&sys_buf + DATA_TRX_ID_LEN is not &(sys_buf + DATA_TRX_ID_LEN).

create_table_def(), ha_innobase::create(): Defer fts_optimize_add_table()
until after the table has been successfully created.

row_fts_merge_insert(): Correctly initialize DB_ROLL_PTR to a safe value
that will not be dereferenced by MVCC.

This patch changes how old rows in mysql.gtid_slave_pos* tables are deleted.
Instead of doing it as part of every replicated transaction in
record_gtid(), it is done periodically (every @@gtid_cleanup_batch_size
transaction) in the slave background thread.

This removes the deletion step from the replication process in SQL or worker
threads, which could speed up replication with many small transactions. It
also decreases contention on the global mutex LOCK_slave_state. And it
simplifies the logic, eg. when a replicated transaction fails after having
deleted old rows.

With this patch, the deletion of old GTID rows happens asynchroneously and
slightly non-deterministic. Thus the number of old rows in
mysql.gtid_slave_pos can temporarily exceed @@gtid_cleanup_batch_size. But
all old rows will be deleted eventually after sufficiently many new GTIDs
have been replicated.

only --datadir option is tested in this test.
Other options (notably --password and --service) would need another,
more comprehensive test.

merge_role_db_privileges() was remembering pointers into Dynamic_array
acl_dbs, and later was using them, while pushing more elements into the
array. But pushing can cause realloc, and it can invalidate all pointers.

Fix: remember and use indexes of elements, not pointers.

sorting them by usernames first, and then by get_sort() value.


Search functions now use binary search to find the the first entry with
given name. Then, linear search is done, until the first match.

We must use the original page format for inserting metadata records.
This is because in case the instant ALTER TABLE operation needs to
be rolled back (due to crash recovery), all pages must remain in
the original format.

btr_cur_optimistic_insert(): Avoid reorganize for inserting the
metadata record.

btr_cur_optimistic_update(): Initialize rec,offsets correctly
after possible page format conversion.

innobase_instant_try(): Do not store NULL values for those
metadata record columns that were originally declared NOT NULL.

dict_index_t::get_n_nullable(): Refer to dict_col_t::was_not_null()
in order to keep the original format intact.

lock_move_reorganize_page(): Allow the two pages to be in
different formats.

rec_convert_dtuple_to_rec_comp(), rec_convert_dtuple_to_rec_comp():
Also observe the DATA_WAS_NOT_NULL flag.

btr_page_create(): Add the parameter 'bool flexible', similar
to btr_page_empty(). In this way, btr_root_raise_and_insert()
and btr_page_split_and_insert() can avoid converting the page
when splitting it. If a page were converted to flexible format
during the split, the page split could theoretically fail
due to the growth of the record header.

Remove some bogus checks for os_has_said_disk_full, to avoid
dereferencing NULL pointers.

btr_page_split_and_insert(): If needed, convert the page to flexible
format right before inserting the record. Do not convert pages when
inserting a metadata record, because we want to be able to roll back
the transaction.

btr_can_merge_with_page(): Convert the page to flexible format if needed.

btr_lift_page_up(): Preserve the format of the page.

page_copy_rec_list_convert(): On page overflow, delete the records
that were copied so far, and return an error.

btr_page_reorganize_low(): Change the first parameter type from bool
to the new type reorganize_t, so that REORGANIZE_CONVERT can be
distinguished from the special case REORGANIZE_KEEP_FORMAT.

btr_cur_optimistic_delete_func(), btr_cur_pessimistic_delete():
Invoke REORGANIZE_KEEP_FORMAT when rolling back the insert of
a metadata record during the rollback of instant ALTER TABLE.
We must preserve the original format in that case.

page_cur_insert_rec_write_log(): Add a const qualifier.

page_copy_rec_list_end_no_locks(): Propagate the error from
page_copy_rec_list_convert().

page_copy_rec_list_to_created_page_write_log(),
page_copy_rec_list_end_to_created_page(): Moved to a different
compilation unit.

btr_cur_optimistic_insert(), btr_cur_optimistic_update():
Propagate btr_page_reorganize() failure to the caller.

btr_cur_pessimistic_update(): If page format conversion is needed
and it fails, fall back to btr_cur_pessimistic_insert().

btr_insert_into_right_sibling(): Convert the page before insertion if needed.

btr_cur_pessimistic_delete(): Convert the page before btr_compress().

btr_compress(): Assert that the page be in the correct format.

btr_cur_optimistic_update(), btr_cur_pessimistic_update():
Convert the page to flexible format if needed.

dtuple_convert_big_rec(): Use the least efficient format.

The previous patch 269da4bf was
actually for MDEV-8894 (not for MDEV-5377). It was erroneously
pushed with a wrong title.

This patch is a small cleanup for MDEV-8894.
CREATE TABLE is now not a part of binary logs recorded with MySQL,
only INSERT statements are. This will allow to reuse the same binary
logs in combinations with different CREATE TABLE statements,
to tests different data types.

MDEV-5377 Row-based replication of MariaDB temporal data types with FSP>0 into a different column type