Community
Participate
Working Groups
An if() pointcut inside a perthis() causes the following dump: ABORT Exception thrown from AspectJ 1.2rc2 This might be logged as a bug already -- find current bugs at http://bugs.eclipse.org/bugs/buglist.cgi?product=AspectJ&component=Compiler Bugs for exceptions thrown have titles File:line from the top stack, e.g., "SomeFile.java:243" If you don't find the exception below in a bug, please add a new bug at http://bugs.eclipse.org/bugs/enter_bug.cgi?product=AspectJ To make the bug a priority, please include a test program that can reproduce this exception. null java.lang.NullPointerException at org.aspectj.weaver.patterns.IfPointcut.concretize1(IfPointcut.java:171) at org.aspectj.weaver.patterns.IfPointcut.concretize(IfPointcut.java:143) at org.aspectj.weaver.patterns.AndPointcut.concretize1(AndPointcut.java:88) at org.aspectj.weaver.patterns.Pointcut.concretize(Pointcut.java:135) at org.aspectj.weaver.patterns.Pointcut.concretize(Pointcut.java:122) at org.aspectj.weaver.patterns.PerObject.concretize(PerObject.java:84) at org.aspectj.weaver.CrosscuttingMembers.setPerClause(CrosscuttingMembers.java:207) at org.aspectj.weaver.ResolvedTypeX.collectCrosscuttingMembers(ResolvedTypeX.java:329) at org.aspectj.weaver.CrosscuttingMembersSet.addOrReplaceAspect(CrosscuttingMembersSet.java:57) at org.aspectj.weaver.bcel.BcelWeaver.prepareForWeave(BcelWeaver.java:322) at org.aspectj.ajdt.internal.compiler.AjCompilerAdapter.weave(AjCompilerAdapter.java:222) at org.aspectj.ajdt.internal.compiler.AjCompilerAdapter.afterCompiling(AjCompilerAdapter.java:114) at org.eclipse.jdt.internal.compiler.Compiler.compile(Compiler.java:376) at org.aspectj.ajdt.internal.core.builder.AjBuildManager.performCompilation(AjBuildManager.java:601) at org.aspectj.ajdt.internal.core.builder.AjBuildManager.doBuild(AjBuildManager.java:160) at org.aspectj.ajdt.internal.core.builder.AjBuildManager.batchBuild(AjBuildManager.java:94) at org.aspectj.ajdt.ajc.AjdtCommand.doCommand(AjdtCommand.java:102) at org.aspectj.ajdt.ajc.AjdtCommand.runCommand(AjdtCommand.java:53) at org.aspectj.tools.ajc.Main.run(Main.java:280) at org.aspectj.tools.ajc.Main.runMain(Main.java:217) at org.aspectj.tools.ajc.Main.main(Main.java:79) Here is a small test program to demonstrate the bug. Just try compiling it. public class Main { public static void main(String [] args) { System.out.println("Hello World"); } } // Note the use of an if inside a perthis, causes the ajc compiler to // throw an exception aspect Aspect perthis(call (* *(..)) // if you comment out the following line, everything is ok, so // it seems tobe the use of if in the perthis that is not handled correctly && if(4==3) ) { before () : call(* *..*(..)) && !within(Aspect*) { System.out.println("Advice 1"); } }
I'm still working on this one, but if you put the if() into a pointcut definition, this will fix the problem: aspect A perthis(definedPC()) { pointcut definedPC(): call(* *(..)) && if(4==3); ... It is to do with the 'scope' within which we are attempting to concretize the if (). The code seems to cope with two cases: 1) The if() is directly specified against advice. 2) The if() is in a declared pointcut definition. This is a third case where there is no surrounding advice or pointcut def.
I managed to put a fix in that meant we did not blow up at compile time. Unfortunately this then led to a situation where we crash at runtime. It turns out that we generate the implementation of the if_0() method (which contains the test you supplied in your if() PCD) when we do a postParse of the PointcutDeclaration (Driven through CompilationUnitScope.buildFieldsAndMethods ()). If the if() test is inlined in the per clause, we don't postparse the pointcut and never generate the if_0() method.
Ensuring the if() test method is generated requires a significant change that I don't want to make between an RC and a final release of 1.2. So, to prevent the compilation failing with an NPE I have added a compilation check to see if you are using if() in a per clause. If you are, you will see: "if() pointcut designator cannot be used directly in a per clause (compiler limitation). Create a named pointcut containing the if() and refer to it" I've added tests for all the various per clause types. I will then make this bug an enhancement request for post 1.2
The limitation that was the final resolution to this bug is not severe and easily worked around. No immediate plans to do anything further to alleviate this restriction. I'm marking as "FIXED" since that best describes the status of the main issue addressed by the bug report.
