ASTConverterSuperAfterStatements.test004 fails on master

[srikanth-sankaran]

ASTConverterSuperAfterStatements.test004 fails on master

[srikanth-sankaran]

@jarthana, @mpalat - I am following up.

[stephan-herrmann]

I suggest you disable the test by prefixing with _ and merge a commit to master - otherwise we are going to see a failing tests on all PRs until this is resolved.

Yes please. All PR builds are failing due to this for more than 4 days now.

[srikanth-sankaran]

Code generated for the constructor is

      stack=2, locals=2, args_size=1
         0: new           #8                  // class X$1
         3: dup
         4: invokespecial #10                 // Method X$1."<init>":()V
         7: astore_1
         8: aload_0
         9: invokespecial #12                 // Method java/lang/Object."<init>":()V
        12: return

In the call stack below, we compute the receiverAndArgsSize to be 2 - that looks suspect to me compared to the signature of the <init> method invokespecialed which is **()V**

TypeAnnotationCodeStream(CodeStream).invoke(byte, MethodBinding, TypeBinding, TypeReference[]) line: 4785	
TypeAnnotationCodeStream.invoke(byte, MethodBinding, TypeBinding, TypeReference[]) line: 132	
QualifiedAllocationExpression.generateCode(BlockScope, CodeStream, boolean) line: 220	
LocalDeclaration.generateCode(BlockScope, CodeStream) line: 174	
ConstructorDeclaration.internalGenerateCode(ClassScope, ClassFile) line: 459	
ConstructorDeclaration.generateCode(ClassScope, ClassFile) line: 310	
TypeDeclaration.generateCode(ClassFile) line: 756	
TypeDeclaration.generateCode(CompilationUnitScope) line: 826	
CompilationUnitDeclaration.generateCode() line: 412	
CompilationUnitProblemFinder(Compiler).resolve(CompilationUnitDeclaration, ICompilationUnit, boolean, boolean, boolean) line: 1074	
CompilationUnitProblemFinder.process(CompilationUnit, SourceElementParser, WorkingCopyOwner, Map<String,CategorizedProblem[]>, boolean, int, IProgressMonitor) line: 272	
CompilationUnit.buildStructure(OpenableElementInfo, IProgressMonitor, Map<IJavaElement,IElementInfo>, IResource) line: 186	
CompilationUnit(Openable).generateInfos(IElementInfo, Map<IJavaElement,IElementInfo>, IProgressMonitor) line: 245	
CompilationUnit(JavaElement).openWhenClosed(IElementInfo, boolean, IProgressMonitor) line: 585	
CompilationUnit.makeConsistent(int, boolean, int, Map<String,CategorizedProblem[]>, IProgressMonitor) line: 1137	
ReconcileWorkingCopyOperation.makeConsistent(CompilationUnit) line: 166	
ReconcileWorkingCopyOperation.executeOperation() line: 92	
ReconcileWorkingCopyOperation(JavaModelOperation).run(IProgressMonitor) line: 740	
ReconcileWorkingCopyOperation(JavaModelOperation).runOperation(IProgressMonitor) line: 805	
CompilationUnit.reconcile(int, int, WorkingCopyOwner, IProgressMonitor) line: 1311	
ASTConverterSuperAfterStatements(AbstractASTTests).buildASTs(int, String, ICompilationUnit, boolean, boolean, boolean) line: 464	
ASTConverterSuperAfterStatements(AbstractASTTests).buildASTs(String, ICompilationUnit, boolean, boolean, boolean) line: 530	
ASTConverterSuperAfterStatements(AbstractASTTests).buildAST(String, ICompilationUnit, boolean, boolean, boolean) line: 378	
ASTConverterSuperAfterStatements(AbstractASTTests).buildAST(String, ICompilationUnit, boolean) line: 369	
ASTConverterSuperAfterStatements(AbstractASTTests).buildAST(String, ICompilationUnit) line: 356	
ASTConverterSuperAfterStatements.test004() line: 224	

In short I think the warning is good and is pointing to a problem in the product.

[srikanth-sankaran]

In this call stack below codeStream.stackDepth becomes -1 which is illegal.

TypeAnnotationCodeStream(CodeStream).astore_1() line: 552	
TypeAnnotationCodeStream(CodeStream).store(LocalVariableBinding, boolean) line: 7528	
LocalDeclaration.generateCode(BlockScope, CodeStream) line: 182	
ConstructorDeclaration.internalGenerateCode(ClassScope, ClassFile) line: 459	
ConstructorDeclaration.generateCode(ClassScope, ClassFile) line: 310	
TypeDeclaration.generateCode(ClassFile) line: 756	
TypeDeclaration.generateCode(CompilationUnitScope) line: 826	
CompilationUnitDeclaration.generateCode() line: 412	

[srikanth-sankaran]

Manoj, please follow up. Stack traces above point to the problem spots.

[mpalat]

Manoj, please follow up. Stack traces above point to the problem spots.

Sure Srikanth.. let me take a look

[srikanth-sankaran]

I suggest you disable the test by prefixing with _ and merge a commit to master - otherwise we are going to see a failing tests on all PRs until this is resolved. Thanks. Not just jdt/core others will see this failure too and may be puzzled: see #2194

[stephan-herrmann]

Found you! To check why my PR build failed I incrementally merged commits from master into BETA_JAVA22. It was when I merged #2171 that this test started to fail. Looks like this is old news for you guys :)

On the way I was puzzled about the relation between BETA_JAVA22 and #2181: Neither EGit nor github seems to show any relation although reportedly the branch was merged, Egit shows it as if the entire branch BETA_JAVA22 was squashed into a single commit??
Anybody else seeing this, too?
Right when I was about to complain, I discovered that blame still works across this missing merge-link. How is that possible?

[srikanth-sankaran]

@jarthana could you answer this question please ? Thanks.

[stephan-herrmann]

On the way I was puzzled about the relation between BETA_JAVA22 and #2181: Neither EGit nor github seems to show any relation although reportedly the branch was merged, Egit shows it as if the entire branch BETA_JAVA22 was squashed into a single commit??

Never mind, my EGit history view had "Show first parent only" enabled (don't know why). Disabling that option revealed the missing link:

Sorry for the noise.