On Thursday 08 April 2010 11:05:30 Stephane Bortzmeyer wrote:
On Wed, Apr 07, 2010 at 08:20:37AM +0200,
Vincent Danjean <vdanjean.ml@free.fr> wrote
a message of 63 lines which said:
> I've no strong opinion about the default value for
> net.ipv6.bindv6only. However, I think that any application that
> breaks if the default value is 0 or 1 is broken and a bug must be
> filled..
You mean that applications should use the option IPV6_V6ONLY of RFC
3493, section 5.3, therefore not depending on the system-wide value?
Could be hard to convince the developers to do so, and in my opinion
distribution-specific patches should be used only when necessary, not as a
normal way to proceed.
Unfortunately i don't have access to the POSIX specification, but the
opengroup says that by default bindv6only is 0.
http://www.opengroup.org/onlinepubs/9699919799/functions/V2_chap02.html#tag_15_10_20_02
So basically the developers will assume this value and will probably reject
bugreports claiming that the assumption is wrong.