Bugzilla – Full Text Bug Listing
| Summary: | perthis inserts perObjectField to all compiled classes | ||
|---|---|---|---|
| Product: | [Tools] AspectJ | Reporter: | Viktor Kuzmin <kva> |
| Component: | Compiler | Assignee: | aspectj inbox <aspectj-inbox> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | CC: | aclement |
| Version: | 1.6.11 | ||
| Target Milestone: | 1.6.12 | ||
| Hardware: | PC | ||
| OS: | Linux | ||
| Whiteboard: | |||
The current implementation is working 'as designed'. But I agree it isn't optimal. I've made the changes to cope with this specific situation (a method execution pointcut that specifies annotations), it is a little ugly because the weaver is not intended to do matching in this way (it usually uses a vague test up front to determine which types might have an aspect, giving them the field, then does more accurate matching later - the extra field is unused when the accurate match actually fails). So 1.6.12 will do a better job here. Do you find the extra field actually breaks something? Or are you just observing that it exists where you don't want it? (In reply to comment #1) > Do you find the extra field actually breaks something? Or are you just > observing that it exists where you don't want it? Extra fields are added to all objects -> they eat a little bit more space but the main problem is that they make harder to do introspection. |
I need to have perthis aspect for classes which have annotated methods: @Aspect("perthis(transactional()) public class TransactionalAspect { @Pointcut("execution(@Transactional * * (..))") public void transactional() { } } It is supposed that perObjectField should be created only for classes which have annotated methods, but currently perObjectField is added to each class.