RFR: 8247463: [TESTBUG] java/nio/channels/etc/OpenAndConnect.java fails due to IPv6 not available
Daniel Fuchs
daniel.fuchs at oracle.com
Fri Jun 12 12:58:01 UTC 2020
Hi Jie,
One more step :-)
It looks to me as if several cases are misplaced. For instance:
159 { INET6, IA6ANYLOCAL, null, IA6LOOPBACK },
160 { INET6, IA6ANYLOCAL, INET6, IA6LOOPBACK },
162 { null, IA6ANYLOCAL, null, IA6LOOPBACK },
164 { null, IA6ANYLOCAL, INET6, IA6LOOPBACK },
181 { null, IA4LOOPBACK, INET6, IA6ANYLOCAL },
182 { null, IA6LOOPBACK, INET6, IA6ANYLOCAL },
183 { null, IA6LOOPBACK, null, IA6ANYLOCAL },
184 { null, IA4LOOPBACK, null, IA6ANYLOCAL }
don't require that IA6LOCAL is configured. So they could be tested
even if IA6LOCAL is null.
best regards,
-- daniel
On 12/06/2020 13:34, jiefu(傅杰) wrote:
> Hi Daniel,
>
> Thanks for your review and nice comments.
>
> Updated: http://cr.openjdk.java.net/~jiefu/8247463/webrev.01/
>
> Please review it and give me some advice.
>
> Thanks a lot.
> Best regards,
> Jie
>
> On 2020/6/12, 6:21 PM, "Daniel Fuchs" <daniel.fuchs at oracle.com> wrote:
>
> Hi Jie,
>
> This sounds reasonable to me. However - I wonder if a better way
> to look at things would be to use IPSupport.hasIPv4() and
> IPSupport.hasIPv6() - in conjunction with IA4LOCAL == null and
> IA6LOCAL == null to decide what test case is appropriate to
> run.
>
> AFAIU, a machine may have no IPv6 local address configured (and
> IA6LOCAL would be null), but might still have the IPv6 stack
> and be able to use the IPv6 loopback.
>
> So I'd prefer if a finer grained strategy was used to maximize the
> number of tests that are performed on a given machine.
>
> best regards,
>
> -- daniel
>
>
> On 12/06/2020 08:14, jiefu(傅杰) wrote:
> > Hi all,
> >
> > The test always fails in some of our CI/CD which do not support IPv6.
> >
> > It would be better to fix it.
> >
> > JBS: https://bugs.openjdk.java.net/browse/JDK-8247463
> >
> > Webrev: http://cr.openjdk.java.net/~jiefu/8247463/webrev.00/
> >
> > Thanks a lot.
> >
> > Best regards,
> >
> > Jie
> >
>
>
>
>
More information about the nio-dev
mailing list