Currently it is necessary to query the currentOp output in order to determine which operations belong to a client. It would be helpful if the client could obtain (or define) the op id (on cursor creation for example, or in a command result) so that the client could, for example 'clean up' after itself in the case of failure (or long running query).

For a practical example, the shell currently searches for all operations with it's own client address, in order to kill them when ctrl+c. If the shell knew the op ids of the operations that it was running, it would be a cleaner way of killing those operations on ctrl+c