Weaving Multiple Aspects in Sequence Diagrams - Transactions on Aspect-Oriented Software Development - page 179

In a nutshell, for each definition of apartofabSD , there is a corresponding
definition of join point. In Fig. 7, if we consider the pointcut depicted, it is easy

to see that the messages 2areajoinpointifwetaketheenclosed

part, the safe part or the general part as definition of part of a bSD, because
there exists a bSD isomorphism between the pointcut and an enclosed part, a
safe part or a general part of

m 1and m

M

2.

3.3 Successive Join Points

To define the notion of successive join points the simple definition of join point
is not precise enough. Indeed, in Fig. 8, the pointcut P 1 matches two different

parts of M 1, but these parts become entangled. Let us consider now the pointcut

P 2. If we take the definition of general part as the definition of

part, there are four possible join points. Indeed, the first message

2 and the bSD M

a and the first

message b can form a join point, as can the second message a and the second

message b , but the first message a with the second message b or the second

a b

message with the first message can also form join points.

Fig. 8. Multiple possible matching

These multiple conflicting join points pose a problem. Among the conflicting
join points, which should be chosen? Considering all these potential join points
might not make sense from the point of view of someone using aspect weaving
tools. In practice, we can reasonably expect that when a join point is detected,
no elements of this join point can be used to form another join point. Roughly
speaking, we can define successive join points as a set of disjoint join points
taken in sequence.