To prevent trashing with many notifications, there are three strategies:

    1. - disable notifications, to the operations, re-enable the notifications.
    2. this has the disadavantage that you might not know what notifications to
    3. send when they are re-enabled. One solution could be to spool them,
    4. and at re-enable, merge the duplicates and send out the minimum.
    5. - have coarse grained operations, operating on large sets and sending out
    6. only one notification at the end.
    7. - Have fine grained modification routines with an option notify that allows
    8. to decide when to send the notification and when not to.
    9. - Have the model be a centralizer of the notification delivery, but have notifyObserver called
    10. externally.
    11. - have a smart signal that can be put in a "trasaction on" mode, and accumulates the
    12. notifications, and then release the notification when a "commit" is issued