This commit removes the `ORDER BY option_id` part of the query to remove the version transients from the `wp_options` table. This change should help users having performance issues with this query (see for example #21215) and, as far as I can see, ordering is not necessary in this context.
Removing the `ORDER BY` from the query means that MySQL won't have to use filesort:
```
MySQL [wccore]> explain DELETE FROM wp_options WHERE option_name LIKE '\\_transient\\_%1510258328' ORDER BY option_id LIMIT 1000;
+----+-------------+------------+------------+-------+---------------+-------------+---------+-------+------+----------+-----------------------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+------------+------------+-------+---------------+-------------+---------+-------+------+----------+-----------------------------+
| 1 | DELETE | wp_options | NULL | range | option_name | option_name | 766 | const | 67 | 100.00 | Using where; Using filesort |
+----+-------------+------------+------------+-------+---------------+-------------+---------+-------+------+----------+-----------------------------+
1 row in set (0.02 sec)
MySQL [wccore]> explain DELETE FROM wp_options WHERE option_name LIKE '\\_transient\\_%1510258328' LIMIT 1000;
+----+-------------+------------+------------+-------+---------------+-------------+---------+-------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+------------+------------+-------+---------------+-------------+---------+-------+------+----------+-------------+
| 1 | DELETE | wp_options | NULL | range | option_name | option_name | 766 | const | 67 | 100.00 | Using where |
+----+-------------+------------+------------+-------+---------------+-------------+---------+-------+------+----------+-------------+
1 row in set (0.00 sec)
```
This commit changes wp_woocommerce_sessions primary key to the bigint field session_id. Before it was the char(32) field session_key. Doing this change primarily as it should reduce the occurrence of deadlocks, but also because it is not a good practice to use a char(32) field as the primary key of a table.
If the child product doesn't have weight or dimensions, the value is not being cached correctly as a ```false``` value is being stored and it always recheck that on every page load.
Peter Fabian
Jeff Stieler
Gerhard Potgieter
claudiulodro
Claudio Sanches
Jesse Pearson
Rodrigo Primo
Robert DeVore
Manos Psychogyiopoulos
Nikhil Barar
Ruben
Paul Dechov