Community
Participate
Working Groups
Build ID: M20080911-1700 Steps To Reproduce: class Test { Object foo(Object o) { return new Object() { /** @see #o */ void x() {} }; } }
Problem was that, in this peculiar case, reference to #o is valid for the compiler but javadoc resolution try to report a problem. This case (fieldBinding.problemId()==0) was not expected in ProblemReporter.javadocInvalidField(...) method, hence got the missing compiler implementation error!
Depending on the fix, we may want to backport it to 3.4.x. This kind of error is quite bad. Is this a regression ?
Created attachment 118039 [details] Proposed patch Patch is quite simple but this is not
... this is not a regression. It was already there in 3.3.2 and I guess it should be the same for previous releases as there was no change in this area for a long time...
Created attachment 118571 [details] New proposed patch Better patch which creates a problem field binding instead of tolerating NoError while reporting the problem...
Released for 3.5M4 in HEAD stream.
Verified for 3.5M4 using I20081209-0100