Next we shall discuss the use of around advice, which provides you with greater control than do before and after advice. Let us suppose that a function that has some around advice is called. The arguments to the function are passed to the code associated with the first piece of around advice in the ordering, and the values returned by that piece of advice are the results of the function. There is no requirement for the advice to invoke any other pieces of advice, nor to call the original definition of the function.
However the code for any piece of around advice has access to the next member of the ordering, which it may invoke any number of times by calling
call-next-advice
. So it is possible for each piece of around advice to call its successor in the ordering if this is desired, and then the bits of around advice are called in turn in a similar fashion to our earlier description for before and after advice. However in the case of around advice the decision whether or not to call the next piece of advice is directly under your control, and you are free to modify the arguments received by the piece of advice, and to choose the arguments to be given to the next piece of advice if it is called.
If the last piece of around advice in the ordering calls
call-next-advice
, then it invokes the combination of before and after advice and the original definition that was discussed earlier. That is, the arguments to the call are given in the sequence described above to each of the before pieces of advice, then to the original definition and then to the after pieces of advice. The call to
call-next-advice
returns with the values produced by the last of these subsidiary calls, and the around advice may use these values in any way.