Javac Compiler Bug...

gistfile1.txt

Potential bug in JavaC.  Surely the line should be:

    annotationProcessingOccurred = c.annotationProcessingOccurred == true;

and do a comparison instead of multi-assignment ( and only assign it if annotationProcessingOccurred is current false, as I don't think we want to reassign a true setting to false here).

Open JDK 6:

http://grepcode.com/file/repository.grepcode.com/java/root/jdk/openjdk/6-b14/com/sun/tools/javac/main/JavaCompiler.java#JavaCompiler

995  JavaCompiler c = procEnvImpl.doProcessing(context, roots, classSymbols, pckSymbols);
996  if (c != this)
997      annotationProcessingOccurred = c.annotationProcessingOccurred = true;
998  return c;

Open JDK 7

http://grepcode.com/file/repository.grepcode.com/java/root/jdk/openjdk/7-b147/com/sun/tools/javac/main/JavaCompiler.java?av=f

1106  JavaCompiler c = procEnvImpl.doProcessing(context, roots, classSymbols, pckSymbols);
1107  if (c != this)
1108      annotationProcessingOccurred = c.annotationProcessingOccurred = true;
1109  // doProcessing will have handled deferred diagnostics
1110  Assert.check(c.log.deferDiagnostics == false
1111          && c.log.deferredDiagnostics.size() == 0);
1112  return c;'

The later is my current guess after thinking about it more, of course - I don't know WHY you'd have multiple instances of a javaCompiler so this might not actually be a problem. I have an odd thought that when annotation processing occurs, another javaCompiler instance is possibly created to compile any generated sources, which would mean the code here is probably correct - but kinda smelly code.