On Jul 24, 2014, at 12:12 AM, Mike Duigou <mike.duigou@oracle.com> wrote:
Hello all;
This change fixes an issue with the instance returned by Collections.synchronizedNavigableSet().tailSet(Object,boolean) which synchronizes on wrong object, itself rather than the same mutex as it's source.
jbsbug: https://bugs.openjdk.java.net/browse/JDK-8048209 webrev: http://cr.openjdk.java.net/~mduigou/JDK-8048209/0/webrev/
Since I was concerned that there might be other lurking errors I opted to create a fairly thorough new test which, thankfully, revealed no additional problems.
Looks good, i like the test re-structuring, just some minor comments on that. - there are some commented out asserts. - take or leave the following... the providers could be simplified a little if you return a Stream rather than a Collection for those make methods: Supplier<..> s = () -> Collections.synchronizedSortedSet(new TreeSet<>(BASE_COLLECTION)); Stream.of(s.get(), s.get().headSet("Foxtrot"), ...) map(c -> new Object[] {c}) and also use Stream.concat. Paul.