I've added testing exceptions to IPv6 as well, PTAL. This required hardcoding IPv4 and IPv6 loopback addresses for the IPv4 and IPv6 code paths. Else DatagramChannel.join() fails since you can't mix IPv4 and IPv6 addresses, and InetAddress.getLocalHost() can only be one of those.
http://cr.openjdk.java.net/~aeubanks/8224019/webrev.01/ On Thu, May 16, 2019 at 2:52 AM Daniel Fuchs <daniel.fu...@oracle.com> wrote: > Hi Arthur, > > Shouldn't there be a version of > 221 exceptionTests(nif); > that works with IPv6? > > If no machine has IPv4 available, then these exceptions > won't be tested any more. > > I suspect the test made the assumption that IPv4 was always available, > and therefore testing exceptions with IPv4 addresses was sufficient. > Now if IPv4 is not available, then maybe those exceptional > cases should be tested with IPv6 instead? > > (Note: you can log a follow-up issue if you don't want to tackle > that immediately) > > best regards, > > -- daniel > > > On 16/05/2019 01:45, Arthur Eubanks wrote: > > bug: https://bugs.openjdk.java.net/browse/JDK-8224019 > > webrev: > http://cr.openjdk.java.net/~aeubanks/8224019/webrev.00/index.html > > > > test/jdk/java/nio/channels/DatagramChannel/BasicMulticastTests.java > > assumes NetworkConfiguration.probe().ip4MulticastInterfaces().iterator() > > always has at least one NetworkInterface. Instead it should do something > > like the IPv6 code right after it where it checks that the iterator is > > not empty. > >
