Re: free_space_realloc and measuring potentical benifit of free space optimization