in short we now record whenever the slave I/O thread ignores a master's event because of its server id,
and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we
do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored*
executed one, which may not be the last of the master's binlog (and so the slave *looks* behind
the master though it's data-wise it's not).
Bug #10308: Parse 'purge master logs' with subselect correctly.
subselect.test:
Bug #10308: Test for 'purge master logs' with subselect.
subselect.result:
Bug #10308: Test result for 'purge master logs' with subselect.
When fixing Item_func_plus in ORDER BY clause field c is searched in all
opened tables, but because c is an alias it wasn't found there.
This patch adds a flag to select_lex which allows Item_field::fix_fields()
to look up in select's item_list to find aliased fields.
After SHOW TABLE STATUS last_insert_id wasn't cleaned, and next select
erroneously rewrites WHERE condition and returs a row;
5.0 isn't affected because of different SHOW TABLE STATUS handling.
last_insert_id cleanup added to mysqld_extend_show_tables().
"CHARACTER SET", "COLLATE", and "DEFAULT" are always
printed(excepting MODE_MYSQL323 and MODE_MYSQL40)
"AUTO_INCREMENT", "ON UPDATE CURRENT_TIMESTAMP" are printed only
if NO_FIELD_OPTIONS is not set.
resolve_const_item() assumed to be not called for Item_row items. For
ensuring that DBUG_ASSERT(0) was set there.
This patch adds section for Item_row items. If it can it recursively calls
resolve_const_item() for each item the Item_row contains. If any of the
contained items is null then whole Item_row substitued by Item_null. Otherwise
it just returns.
The problem was an ab-use of last_rkey_length.
Formerly we saved the packed key length (of the search key)
in this element. But in certain cases it got replaced by
the (packed) result key length.
Now we use a new element of MI_INFO to save the packed key
length of the search key.
set auto_increment_field_not_null to true to preserve zero values
for autoincrement fields during LOAD DATA execution if
NO_AUTO_VALUE_ON_ZERO sql mode is set
thd->allow_sum_func was left 'true' after previous statement thus allowing
sum funcs to be present in conditions.
thd->allow_sum_func should be set to 0 for each query and each prepared
statement reinitialization. This is done in lex_start() and
reset_stmt_for_execute().
index doesn't return correct result
item_cmpfunc.cc:
Use charset of LIKE to decide whether
to use 8bit or Unicode "escape" value.
But use charset of "escape" to scan escape character.
strings/ctype-xxx.c:
We cannot reduce "end" pointer using charpos(),
because of possible escape characters in the string.
Limit the loop using count of written characters instead.
ctype_like_escape.inc:
new file
mysql-test/t/ctype_xxx:
mysql-test/r/ctype_xxx:
Adding test case.
The problem was in that the MIN/MAX optimization in opt_sum_query was
replacing MIN/MAX functions with their constant argument without
taking into account that a query has no result rows.
ctype_latin1.test, ctype_latin1.result:
adding test case
ctype-latin1.c:
Fixing ctype array to treat extended cp1252
letters as valid identifiers on server side,
and as valid "isprint" characters (e.g. on client side).
For queries with GROUP BY and without hidden GROUP BY fields DISTINCT is
optimized away becuase such queries produce result set without duplicates.
But ROLLUP can add rows which may be same to some rows and this fact was
ignored.
Added check so if ROLLUP is present DISTINCT can't be optimized away.
present): the problem originally was that the tables in auxilliary_tables did not have
the correct real_name, which caused problems in the second call to tables_ok().
The fix corrects the real_name problem, and also sets the updating flag properly,
which makes the second call to tables_ok() unnecessary.
produce warning for 'create database if not exists' if database exists
do not update database options in this case
produce warning for 'create table if not exists' if table exists
This fix is cancellation of ChangeSet
1.2329 05/07/12 08:35:30 reggie@linux.site +8 -0
Bug 7142 Show Fields from fails using Borland's dbExpress interface
The reason is we can't fix bug#7142 without
breaking of existing applications/APIs that worked fine with earlier 4.1
bug 7142 is fixed in 5.0
Fixed bug #12885.
Forced inheritence of the maybe_null flag for the expressions
containing GROUP BY attributes in selects with ROLLUP.
olap.test, olap.result:
Added test case for bug #12885.
Added test cases for bug #12863.
item_sum.cc, item_sum.h:
Fixed bug #12863.
Added a flag to Item_func_group_concat set to FALSE after
concatenation of the first element of a group.
Fixed bugs #12101, #12102: wrong calculation of not_null_tables()
for some expressions.
The classes Item_func_between, Item_func_if, Item_func_in are modified.
Item_func_between/in objects can represent now [NOT]BETWEEN/IN expressions.
The class Item_func_opt_neg is added to factor out the functionality
common for the modified classes Item_func_between and Item_func_in.
item_cmpfunc.cc:
Fixed bugs #12101, #12102: wrong calculation of not_null_tables()
for some expressions.
Added Item_func_between::fix_fields(), Item_func_if::fix_fields(),
Item_func_in::fix_fields(). They correct generic calculation of
the not_null_tables attribute when it is needed.
Modified Item_func_between::val_int(), Item_func_in::val_int().
opt_range.cc:
Fixed bugs #12101, #12102: wrong calculation of not_null_tables()
for some expressions.
The function get_mm_tree() is modified. There cannot be NOT before
BETWEEN/IN anymore. Rather Item_func_between/in objects can represent
now [NOT]BETWEEN/IN expressions.
sql_yacc.yy:
Fixed bugs #12101, #12102: wrong calculation of not_null_tables()
for some expressions.
Item_func_between/in objects can represent now [NOT]BETWEEN/IN expresions.
join_outer.result:
Fixed some testcases results (bugs #12101, #12102)
join_outer.test:
Added testcases for bugs #12101, #12102
adding test case.
item_cmpfunc.cc:
Bug#12611
ESCAPE + LIKE do not work when the escape char is a multibyte one
Additional fix for 8bit character sets:
escape character must be converted into
operation character set.
multi-threaded environment".
To avoid deadlocks between several simultaneously run account management
commands (particularly between FLUSH PRIVILEGES/SET PASSWORD and GRANT
commands) we should always take table and internal locks during their
execution in the same order. In other words we should first open and lock
privilege tables and only then obtain acl_cache::lock/LOCK_grant locks.
Item_func_group_concat::fix_fields() set maybe_null flag to 0, and set it to
1 only if some of it's arguments may be null. When used in subquery in tmp
table created field which can't be null. When no data retireved result field
have to be set to null and error mentioned in bug report occurs. Also this
bug can occur if selecting from not null field in empty table.
Function group_concat now marked maybe_null from the very beginning not only
if some of it's argument may be null.
item_cmpfunc.cc:
Pass unicode value as "escape" argument to my_wildcmp
if a multibyte character set is used.
For single byte character set nothing has changed:
native (non-unicode) character code is still passed.
ctype_utf8.result, ctype_utf8.test:
adding test case
Item::tmp_table_field_from_field_type() and create_tmp_field_from_item()
was converting string field to blob depending on byte-wise length instead of
character length, which results in converting valid varchar string with
length == 86 to longtext.
Made that functions above take into account max width of character when
converting string fields to blobs.
character_set_results is nullable, but value_ptr returns string "NULL"
set_var.cc:
Create Item_null instead of Item_string for NULL values
variables.result, variables.test:
adding test case
Cannot convert the charset of a GROUP_CONCAT result:
item_sum.cc:
"result" character set was not set into proper value.
func_gconcat.result, func_gconcat.test:
Fixing tests accordingly.
Added test cases for bug #12625.
sql_select.cc:
Fixed bug #12625.
Fixed invalid removal of constant items from the DISTINCT
list in the function create_distinct_group.
We binlog the DROP TABLE for each table that was actually dropped. Per Sergei's
suggestion a fixed buffer for the DROP TABLE query is pre-allocated from THD pool, and
logging now is done in batches - new batch is started if the buffer becomes full.
Reduced memory usage by reusing the table list instead of accumulating a list of
dropped table names. Also fixed the problem if the table was not actually dropped, eg
due to permissions. Extended the test case to make sure batched query
logging does work.
Split table version into 2 (major, minor)
Impl. signaling to API when table has been altered
Allow running transactions to use any minor number for transactions
Fixed bug #11479.
The JOIN::reinit method cannot call setup_tables
after the optimization phase since this function
removes some optimization settings for joined
tables. E.g. it resets values of the null_row flag to 0.
subselect.result, subselect.test:
Added a test case for bug #11479.
Added a test case for bug #12392.
item_cmpfunc.cc:
Fixed bug #12392.
Missing handling of rows containing NULL components
when evaluating IN predicates caused a crash.
- Fixed some error condtion when handling dates with 'T'
- Added extra test for bug #11867 (Wrong result with "... WHERE ROW( a, b ) IN ( SELECT DISTINCT a, b WHERE ...)" to show it's not yet fixed
- Safety fixes and cleanups
BUG #11104
Took out the offset-=delimiter_length-1 out of the for loop. It was causing
basically this:
select substring_index('the king of the the hill', 'the', -2) to not work.
The first iteration, offset would be initialised to 24, then strstr would
point at 'the king of the the* hill' ('*'means right before the
character following), returning a offset of 16. The for loop would then
decrement offset by two (3 - 1), to 14, now pointing at
"the king of th*e the hill", _skipping_ past the 'e' in the second to last
'the', and therefore strstr would never have a chance of matching the
second to last 'the', then moving on to the 'the' at the begginning of the
string!
In a nutshell, offset was being decremented by too great a value, preventing
the second to last 'the' from being ever found, hence the result of
'king of the the hill' from the query that is reported in the bug report
func_str.test:
BUG #11104
Added tests to make sure fix addresses issues in original bug report
func_str.result:
BUG #11104
New results for new tests
adding test case
item_strfunc.cc:
Bug#12351
CONCAT with USER()/DATEBASE() and
a column gets strange results.
Mark created Item_str as constant, so CONCAT
cannot reuse it for optimization purposes.
Changed assembler functions to not access global variables or variables in text segement
Added wrapper function in C to longlong2str() to pass _dig_vec_upper as an argument
Added test case for bug #11867.
Fixed results for two existing test cases.
subselect.test:
Added test case for bug #11867.
item_subselect.cc:
Fixed bug #11867.
Added missing code in Item_in_subselect::row_value_transformer
that caused problems for queries with
ROW(elems) IN (SELECT DISTINCT cols FROM ...).
When creating temporary table for UNION, pass TMP_TABLE_FORCE_MYISAM flag to
create_tmp_table if we will be using fulltext function(s) when reading from the
temp. table.
item.cc:
item.h:
Adding Item_param::safe_charset_converter,
not to return collation mix error if
parameter can be converted into operation
character set.
ctype_utf8.result:
adding test case
ctype_utf8.test:
adding test case
length.
When temporary field created for DATE(LEFT(column,8)) expression, max_length
value is taken from Item_date_typecast, and it is getting it from underlaid
Item_func_left and it's max_length is 8 in given expression. And all this
results in stripping last 2 digits.
To Item_date_typecast class added its own fix_length_and_dec() function
that sets max_length value to 10, which is proper for DATE field.
fixing tests accordingly
item.cc:
Bug #10892 user variables not auto cast for comparisons
When mixing strings with different character sets,
and coercibility is the same, we allow conversion
if one character set is superset for other character set.
Added a test case for bug #12095.
sql_class.h:
Fixed bug #12095: a join query with GROUP_CONCAT over a single row table.
Added a flag to the TMP_TABLE_PARAM class forcing to put constant
items generated after elimination of a single row table into temp table
in some cases (e.g. when GROUP_CONCAT is calculated over a single row
table).
bk ci sql/item_sum.cc
Fixed bug #12095: a join query with GROUP_CONCAT over a single row table.
If GROUP_CONCAT is calculated we always put its argument into a temp
table, even when the argument is a constant item.
sql_select.cc:
Fixed bug #12095: a join query with GROUP_CONCAT over one row table.
If temp table is used to calculate GROUP_CONCAT the argument should
be always put into this table, even when it is a constant item.
Fixed bug #12144.
Added an optimization that avoids key access with null keys for the 'ref'
method when used in outer joins. The regilar optimization with adding
IS NOT NULL expressions is not applied for outer join on expressions as
the predicates of these expressions are not pushed down in 4.1.
null_key.result, null_key.test:
Added a test case for bug #12144.
Adding test
item_sum.cc:
Adding a call for collation/charset aggregation,
to collect attributes from the arguments. The actual bug fix.
item_func.h, item_func.cc, item.h, item.cc:
- Removing collation aggrgation functions from Item_func class
in item.cc, and adding it as non-class functions in item.cc
to be able to reuse this code for group_concat.
- Adding replacement for these functions into Item_func class
as wrappers for moved functions, to minizize patch size,
Bug #11987
mysql will truncate the text when the text contain GBK char:"0xA3A0" and "0xA1"
Allow to store and retrieve even unassigned GBK codes.
Like we did in Big5 earlier.
have_gbk.inc, have_gbk.require, ctype_gbk.result, ctype_gbk.test:
new file
Index.xml:
Fixing latin1 comment:
it is actually cp1252, not iso-8859-1
ctype_latin1.result:
changeing test results accordingly.
ctype-latin1.c:
Fixed to- and from-Unicode conversion maps
for better Unicode round trip of undefined
characters.
LOCK_thread_count)" and for BUG#12004 "SHOW BINARY LOGS reports 0
for the size of all binlogs but the current one".
There are a lot of 4.1->5.0 unmerged changes (hardest are in the optimizer), can't merge; still pushing in 4.1 because my changes
are very small. Feel free to ask me if you have problems merging them.
Added a test case for bug #11745.
sql_select.cc:
Fixed bug # 11745.
Added support of where clause for queries with FROM DUAL.
sql_yacc.yy:
Fixed bug # 11745.
Added optional where clause for queries with FROM DUAL.
"Negative integer keys incorrectly substituted for 0 during range analysis."
The problem is that the range optimizer incorrectly replaces any negative
constant with '0' for all types except BIGINT because the method save_in_field()
casts negative integers to non-negative. This causes incorrect query
results where (0 = any_negative_number).
The problem caused by this bug is fixed by the patch for BUG#11185.
That patch constitutes an optimization due to which the problem code is
never called with negative constants. This patch adds a test so we are sure
that the problem does not reappear.
disabled if ref is built with a key from the updated table
Problem was in add_not_null_conds() optimization function.
It contains following code:
JOIN_TAB *referred_tab= not_null_item->field->table->reginfo.join_tab;
...
add_cond_and_fix(&referred_tab->select_cond, notnull);
For UPDATE described in bug report referred_tab is 0 and dereferencing it
crashes the server.
of system vars at PREPARE time": implement a special Item
to handle system variables. This item substitutes itself with
a basic constant containing variable value at fix_fields.
than in previous 4.1.x
Wrongly applied optimization were adding NOT NULL constraint which results in
rejecting valid rows and reduced result set.
The problem was that add_notnull_conds() while checking subquery were adding
NOT NULL constraint to left joined table, to which, normally, optimization
don't have to be applied.
- Fixed problem, only detect comment if the # is on start of line AND starting line of the current command.
- Wrote tests for most of the mysqltest commands, added stricter checking of correct syntax.
No separate typecode for MEDIUMTEXT/LONGTEXT is added, as we
have no sound decision yet what typecodes and for what types are
sent by the server (aka what constitutes a distinct type in MySQL).
data": remove the fix for another bug (8807) that
added OUTER_REF_TABLE_BIT to all subqueries that used a placeholder
to prevent their evaluation at prepare. As this bit hanged in
Item_subselect::used_tables_cache for ever, a constant subquery with
a placeholder was never evaluated as such, which caused wrong
choice of the execution plan for the statement.
- to fix Bug#8807 backport a better fix from 5.0
- post-review fixes.
adding test case
sql_table.cc:
sql_table.cc:
- do not create a new item when charsets are the same
- return ER_INVALID_DEFAULT if default value cannot
be converted into the column character set.
item.cc:
- Allow conversion not only to Unicode,
but also to and from "binary".
- Adding safe_charset_converter() for Item_num
and Item_varbinary, returning a fixed const Item.
The problem here is that columns that have an especially long type
such as an enum type with many options would be longer than 40 chars
but the type column returned from show columns always was defined
as varchar(40).
This is fixed in 5.0 using info schema.
Bug#11591
CHAR column with utf8 does not work properly
(more chars than expected)
do_cut_string didn't call well_formed_length,
and copied all data, which was wrong in the
case of multibyte character set.
ctype_utf8.result, ctype_utf8.test:
adding test case
New BitKeeper file ``mysql-test/include/ctype_innodb_like.inc''
Many files:
bug#11650: LIKE pattern matching using prefix index doesn't return correct result
min and max values were too long in the case of prefix key.
Fix my_like_range functions not to exceed prefix length.
ctype_innodb_like.inc:
new file
Added a test case for bug #11543.
sql_select.cc:
Fixed bug #11543.
A ROLLUP query could return a wrong result set when
its GROUP BY clause contained references to the same
column.
(IN() remove NULL rows only for tables from first argument (value which we looking for in IN() list) but not for tables from IN() list)
Also it will be better change Item::not_null_tables() to prohibit this optimisation by default for new created items in 5.0 or 5.1.
Added a test case for bug #11414.
sql_select.cc:
Fixed bug #11414: crash on Windows with some simple
GROUP BY queries.
It happened to an allocation of an array containing
0 Copy_field elements in setup_copy_fields.
The bug had been already fixed in 5.0.
Fixing tests accordingly.
ctype-ucs2.c:
The same fix for UCS2.
ctype-utf8.c:
Bug #9557
MyISAM utf8 table crash
The problem was that my_strnncollsp_xxx could
return big value in the range 0..0xffff.
for some constant pairs it could return 32738,
which is defined as MI_FOUND_WRONG_KEY in
myisamdef.h. As a result, table considered to
be crashed.
Fix to return -1,0 or 1.
#9728 'Decreased functionality in "on duplicate key update
#8147 'a column proclaimed ambigous in INSERT ... SELECT .. ON DUPLICATE'
This ensures fields are uniquely qualified and also that one can't update other tables in the ON DUPLICATE KEY UPDATE part
Moved the key statistics update to info().
The table is not locked in open(). This made wrong stats possible.
No test case for the test suite.
This happens only with heavy concurrency.
A test script is added to the bug report.
Added a test case for bug #10124.
sql_select.h, item_subselect.cc, sql_select.cc:
Fixed bug #10124.
The copy method of the store_key classes can return
STORE_KEY_OK=0, STORE_KEY_FATAL=1, STORE_KEY_CONV=2 now.
field.cc:
Fixed bug #10124.
When ussuing a warning the store methods return 2 instead of 1 now.
The source of the problem is in Field_longlong::cmp. If 'this' is
an unsigned number, the method casts both the current value, and
the constant that we compare with to an unsigned number. As a
result if the constant we compare with is a negative number, it
wraps to some unsigned number, and the comparison is incorrect.
When the optimizer chooses the "range" access method, this problem
causes handler::read_range_next to reject the current key when the
upper bound key is a negative number because handler::compare_key
incorrectly considers the positive and negative keys to be equal.
The current patch does not correct the source of the problem in
Field_longlong::cmp because it is not easy to propagate sign
information about the constant at query execution time. Instead
the patch changes the range optimizer so that it never compares
unsiged fields with negative constants. As an added benefit,
queries that do such comparisons will execute faster because
the range optimizer replaces conditions like:
(a) (unsigned_int [< | <=] negative_constant) == FALSE
(b) (unsigned_int [> | >=] negative_constant) == TRUE
with the corresponding constants.
In some cases this may even result in constant time execution.
Remove changes made by bug fix#8147. They strips list of insert_table_list to
only insert table, which results in error reported in bug #9728.
Added flag to Item to resolve ambigous fields reported in bug #8147.
