added setting of 'neg' in Item_param::set_time() (looks like the only forgotten
member).
It's the second place I find where 'neg' was forgotten.
The symptom was unexpected negative times in the binary log when running
tests/client_test.c (test_date() in fact):
# at 43009
#040210 15:46:42 server id 1 log_pos 43009 Query thread_id=1 exec_time=0 error_code=0
SET TIMESTAMP=1076424402;
INSERT INTO test_date VALUES('2000-01-10 11:16:20','-11:16:20' etc
Don't show PSEUDO_THREAD_ID in SHOW VARIABLES because:
- we don't want people to discover this variable as it could never do good to
set it (it was designed for use by mysqlbinlog only, so that a thread can have
several temp tables of the same name at the same time)
- if we show it in SHOW VARIABLES, Mysql Administrator will display it and
this will force us to put a description, so all MySQL Administrator user
will be aware of this variable, some may have the idea to set it with a SET
command, and then it will cause bad things.
The variable is still settable, and still visible with SELECT @@.
Done clean-up in prep stmt API functions:
1) Removed some checks that were performed only in debug version
were making debug version more tolerable to user errors than
production (and thus caused problems for example masking some
bugs).
2) Also removed some other checks to make prep stmt API
consistent with the rest of C API (this also in line with
general politics - make checks in only those places where
errors are very common and hard to spot).
Allow always DROPping of a table which is only referenced by FOREIGN KEY constraints from the same table
Many files:
Do not let REPLACE to perform internally an UPDATE if the table is referenced by a FOREIGN KEY: the manual says that REPLACE must resolve a duplicate key error semantically with DELETE(s) + INSERT, and not by an UPDATE; the internal update caused foreign key checks and cascaded operations to behave in a semantically wrong way
Fixed output from mysqlbinlog when using --skip-comments
Fixed warnings from valgrind
Fixed ref_length when used with HEAP tables
More efficent need_conversion()
Fixed error handling in UPDATE with not updateable tables
Fixed bug in null handling in CAST to signed/unsigned
Fix bug #2167: generate foreign key id's locally for each table, in the form databasename/tablename_ibfk_number; if the user gives the constraint name explicitly remember it; these changes should ensure that foreign key id's in a slave are the same as in the master, and DROP FOREIGN KEY does not break replication
sync0sync.c:
UNIV_SYNC_DEBUG caused assertion in the creation of the doublewrite buffer, if we do not allow thousands of latches per thread
If AUTOCOMMIT=1, then we do not need to make a plain SELECT set shared locks even on the SERIALIZABLE isolation level, because we know the transaction is read-only: a read-only transaction can always be performed on the REPEATABLE READ level, and that does not endanger the serializability
