|
Martin edited on 2006-07-18 23:05:23.0
|
|
I could fix it for the given case, and in all cases where both superclass and inheriting trait are Scala classes. However, I noted that the `protected' gets eliminated in the javap output. Iulian, is the backend doing that, and under what circumstances? My problem is that if the `protected' is *not* eliminated then we still might have a problem with traits that inherit from Java classes.
|
|
Martin edited on 2006-07-18 23:34:31.0
|
|
OK, I verified that it does work if Test is a Java class with a protected. So the bug is closed now. However, I
still have two questions:
1. Why is the backend eliminating `protected' modifiers?
2. It seems the JVM is not enforcing the protected access rule that the access must happen from within a class that is a subclass of the protected function. Note that the access
to `foo' happens in class Test2$class which is not a
subclass of Test.
|
|
Martin edited on 2006-07-24 16:38:08.0
|
|
Iulian indicated that the JRE does indeed throw a verify error. I leave the bug open as a reminder.
|
|
Martin edited on 2006-08-22 16:31:32.0
|
|
|
|
Iulian edited on 2007-03-01 10:46:37.0
|
|
Seems to have been fixed by some other bug fix.
|