This looks good from a functional perspective. But the code in store_temporary_table_record() is a bit messy, and the patch even adds a bit to that mess instead of cleaning it up.
There are 2 copies of the temp. table handler:
- 'handle' which was only used to get the storage engine type, and now, to clone file_handle_2
- 'file' which was used to get the actual stat values, and is now unused.
What about making 'file' a pointer to the cloned handler object. That would result in a bit cleaner code and less lines to touch?
Laurynas,
This looks good from a functional perspective. But the code in store_temporary _table_ record( ) is a bit messy, and the patch even adds a bit to that mess instead of cleaning it up.
There are 2 copies of the temp. table handler:
- 'handle' which was only used to get the storage engine type, and now, to clone file_handle_2
- 'file' which was used to get the actual stat values, and is now unused.
What about making 'file' a pointer to the cloned handler object. That would result in a bit cleaner code and less lines to touch?