Community
Participate
Working Groups
Source version v_725. The following extract shows that the second test against Binding.NO_TYPE_VARIABLES is extraneous, assuming that we are exempt from multi-threading side effects. public MethodBinding findExactMethod( ... MethodBinding exactMethod = ... if (exactMethod != null && exactMethod.typeVariables == Binding.NO_TYPE_VARIABLES && !exactMethod.isBridge()) { ... if (exactMethod.typeVariables != Binding.NO_TYPE_VARIABLES || invocationSite.genericTypeArguments() != null) { ... } return null; } A variant happens within the block. I believe that this is a consequence of a constants rationalization effort that replaced a code in which tests yielding the same results looked different. Now they are clearly redundant. This is a follow-up on bug 164094 that pointed that code area out for another reason.
It is in Scope#findExactMethod indeed.
Created attachment 54769 [details] Fix
Released for 3.3 M4.
Verified for 3.3M4 with I20061211-1119