Community
Participate
Working Groups
With the work done to split matching and weaving, we can build a JDTWorld that will enable matching without compilation. Preliminary work has started but it has uncovered that we are sorely lacking reusable pointcut test suites for the world, particular that use Java5 features (generics/annotations in pointcuts). Due to the sheer scale of the issue we need to create a roadmap for this: 1. Complete coverage of execution() pointcuts (Spring IDE wishes to exploit the JDTWorld) 2. Create a JDTTypeWalker that can walk a JDT element and create all the join points that occur within any given piece of code (initially just create execution join points) 3. Wire these two together and build the higher level API 4. Expand into further pointcuts (testcases and in JDTTypeWalker), first without Java5 5. Expand into further pointcuts (testcases and in JDTTypeWalker), with Java5 Note that in terms of actual new code to write, the only code that will be included in distributions is the JDTWorld and JDTTypeWalker - there is probably not much to do within the pointcut classes themselves, we know they work fine - it is just the tests that need writing so we can verify the behaviour across new World implementations rather than only with a BcelWorld.
Trying to figure out how AJDT can use this. Currently, AJDT's access to the structure model is primarily through IHierarchy and IRelationshipMap objects. Could a JDTWorld object (eventually) be used to create and populate IHierarchy and IRelationshipMap objects? This would be the ideal way for AJDT to manage crosscutting relationships for LTW. I am going to open a new enhancement request to discuss this.
My preliminary thoughts on how AJDT mught be able to use this are now covered in bug 265029.
unsetting the target field which is currently set for something already released