On 30 Jun 2015, at 06:56, Tim Rice <tim@xxxxxxxxxxxxxxxx> wrote:
> On Tue, 30 Jun 2015, Damien Miller wrote:
>
>> On Mon, 29 Jun 2015, Tim Rice wrote:
>>
>>> Seems a shame to disable the whole test.
>>> Is this too ugly?
>>
>> don't mind, but there were a couple of bugs (tests reverse, > instead of >>)
>
> Actually the test was correct. Skip or append IPv6 bits.
> But "> instead of >>" was a mistake.
> Oops.
Note that the proposed test tests whether IPv6 code can be
compiled, not whether it's supported or enabled on the host.
For instance, the following sysctl's will turn IPv6 support]
off under Linux:
net.ipv6.conf.all.disable_ipv6 = 1
net.ipv6.conf.default.disable_ipv6 = 1
net.ipv6.conf.lo.disable_ipv6 = 1
but the test would still be run. I believe OpenBSD would
perform similarly if the kernel were compiled without
IPv6 support.
I think the real test would involve running a tiny program
to check you can bind to an IPv6 loopback address as well
as checking compile time support.
--
Alex Bligh
_______________________________________________
openssh-unix-dev mailing list
openssh-unix-dev@xxxxxxxxxxx
https://lists.mindrot.org/mailman/listinfo/openssh-unix-dev
