You don't need to worry about enforcing locks, Milton will do that for you at the protocol level. And its probably better if you dont because there's some ambiguity about exactly who/what can operate on a locked resource. Milton has settled on a slightly fuzzy-logic implementation which is known to work across clients but if you do your own enforcement that could interfere.
The one time you should do your own checking for locks is if you want performance optimised deletes of collections.
Why deleting collections can be slow, and what to do about it
When a webdav client deletes a collection resource they should just issue a delete on the collection itself, and not its members (in reality some OS clients delete each member and there's nothing we can do about that).
But if a collection is deleted, its possible that any member of that collection is locked which should prevent the delete. Or, its possible that a user might have permission to delete the collection, but not some member of it - which should also prevent the delete.
To make sure these conditions are satisfied milton recursively deletes each member within the collection explicitly, thuse invoking its lock and permission checks. This can of course be very slow, such as if the collection contains thousands of resources which might happen to be database rows.
However, the application might be able to do the same check is a much more optimised way. It could check for any locks or permissions within the whole collection with a single SQL call (for example). So to support this Milton has an optional interface - DeletableCollectionResource
If you implement this on your collection milton will NOT do recursive deletes, it will just call the isLockedOut method and then, if ok, delete the collection.