of bugs easier (MDEV-8919, MDEV-10304, MDEV-10305, MDEV-10307)
- Adding Item::push_note_converted_to_negative_complement() and
Item::push_note_converted_to_positive_complement()
- Adding virtual methods Item::val_int_signed_typecast() and
Item::val_int_unsigned_typecast()
- Moving COLUMN_GET() related code from
Item_func_signed::val_int() and Item_func_unsigned::val_int() to
Item_dyncol_get::val_int_signed_typecast() and
Item_dyncol_get::val_int_unsigned_typecast()
- Moving Item_func_signed::val_int_from_str() to Item::val_int_from_str()
and changing it to get the value from "this" instead of args[0].
The patch does not change behaviour. It's only to simplify fixing of the
mentioned bugs. It will also simplify switching the CAST related code to
use the type handler infrastructure easier (soon).
Window functions need to have their own column in the work (temp) table,
like aggregate functions do.
They don't need val_int() -> val_int_result() conversion though, so they
should be wrapped with Item_direct_ref, not Item_aggregate_ref.
Item_func_or_sum.
Implemented method update_used_tables for class Item_findow_func.
Added the flag Item::with_window_func.
Made sure that window functions could be used only in SELECT list
and ORDER BY clause.
Added test cases that checked different illegal placements of
window functions.
1. Fixing Field_time::get_equal_const_item() to pass TIME_FUZZY_DATES
and TIME_INVALID_DATES to get_time_with_conversion().
This is needed to make the recursively called Item::get_date() return
non-NULL values on garbage input. This makes Field_time::get_equal_const_item()
work consistently with how Item::val_time_packed() works.
2. Fixing Item::get_date() to return TIME'00:00:00' rather than
DATE'0000-00-00' on empty or garbage input when:
- TIME_FUZZY_DATES is enabled
- The caller requested a TIME value (by passing TIME_TIME_ONLY).
This is needed to avoid conversion of DATE'0000-00-00' to TIME
in get_time_with_conversion(), which would erroneously try to subtract
CURRENT_DATE from DATE'0000-00-00' and return TIME'-838:59:59' rather than
the desired zero value TIME'00:00:00'.
#1 and #2 fix these type of scripts to return one row with both
MyISAM and InnoDB, with and without an index on t1.b:
CREATE TABLE t1 (a ENUM('a'), b TIME, c INT, KEY(b));
INSERT INTO t1 VALUES ('','00:00:00',0);
SELECT * FROM t1 WHERE b='';
SELECT * FROM t1 WHERE a=b;
SELECT * FROM t1 IGNORE INDEX(b) WHERE b='';
SELECT * FROM t1 IGNORE INDEX(b) WHERE a=b;
Additionally, #1 and #2 fix the originally reported in MDEV-9604 crash
in Item::save_in_field(), because now execution goes through a different
path, so save_in_field() is called for a Item_time_literal instance
(which is non-NULL) rather than a Item_cache_str instance (which could
return NULL without setting null_value).
3. Fixing Field_temporal::get_equal_const_item_datetime() to enable
equal field propagation for DATETIME and TIMESTAMP in case of
comparison (e.g. when ANY_SUBST), for symmetry with
Field_newdate::get_equal_const_item(). This fixes a number of problems
with empty set returned on comparison to empty/garbage input.
Now all SELECT queries in this script return one row for MyISAM and InnoDB,
with and without an index on t1.b:
CREATE TABLE t1 (a ENUM('a'), b DATETIME, c INT, KEY(b));
INSERT INTO t1 VALUES ('','0000-00-00 00:00:00',0);
SELECT * FROM t1 WHERE b='';
SELECT * FROM t1 WHERE a=b;
SELECT * FROM t1 IGNORE INDEX(b) WHERE b='';
SELECT * FROM t1 IGNORE INDEX(b) WHERE a=b;
when doing set_field_to_new_field (from switch_to_nullable_trigger_fields())
make sure that the field we're about to change actually belongs
to the right table (otherwise we cannot dereference new_field[]
array as the wrong table might have more fields than
new_field[] has elements)
Case: table with a NOT NULL field, BEFORE UPDATE trigger,
and UPDATE with a subquery that uses GROUP BY on that
NOT NULL field, and needs a temporary table for it.
Because of the BEFORE trigger, the field becomes nullable
temporarily. But its Item_field (used in GROUP BY) doesn't.
When working with the temptable some code looked at
item->maybe_null, some - at field->null_ptr.
The fix: make Item_field nullable when its field is.
This triggers an assert. The group key size is calculated
before the item is made nullable, so the group key doesn't
have a null byte. The fix: make fields/items nullable
before the group key size is calculated.
"Re-factor the code for post-join operations".
The patch mainly contains the code ported from mysql-5.6 and
created for two essential architectural changes:
1. WL#5558: Resolve ORDER BY execution method at the optimization stage
2. WL#6071: Inline tmp tables into the nested loops algorithm
The first task was implemented for mysql-5.6 by Ole John Aske.
It allows to make all decisions on ORDER BY operation at the optimization
stage.
The second task implemented for mysql-5.6 by Evgeny Potemkin adds JOIN_TAB
nodes for post-join operations that require temporary tables. It allows
to execute these operations within the nested loops algorithm that used to
be used before this task only for join queries. Besides these task moves
all planning on the execution of these operations from the execution phase
to the optimization phase.
Some other re-factoring changes of mysql-5.6 were pulled in, mainly because
it was easier to pull them in than roll them back. In particular all
changes concerning Ref_ptr_array were incorporated.
The port required some changes in the MariaDB code that concerned the
functionality of EXPLAIN and ANALYZE. This was done mainly by Sergey
Petrunia.
NOT NULL constraint must be checked *after* the BEFORE triggers.
That is for INSERT and UPDATE statements even NOT NULL fields
must be able to store a NULL temporarily at least while
BEFORE INSERT/UPDATE triggers are running.
* move common code to a new set_bad_null_error() function
* move repeated comparison out of the loop
* remove unused code
* unused method Table_triggers_list::set_table
* redundant condition (if (table) after table was dereferenced)
* add an assert
The following left in semi-improved state to keep patch size reasonable:
- Field operator new: left thd_alloc(current_thd)
- Sql_alloc operator new: left thd_alloc(thd_get_current_thd())
- Item_args constructors: left thd_alloc(thd)
- Item_func_interval::fix_length_and_dec(): no THD arg, have to call current_thd
- Item_func_dyncol_exists::val_int(): same
- Item_dyncol_get::val_str(): same
- Item_dyncol_get::val_int(): same
- Item_dyncol_get::val_real(): same
- Item_dyncol_get::val_decimal(): same
- Item_singlerow_subselect::fix_length_and_dec(): same
Problem:
At the end of first execution select_lex->prep_where is pointing to
a runtime created object (temporary table field). As a result
server exits trying to access a invalid pointer during second
execution.
Analysis:
While optimizing the join conditions for the query, after the
permanent transformation, optimizer makes a copy of the new
where conditions in select_lex->prep_where. "prep_where" is what
is used as the "where condition" for the query at the start of execution.
W.r.t the query in question, "where" condition is actually pointing
to a field in the temporary table. As a result, for the second
execution the pointer is no more valid resulting in server exit.
Fix:
At the end of the first execution, select_lex->where will have the
original item of the where condition.
Make prep_where the new place where the original item of select->where
has to be rolled back.
Fixed in 5.7 with the wl#7082 - Move permanent transformations from
JOIN::optimize to JOIN::prepare
Patch for 5.5 includes the following backports from 5.6:
Bugfix for Bug12603141 - This makes the first execute statement in the testcase
pass in 5.5
However it was noted later in in Bug16163596 that the above bugfix needed to
be modified. Although Bug16163596 is reproducible only with changes done for
Bug12582849, we have decided include the fix.
Considering that Bug12582849 is related to Bug12603141, the fix is
also included here. However this results in Bug16317817, Bug16317685,
Bug16739050. So fix for the above three bugs is also part of this patch.
The problem was that GROUP BY code created Item_field objects
that referred to fields in the temp. tables used for GROUP BY.
Item_ref and set_items_ref_array() call caused pointers to temp.
table fields to occur in many places.
This patch introduces Item_temptable_field, which can handle
item->print() calls made after the underlying table is freed.
Item_string::clone_item() creates a new Item_string that
points exactly to the same buffer that the original one does.
Later, Item_string::print() uses c_ptr() for the original Item_string,
which reallocs the original buffer, and the clone remain with
the old freed buffer.
Refactoring the code not to use c_ptr() in Item_string::print().
1. Removing the legacy code that disabled equal field propagation in cases
when comparison is done as VARBINARY. This is now correctly handled by
the new propagation code in Item_xxx::propagate_equal_fields() and
Field_str::can_be_substituted_to_equal_item (the bug fix).
2. Also, removing legacy (pre-MySQL-4.1) Arg_comparator methods
compare_binary_string() and compare_e_binary_string(), as VARBINARY
comparison is correcty handled in compare_string() and compare_e_string() by
the corresponding VARBINARY collation handler implemented in my_charset_bin.
(not really a part of the bug fix)
WHERE COALESCE(time_column)=TIME('00:00:00')
AND COALESCE(time_column)=DATE('2015-09-11')
MDEV-8814 Wrong result for WHERE datetime_column > TIME('00:00:00')
