On Thu, 15 Oct 2026 11:32:24 GMT, Alan Bateman <[email protected]> wrote:
>> Remove an old hack that prevented binding to 127.x.x.255 addresses on Linux. >> The hack was intended to work around a bug in the JCK that assumed that >> 127.x.x.255 addresses are used for broadcasting and must not be bindable. >> >> Added 2 new tests that fail on Linux without the fix, pass with the fix. >> Tier1, tier2, JCK java.net and java.nio tests continue to pass. >> >> --------- >> - [x] I confirm that I make this contribution in accordance with the >> [OpenJDK Interim AI Policy](https://openjdk.org/legal/ai). > > test/jdk/java/nio/channels/DatagramChannel/BindLoopback.java line 51: > >> 49: } >> 50: try (DatagramChannel channel = >> DatagramChannel.open(StandardProtocolFamily.INET)) { >> 51: channel.bind(new InetSocketAddress("127.0.0.255", 0)); > > The test binds to a very specific loopback address. So I think we need to > find a better name for the test, or beef it up to test a number of different > loopback addresses (both IPv4 and IPv6) That's a good point. I'd rather not extend the test, especially since it's only running on Linux (it's failing on Mac, and I suppose it may fail on other platforms too). Would B8394580 be an acceptable name for this test, or are we no longer using this kind of names? ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/33378#discussion_r4283034831
