I've noticed that the number of ReferenceType objects in an application that uses Spring 2.5.6 PointCuts with AspectJ (1.6.5) appears to be consistent with the following model:
k * P * T where: P is the number of PointCuts defined (15 in my case) T is the number of Types analysed (4500 in my case) and k is some overhead per-type analysed (10 in my case). So, in an application which perhaps 100 types actually need the pointcut applied, I am carrying a cost of k * P * T ~ 675000 ReferenceType objects. The * P multiplier appears to result from each PointCut getting its own ReflectionWorld and hence its own Type map. I would expect the long term storage consumed by the AspectJ type analysis to be linear in the number of intercepted types, not the number of analysed types and ideally some of the type analysis to be shared between point cuts. I take it that 1.6.7 might substantially improve this, is that so? jon. _______________________________________________ aspectj-users mailing list aspectj-users@eclipse.org https://dev.eclipse.org/mailman/listinfo/aspectj-users
