB join_tab,uint tables,
		COND *cond, table_map normal_tables,List<Item_func_match> &ftfuncs)
		update_ref_and_keys(THD thd, DYNAMIC_ARRAY keyuse,JOIN_TAB *join_tab,
		uint tables, COND *cond, table_map normal_tables,
		List<Item_func_match> &ftfuncs)
		{
		uint and_level,i,found_eq_constant;

		@@ -1451,8 +1438,7 @@ update_ref_and_keys(DYNAMIC_ARRAY keyuse,JOIN_TAB join_tab,uint tables,
		KEY_FIELD key_fields,end;

		if (!(key_fields=(KEY_FIELD*)
		my_malloc(sizeof(key_fields[0])*
		(current_thd->cond_count+1)*2,MYF(0))))
		thd->alloc((sizeof(key_fields[0])thd->cond_count+1)2)))
		return TRUE; /* purecov: inspected */
		and_level=0; end=key_fields;
		if (cond)
		@@ -1466,14 +1452,10 @@ update_ref_and_keys(DYNAMIC_ARRAY keyuse,JOIN_TAB join_tab,uint tables,
		}
		}
		if (init_dynamic_array(keyuse,sizeof(KEYUSE),20,64))
		{
		my_free((gptr) key_fields,MYF(0));
		return TRUE;
		}
		/* fill keyuse with found key parts */
		for (KEY_FIELD *field=key_fields ; field != end ; field++)
		add_key_part(keyuse,field);
		my_free((gptr) key_fields,MYF(0));
		}

		if (ftfuncs.elements)
		@@ -1894,7 +1876,7 @@ cache_record_length(JOIN *join,uint idx)
		{
		uint length;
		JOIN_TAB pos,end;
		THD *thd=current_thd;
		THD *thd=join->thd;

		length=0;
		for (pos=join->best_ref+join->const_tables,end=join->best_ref+idx ;
		@@ -2076,7 +2058,7 @@ get_best_combination(JOIN *join)
		}
		else
		{
		THD *thd=current_thd;
		THD *thd=join->thd;
		for (i=0 ; i < keyparts ; keyuse++,i++)
		{
		while (keyuse->keypart != i \|\|
		@@ -4433,7 +4415,8 @@ join_init_read_record(JOIN_TAB *tab)
		{
		if (tab->select && tab->select->quick)
		tab->select->quick->reset();
		init_read_record(&tab->read_record,current_thd, tab->table, tab->select,1,1);
		init_read_record(&tab->read_record, tab->join->thd, tab->table,
		tab->select,1,1);
		return (*tab->read_record.read_record)(&tab->read_record);
		}

		@@ -5265,6 +5248,38 @@ create_sort_index(JOIN_TAB tab,ORDER order,ha_rows select_limit)
		}


		/*
		** Add the HAVING criteria to table->select
		*/

		static bool fix_having(JOIN join, Item *having)
		{
		(*having)->update_used_tables(); // Some tables may have been const
		JOIN_TAB *table=&join->join_tab[join->const_tables];
		table_map used_tables= join->const_table_map \| table->table->map;

		Item* sort_table_cond=make_cond_for_table(*having,used_tables,used_tables);
		if (sort_table_cond)
		{
		if (!table->select)
		if (!(table->select=new SQL_SELECT))
		return 1;
		if (!table->select->cond)
		table->select->cond=sort_table_cond;
		else // This should never happen
		if (!(table->select->cond=new Item_cond_and(table->select->cond,
		sort_table_cond)))
		return 1;
		table->select_cond=table->select->cond;
		DBUG_EXECUTE("where",print_where(table->select_cond,
		"select and having"););
		having=make_cond_for_table(having,~ (table_map) 0,~used_tables);
		DBUG_EXECUTE("where",print_where(*having,"having after make_cond"););
		}
		return 0;
		}


		/*****************************************************************************
		** Remove duplicates from tmp table
		** This should be recoded to add a uniuqe index to the table and remove
		@@ -5305,7 +5320,7 @@ static void free_blobs(Field **ptr)


		static int
		remove_duplicates(JOIN join, TABLE entry,List<Item> &fields)
		remove_duplicates(JOIN join, TABLE entry,List<Item> &fields, Item *having)
		{
		int error;
		ulong reclength,offset;
		@@ -5342,9 +5357,10 @@ remove_duplicates(JOIN join, TABLE entry,List<Item> &fields)
		sortbuff_size)))
		error=remove_dup_with_hash_index(join->thd, entry,
		field_count, first_field,
		reclength);
		reclength, having);
		else
		error=remove_dup_with_compare(join->thd, entry, first_field, offset);
		error=remove_dup_with_compare(join->thd, entry, first_field, offset,
		having);

		free_blobs(first_field);
		DBUG_RETURN(error);
		@@ -5352,19 +5368,19 @@ remove_duplicates(JOIN join, TABLE entry,List<Item> &fields)


		static int remove_dup_with_compare(THD thd, TABLE table, Field **first_field,
		ulong offset)
		ulong offset, Item *having)
		{
		handler *file=table->file;
		char org_record,new_record;
		char org_record,new_record, *record;
		int error;
		ulong reclength=table->reclength-offset;
		DBUG_ENTER("remove_dup_with_compare");

		org_record=(char*) table->record[0]+offset;
		org_record=(char*) (record=table->record[0])+offset;
		new_record=(char*) table->record[1]+offset;

		file->rnd_init();
		error=file->rnd_next(table->record[0]);
		error=file->rnd_next(record);
		for (;;)
		{
		if (thd->killed)
		@@ -5381,6 +5397,12 @@ static int remove_dup_with_compare(THD thd, TABLE table, Field **first_field,
		break;
		goto err;
		}
		if (having && !having->val_int())
		{
		if ((error=file->delete_row(record)))
		goto err;
		continue;
		}
		if (copy_blobs(first_field))
		{
		my_error(ER_OUT_OF_SORTMEMORY,MYF(0));
		@@ -5393,7 +5415,7 @@ static int remove_dup_with_compare(THD thd, TABLE table, Field **first_field,
		bool found=0;
		for (;;)
		{
		if ((error=file->rnd_next(table->record[0])))
		if ((error=file->rnd_next(record)))
		{
		if (error == HA_ERR_RECORD_DELETED)
		continue;
		@@ -5403,19 +5425,19 @@ static int remove_dup_with_compare(THD thd, TABLE table, Field **first_field,
		}
		if (compare_record(table, first_field) == 0)
		{
		if ((error=file->delete_row(table->record[0])))
		if ((error=file->delete_row(record)))
		goto err;
		}
		else if (!found)
		{
		found=1;
		file->position(table->record[0]); // Remember position
		file->position(record); // Remember position
		}
		}
		if (!found)
		break; // End of file
		/* Restart search on next row */
		error=file->restart_rnd_next(table->record[0],file->ref);
		error=file->restart_rnd_next(record,file->ref);
		}

		file->extra(HA_EXTRA_NO_CACHE);
		@@ -5436,7 +5458,8 @@ static int remove_dup_with_compare(THD thd, TABLE table, Field **first_field,
		static int remove_dup_with_hash_index(THD thd, TABLE table,
		uint field_count,
		Field **first_field,
		ulong key_length)
		ulong key_length,
		Item *having)
		{
		byte key_buffer, key_pos, *record=table->record[0];
		int error;
		@@ -5484,6 +5507,12 @@ static int remove_dup_with_hash_index(THD thd, TABLE table,
		break;
		goto err;
		}
		if (having && !having->val_int())
		{
		if ((error=file->delete_row(record)))
		goto err;
		continue;
		}

		/* copy fields to key buffer */
		field_length=field_lengths;
		@@ -5499,6 +5528,7 @@ static int remove_dup_with_hash_index(THD thd, TABLE table,
		if ((error=file->delete_row(record)))
		goto err;
		}
		else
		(void) hash_insert(&hash, key_pos-key_length);
		key_pos+=extra_length;
		}

support-files/Makefile.am

+0 −7

Original line number	Diff line number	Diff line
		@@ -18,7 +18,6 @@
		## Process this file with automake to create Makefile.in

		EXTRA_DIST = mysql.spec.sh \
		mysql-max.spec.sh \
		my-small.cnf.sh \
		my-medium.cnf.sh \
		my-large.cnf.sh \
		@@ -34,7 +33,6 @@ pkgdata_DATA = my-small.cnf \
		my-huge.cnf \
		mysql-log-rotate \
		mysql-@VERSION@.spec \
		mysql-max-@VERSION@.spec \
		binary-configure

		pkgdata_SCRIPTS = mysql.server
		@@ -44,7 +42,6 @@ CLEANFILES = my-small.cnf \
		my-large.cnf \
		my-huge.cnf \
		mysql.spec \
		mysql-max-@VERSION@.spec \
		mysql-@VERSION@.spec \
		mysql-log-rotate \
		mysql.server \
		@@ -55,10 +52,6 @@ mysql-@VERSION@.spec: mysql.spec
		rm -f $@
		cp mysql.spec $@

		mysql-max-@VERSION@.spec: mysql-max.spec
		rm -f $@
		cp mysql-max.spec $@

		SUFFIXES = .sh

		.sh: