On Fri, Mar 23, 2007 at 03:06:39PM +0100, Sjoerd Simons wrote:
> > Obviously, it's not reasonable to enable IPv6 support at the *expense of*
> > IPv4 support.

> Uhm, by default on linux if you listen on the IPv6 wildcard address, you'll 
> get
> IPv4 automagically too. For example my output for ssh is the same:

> $ lsof -p 16733 | grep -i listen
> sshd    16733 root    3u  IPv6 5141804            TCP *:ssh (LISTEN)

> And i can assure you, that works with ipv4 too :)

Right, I had this explained to me after the fact.

> > The IPv6 support in jabber seems to need a total rework.  In the meantime,
> > I'll prepare an NMU to revert this change from the -3.1 NMU.

> I can agree with that, the code isn't really good. But thanks to the fact 
> that linux listens on both IPv4 and IPv6 on an IPv6 socket it doesn't break
> IPv4 compatibility.

> I just tested that you can even listening on multiple ipv4 addresses even 
> works fine (Not the IP protocol vs. the address):

> $ lsof -n -p 9152  | grep -i listen  
> jabberd 9152 jabber    7u  IPv6 6744306  TCP 127.0.0.1:xmpp-client (LISTEN)
> jabberd 9152 jabber    8u  IPv6 6744307  TCP 145.120.11.130:xmpp-client 
> (LISTEN)
> jabberd 9152 jabber   11u  IPv6 6744310  TCP *:xmpp-server (LISTEN)

> So imho the NMU which disables ipv6 should be reverted and probably an 
> modprobe
> ipv6 added to the init script.

On a normal Debian system, the ipv6 module is auto-loaded for you whether
you want it or not.  AFAICS, the reason for the reported failures is that
users were running unofficial kernels without ipv6 support.  You can't fix
that with a modprobe, certainly.

So I think the NMU reverting this change is still the correct course of
action for etch, because:

- enabling the ipv6 support does cause a regression in support for ipv4
  systems, so is not a change that should be made during a freeze
- the change was made in an NMU, so again, regressions are not appropriate
  unless the changes have been specifically ok'ed by the maintainer.

I think this would be a perfectly reasonable change to reintroduce at the
beginning of the lenny cycle, preferably with discussion on debian-devel on
the appropriate way to handle ipv6 support (which might, e.g., result in
someone offering a better patch).

Thanks,
-- 
Steve Langasek                   Give me a lever long enough and a Free OS
Debian Developer                   to set it on, and I can move the world.
[EMAIL PROTECTED]                                   http://www.debian.org/


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to