From: Pekka Savola (pekkas_at_no.spam)
Date: Fri Nov 16 2001 - 23:23:26 PST
On Sat, 17 Nov 2001, Myung-Ki Shin wrote:
> > This does not discuss the mess with network daemons and such in the
> > multiprotocol environment. That is, should one only use AF_INET6 and let
> > IPv4 users connect using mapped addresses or what?
>
> The scope of our draft is to provide guidelines for
> application transition (IPv4 -> IPv6) under Ngtrans mechanisms.
>
> I think 'the mess with network daemons and such in the
> multiprotocol environment' is another issue.
Perhaps 'application' would have to defined more explicitly then. It can
mean all programs run on a node, including server applications.
> > Also, there's much more to it to 4.3 than using getaddrinfo for getting
> > the names.. That is, using sockaddr_storage and friends because you
> > cannot know how much space you have to allocate for the addresses,
> > sockaddr_in or sockaddr_in6.
>
> You're right, we knew that, but the reason why we didn't mention
> "sockaddr_storage" is
> that's out of our subject.
Well, I'm not sure about the subject then; is it not an application
writer's problem to code an IPv4/IPv6 compatible code? You already showed
how to use getaddrinfo properly. The application writer has now to also
deal with the results of getaddrinfo; certain changes to the code are
required.
IMO, perhaps getaddrinfo etc. should be moved to an appendix and tagged
with a disclaimer like "there are also several other issues when dealing
with IPv4/IPv6 capable software: ..., ..., ..." or expanded a bit possibly
like above.
-- Pekka Savola "Tell me of difficulties surmounted, Netcore Oy not those you stumble over and fall" Systems. Networks. Security. -- Robert Jordan: A Crown of Swords
This archive was generated by hypermail 2.1.7 : Fri Oct 06 2006 - 00:00:26 PDT