has anyone worked out good way to handled orphaned files? My current solutions is cumbersome at best. I basically wrap all my mutations in the http post success callback.