RE: IPv6 and draft-petersson-forwarded-for

> -----Original Message-----
> From: Andreas Petersson [mailto:andreas@sbin.se]
> Sent: Tuesday, April 26, 2011 6:01 AM
> To: Dan Wing
> Cc: draft-petersson-forwarded-for@tools.ietf.org; ietf-http-wg@w3.org
> Subject: Re: IPv6 and draft-petersson-forwarded-for
> 
> On Mon, 25 Apr 2011 22:14:58 -0700
> "Dan Wing" <dwing@cisco.com> wrote:
> 
> > Andreas, Martin,
> >
> > The relevant ABNF to construct an IPv6 address is:
> >
> >        node = nodename [":" node-port]
> >        node-port = port | obfport
> >        nodename = IPv4address | IPv6address |
> >                    "unknown" | "hidden" | obfnode
> >        IPv6address = addr6 | "[" addr6 "]"
> >
> > which could result in something like 2001:db8::5555, which has an
> ambiguous
> > meaning -- it could be "2001:db8 port 5555", or simply the IPv6
> address
> > "2001:db8::5555" (without the node-port).
> >
> > The simplest fix is to require the "[]" characters around an IPv6
> address,
> > similar to how HTTP is constructed (RFC3986), like this:
> >        IPv6address = "[" addr6 "]"
> >
> > -d
> >
> >
> >
> 
> Hi,
> 
> Yes, [] around the address is required in the draft in textual form if
> port is present, this will be changed in the next version of the draft
> where [] will be required independent of the presence of port.
> 
> Beside that, should the BNF for IPv6 be borrowed from RFC3986?

I would borrow the BNF for both IPv4 and IPv6 from RFC3986, perhaps
with something like:

  The BNF rules for "IPv6address" and "IPv4address" are defined
  in [RFC3986].

> PHK suggested to use the RFC5952, I think that looks good but it lacks
> a BNF-description.

How about:

  The BNF rules for "IPv6address" and "IPv4address" are defined
  in [RFC3986].  The IPv6address SHOULD comply with textual
  representation recommendations [RFC5952] (e.g., lowercase,
  zero compression).
 
-d


> 
> rgds,
>  Andreas Petersson

Received on Tuesday, 26 April 2011 15:07:52 UTC