Am 12.01.15 um 16:03 schrieb Bertrand Delacretaz: > On Mon, Jan 12, 2015 at 3:34 PM, Robert Munteanu <romb...@apache.org> wrote: >> ...Without looking at the test, I think it's relying on the ordering >> maintained by a HashSet, which has changed with Java 8, and is of >> course nothing to rely on... > > It's something like that but in the actual code, not in the test - > looking at it under SLING-4299 >
I think the test is wrong, the impl shouldn't be required to use an ordered set (unless I'm overlooking something). Carsten -- Carsten Ziegeler Adobe Research Switzerland cziege...@apache.org