Under row-based replication, DELETE FROM will now always be
replicated as individual row deletions, while TRUNCATE TABLE will
always be replicated as a statement.
The check for view security was lacking several points :
1. Check with the right set of permissions : for each table ref that
participates in a view there were the right credentials to use in it's
security_ctx member, but these weren't used for checking the credentials.
This makes hard enforcing the SQL SECURITY DEFINER|INVOKER property
consistently.
2. Because of the above the security checking for views was just ruled out
in explicit ways in several places.
3. The security was checked only for the columns of the tables that are
brought into the query from a view. So if there is no column reference
outside of the view definition it was not detecting the lack of access to
the tables in the view in SQL SECURITY INVOKER mode.
The fix below tries to fix the above 3 points.
counter".
When TRUNCATE TABLE was called within an stored procedure the
auto_increment counter was not reset to 0 even if straight
TRUNCATE for this table did this.
This fix makes TRUNCATE in stored procedures to be handled exactly
in the same way as straight TRUNCATE. We achieve this by rolling
back the fix for bug 8850, which is no longer needed since stored
procedures don't require prelocked mode anymore (and TRUNCATE is
not allowed in stored functions or triggers).
Added missing DBUG_xxx_RETURN statements
Fixed some usage of not initialized variables (as found by valgrind)
Ensure that we don't remove locked tables used as name locks from open table cache until unlock_table_names() are called.
This was fixed by having drop_locked_name() returning any table used as a name lock so that we can free it in unlock_table_names()
This will allow Tomas to continue with his work to use namelocks to syncronize things.
Note: valgrind still produces a lot of warnings about using not initialized code and shows memory loss errors when running the ndb tests
rows for SEs using injector):
Table truncation ("DELETE FROM t1" and "TRUNCATE t1") was logged as
a statement even when the storage engine deletes the rows individually
using the injector.
and new binlog format called "mixed" (which is statement-based except if only row-based is correct,
in this cset it means if UDF or UUID is used; more cases could be added in later 5.1 release):
SET GLOBAL|SESSION BINLOG_FORMAT=row|statement|mixed|default;
the global default is statement unless cluster is enabled (then it's row) as in 5.1-alpha.
It's not possible to use SET on this variable if a session is currently in row-based mode and has open temporary tables (because CREATE
TEMPORARY TABLE was not binlogged so temp table is not known on slave), or if NDB is enabled (because
NDB does not support such change on-the-fly, though it will later), of if in a stored function (see below).
The added tests test the possibility or impossibility to SET, their effects, and the mixed mode,
including in prepared statements and in stored procedures and functions.
Caveats:
a) The mixed mode will not work for stored functions: in mixed mode, a stored function will
always be binlogged as one call and in a statement-based way (e.g. INSERT VALUES(myfunc()) or SELECT myfunc()).
b) for the same reason, changing the thread's binlog format inside a stored function is
refused with an error message.
c) the same problems apply to triggers; implementing b) for triggers will be done later (will ask
Dmitri).
Additionally, as the binlog format is now changeable by each user for his session, I remove the implication
which was done at startup, where row-based automatically set log-bin-trust-routine-creators to 1
(not possible anymore as a user can now switch to stmt-based and do nasty things again), and automatically
set --innodb-locks-unsafe-for-binlog to 1 (was anyway theoretically incorrect as it disabled
phantom protection).
Plus fixes for compiler warnings.
- Encoding itself, implemented as a charset
"filename". Originally planned to use '.'
as an escape character, but now changed to '@'
for two reasons: "ls" does not return
file names starting with '.' considering them
as a kind of hidden files; some platforms
do not allow several dots in a file name.
- replacing many calls of my_snprintf() and
strnxmov() to the new build_table_filename().
- Adding MY_APPEND_EXT mysys flag, to append
an extention rather that replace it.
- Replacing all numeric constants in fn_format
flag arguments to their mysys definitions, e.g.
MY_UNPACK_FILENAME,
- Predictability in several function/methods:
when a table name can appear with or withot .frm
extension. Some functions/methods were changed
so accept names strictly with .frm, other - strictly
without .frm extensions. Several DBUG_ASSERTs were
added to check whether an extension is passed.
Many files:
table name to file name encoding
mysql_priv.h:
Prototypes for new table name encoding tools.
ctype-utf8.c:
Implementing "filename" charset for
table name to file name encoding.
row0mysql.c:
Fixing table name prefix.
mf_format.c:
Adding MY_APPEND_EXT processing.
Many files:
Fixing tests.
my_sys.h:
Adding new flag to append rather than replace an extension.
m_ctype.h:
Adding "filename" charset definition.
Problem #1: INSERT...SELECT, Version for 5.1.
Extended the unique table check by a check of lock data.
Merge sub-tables cannot be detected by doing name checks only.
Problem #1: INSERT...SELECT, Version for 5.0.
Extended the unique table check by a check of lock data.
Merge sub-tables cannot be detected by doing name checks only.
The table opening process now works the following way:
- Create common TABLE_SHARE object
- Read the .frm file and unpack it into the TABLE_SHARE object
- Create a TABLE object based on the information in the TABLE_SHARE
object and open a handler to the table object
Other noteworthy changes:
- In TABLE_SHARE the most common strings are now LEX_STRING's
- Better error message when table is not found
- Variable table_cache is now renamed 'table_open_cache'
- New variable 'table_definition_cache' that is the number of table defintions that will be cached
- strxnmov() calls are now fixed to avoid overflows
- strxnmov() will now always add one end \0 to result
- engine objects are now created with a TABLE_SHARE object instead of a TABLE object.
- After creating a field object one must call field->init(table) before using it
- For a busy system this change will give you:
- Less memory usage for table object
- Faster opening of tables (if it's has been in use or is in table definition cache)
- Allow you to cache many table definitions objects
- Faster drop of table
Not fixed in 4.1 as not critical. Also I'm correcting error checking of multi-UPDATE/DELETE
when it comes to binlogging, to make it consistent with when we rollback the statement.
"Interleaved SPs execution is now binlogged properly, "SELECT spfunc()" is binlogged too.
The known remaining issue is binlogging/replication of "a routine is deleted while it is executed" scenario.
"Process NATURAL and USING joins according to SQL:2003".
* Some of the main problems fixed by the patch:
- in "select *" queries the * expanded correctly according to
ANSI for arbitrary natural/using joins
- natural/using joins are correctly transformed into JOIN ... ON
for any number/nesting of the joins.
- column references are correctly resolved against natural joins
of any nesting and combined with arbitrary other joins.
* This patch also contains a fix for name resolution of items
inside the ON condition of JOIN ... ON - in this case items must
be resolved only against the JOIN operands. To support such
'local' name resolution, the patch introduces a stack of
name resolution contexts used at parse time.
NOTICE:
- This patch is not complete in the sense that
- there are 2 test cases that still do not pass -
one in join.test, one in select.test. Both are marked
with a comment "TODO: WL#2486".
- it does not include a new test specific for the task
cause crash on update".
Let us update "thd" pointer in LEX, all its units and in LEX::result before
executing statement in trigger body, since triggers are associated with TABLE
object and because of this can be used in different threads.
"the server side preparedStatement error for LIMIT placeholder",
which moves all uses of LIMIT clause from PREPARE to OPTIMIZE
and later steps.
After-review fixes.
databases" and basic handling of errors which happen in triggers.
(The bug itself was fixed by several previous patches).
Fixed bug in multi-delete which were exposed by these tests.
#5860 "Multi-table UPDATE does not activate update triggers"
#6812 "Triggers are not activated for INSERT ... SELECT"
#8755 "Trigger is not activated by LOAD DATA".
This patch also implements proper handling of triggers for special forms
of insert like REPLACE or INSERT ... ON DUPLICATE KEY UPDATE.
Also now we don't call after trigger in case when we have failed to
inserted/update or delete row. Trigger failure should stop statement
execution.
I have not properly tested handling of errors which happen inside of
triggers in this patch, since it is simplier to do this once we will be
able to access tables from triggers.
Fixed bug #8528.
Representation for single-table views was made similar to
representation for multi-table views.
view.test:
Added test case for bug #8528.
view.result:
Added test case for bug #8528. Fixed other test cases.
Added a test case for bug #8392.
sql_delete.cc:
Fixed bug #8392.
The bug caused a crash for a delete statement with ORDER BY
that explicitly referred to the modified table.
Split TABLE to TABLE and TABLE_SHARE (TABLE_SHARE is still allocated as part of table, will be fixed soon)
Created Field::make_field() and made Field_num::make_field() to call this
Added 'TABLE_SHARE->db' that points to database name; Changed all usage of table_cache_key as database name to use this instead
Changed field->table_name to point to pointer to alias. This allows us to change alias for a table by just updating one pointer.
Renamed TABLE_SHARE->real_name to table_name
Renamed TABLE->table_name to alias
Renamed TABLE_LIST->real_name to table_name
CREATE DATABASE statement used the current database instead of the
database created when checking conditions for replication.
CREATE/DROP/ALTER DATABASE statements are now replicated based on
the manipulated database.
If we have DELETE with always true WHERE clause we should not use
optimized delete_all_rows() method for tables with DELETE triggers,
because in this case we will lose side-effect of deletion.
Mostly needed for Monty for him getting notion what needed for triggers
from new .FRM format.
Things to be done:
- Right placement of trigger's invocations
- Right handling of errors in triggers (including transaction rollback)
- Support for priviliges
- Right handling of DROP/RENAME table (hope that it will be handled automatically
with merging of .TRG into .FRM file)
- Saving/restoring some information critical for trigger creation and replication
with their definitions (e.g. sql_mode, creator, ...)
- Replication
Already has some known bugs so probably not for general review.
added tests to alter table for "large" alter tables and truncates in ndbcluster
added debug printout in restart() in ndbcluster
added flag THD::transaction.on to enable/disable transaction
more logical table/index_flags
return HA_ERR_WRONG_COMMAND instead of abstract methods where appropriate
max_keys and other limits renamed to max_supported_keys/etc
max_keys/etc are now wrappers to max_supported_keys/etc
ha_index_init/ha_rnd_init/ha_index_end/ha_rnd_end are now wrappers to real {index,rnd}_{init,end} to enforce strict pairing
binlog even if they changed nothing, and a test for this.
This is useful when users use these commands to clean up their master and slave by issuing
one command on master (assume master and slave have slightly different data for some
reason and you want to clean up both).
Note that I have not changed multi-table DELETE and multi-table UPDATE because their
error-reporting mechanism is more complicated.
