From: Andy Clement <aclement@pivotal.io>
Date: Wed, 27 Sep 2017 21:15:38 +0000 (-0700)
Subject: Added commentary on a strange bit of code
X-Git-Tag: V1_9_0_RC3~2^2~60
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=d7cb30379a1da09c2a60a906816f6f6ba29ee103;p=aspectj.git

Added commentary on a strange bit of code

I don't want to change that code until I see more evidence of it
misbehaving but I'm having trouble working out why it is there!
---

diff --git a/org.aspectj.matcher/src/org/aspectj/weaver/ReferenceType.java b/org.aspectj.matcher/src/org/aspectj/weaver/ReferenceType.java
index 65fdf3a95..7dc162a16 100644
--- a/org.aspectj.matcher/src/org/aspectj/weaver/ReferenceType.java
+++ b/org.aspectj.matcher/src/org/aspectj/weaver/ReferenceType.java
@@ -376,10 +376,16 @@ public class ReferenceType extends ResolvedType {
 		if (this.isFinal() || other.isFinal()) {
 			return false;
 		}
+		
+		// 20170927: What is the block of code for? It mentions jls5.5 which isn't on this topic (old version of jls?)
+		// Some possible references: http://docs.oracle.com/javase/specs/jls/se9/jls9.pdf  5.1.6 (narrowing reference conversion)
+		// On Java 9 the test GenericsTests.testAfterReturningWithWildcardVar will fail because this code below
+		// used to find Set and List were the same, but now finds they are not. (so it doesn't put out the unchecked
+		// conversion message). However the code "List l = (List)someSet;" still compiles on 9 - so is this code bogus?
+		
 		// ??? needs to be Methods, not just declared methods? JLS 5.5 unclear
 		ResolvedMember[] a = getDeclaredMethods();
-		ResolvedMember[] b = other.getDeclaredMethods(); // ??? is this cast
-		// always safe
+		ResolvedMember[] b = other.getDeclaredMethods();
 		for (int ai = 0, alen = a.length; ai < alen; ai++) {
 			for (int bi = 0, blen = b.length; bi < blen; bi++) {
 				if (!b[bi].isCompatibleWith(a[ai])) {