initialization crashes server.
Make sure variables are initialized to something (like null) when
the default initialization fails and a continue handler is in effect.
Two handler objects were present, one was used for an insert and the other for a select
The state of the statistics was local to the handler object and thus the other handler
object didn't notice the insert.
Fix included:
1) Add a new variable key_stat_version added to whenever statistics was considered in need
of update (previously key_stats_ok= FALSE in those places)
2) Add a new handler variable key_stat_version assigned whenever key_stats_ok= TRUE was set
previously
3) Fix records_in_range to return records if records <= 1
4) Fix records_in_range to add 2 to rec_per_key to ensure we don't specify 0 or 1 when it isn't
and thus invoking incorrect optimisations.
5) Fix unique key handling for HEAP table in records_in_range
non-deterministic result in the test case for BUG#7947
the bug fix for BUG#7947 now fixed the result of mix_innodb_myisam_binlog test, which
in the past was missing DO RELEASE_LOCK() in the output of SHOW BINLOG EVENTS
test "length" first (otherwise when "length" is 0, the *from invalid access still triggers a Valgrind warning).
I wrote to the Valgrind authors in case this is something fixable in Valgrind (normally the
decision to issue a warning is based on the simulated CPU condition code, which should not be undefined here).
Initialized usable_keys from table->keys_in_use instead of ~0
in test_if_skip_sort_order(). It was possible that a disabled
index was used for sorting.
ctype_utf8.result, ctype_utf8.test:
Adding test case.
item_strfunc.cc:
item_strfunc.h:
Moving the well formed checking code into a method,
to reuse in several Item_func_xxx. Reusing the new
method in Item_func_char and Item_func_charset_conv.
The problem was to continue at the right place in the code after the
test expression in a flow control statement fails with an exception
(internally, the test in sp_instr_jump_if_not), and the exception is
caught by a continue handler. Execution must then be resumed after the
the entire flow control statement (END IF, END WHILE, etc).
Version for 4.0.
It fixes two problems:
1. The cause of the bug was that we did not check the table version for
the HANDLER ... READ commands. We did not notice when a table was
replaced by a new one. This can happen during ALTER TABLE, REPAIR
TABLE, and OPTIMIZE TABLE (there might be more cases). I call the fix
for this problem "the primary bug fix".
2. mysql_ha_flush() was not always called with a locked LOCK_open.
Though the function comment clearly said it must.
I changed the code so that the locking is done when required. I call
the fix for this problem "the secondary fix".
