Community
Participate
Working Groups
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.0.14) Gecko/2009090216 Ubuntu/9.04 (jaunty) Firefox/3.0.14 Build Identifier: 1.6.5 When I compile my aspects defined with the annotations-based approach (aka @AspectJ) with javac and then use ajc to weave statically them into some classes then ajc fails with NullPointerException at org.aspectj.weaver.patterns.ReferencePointcut.concretize1(ReferencePointcut.java:356) Since the aspects are normal .java files with special annotations then I don't see the reason why I couldn't compile them with javac. And even if there is a reason to require their compilation with ajc, it should throw a meaningful and helpful error instead of NPE (e.g. "May be the aspect classes have been compiled with javac instead of ajc as required, recompile them with ajc."). I'll attach a zip file that demonstrates the issue, see the steps to reproduce. Using AspectJ 1.6.5 Reproducible: Always Steps to Reproduce: 1.Unzip the attached aspectj-issue-demonstration.zip to /tmp 2 enter into the created /tmp/aspectj 3.(Optional) re-compile the aspect classes located in /tmp/aspectj/aspects/ by executing the provided javac.sh; or you may use those already present at javac_build/ 3.Run ajc using the provided ajc.sh 4.Ajc will fail with the error stored in ajc.log.
Created attachment 147869 [details] Archiive with source codes and commands demonstrating the issue
> Since the aspects are normal .java files with special annotations then I don't > see the reason why I couldn't compile them with javac. Yes, that is the intended design, you can just use javac. > And even if there is a reason to require their compilation with ajc, > it should throw a meaningful and helpful error instead of NPE Yep, the compiler should never NPE regardless of what the user does.
fix committed. The problem here is that the scope pointcut is blank. The pointcut resolver could not cope with that for an annotation style pointcut. The blank pointcut matches nothing so the effect here should be that the entire compound pointcut doesn't match anything. thanks for the test program.
