Re: [Fwd: Re: About IPv6 private address]

[bill fumerola <billf@mu.org>]

On Tue, Feb 05, 2008 at 04:54:56PM -0800, james woodyatt wrote:
> On Feb 5, 2008, at 15:06, bill fumerola wrote:
> >$ host lurgee.local
> >Host lurgee.local not found: 3(NXDOMAIN)
> 
> This could be regarded as a bug, but it's a tricky case.  The /usr/ 
> bin/host command isn't using the system name resolver.  It's  
> distributed with BIND, so it's basically doing the same thing that / 
> usr/bin/dig is doing.  It makes DNS queries only.  Multicast DNS is  
> derived from DNS, but it's not strictly the same as DNS, so /usr/bin/ 
> host doesn't know anything about it.

yep. i used both a command that uses the system's resolver library and
one that uses standard dns to parallel my question about just how
widespread microsoft's adoption of ipv6-literal.net. e.g. just where
does the 'interception' occur? i don't know how many different versions
of gethostbyname() microsoft has in their libraries. then again, i
shouldn't have to.

much like a non-existant .local nameserver can't resolve anyone's requests
of .local, ipv6-literal.net cannot resolve anyone's requests of resolution
of a lladdr. except requests are going to get further than root servers
and now microsoft has partially reinvented AS112 to deal with it. *sigh*

> At least, the .local domain has a long and ignoble history of not  
> being usable as a global TLD in the DNS protocol.  Apple was already  
> doing something weird and non-standard with it back in the Mac OS 9  
> days, when the DNS protocol was young and dinosaurs yet roamed the  
> Internet, so it's not like they polluted an otherwise pure water  
> table by repurposing it for Multicast DNS.

exactly. ipv6-literal.net is the worst of both worlds.

> Still, IETF does have a problem.  There are *two* Multicast DNS  
> protocols.  There is RFC 4795 (informational) for Link-Local  
> Multicast Name Resolution, which is implemented in Microsoft products  
> but not much else.  And there is <http://tools.ietf.org/html/draft- 
> cheshire-dnsext-multicastdns>, the Multicast DNS protocol in Bonjour,  
> which enjoys widespread industry adoption, in multiple  
> implementations, currently shipping in a myriad of commercial  
> products, but has been languishing as an under-loved individual  
> submission to the DNSEXT working group since before the Great Cataclysm.

[ here is where i get off-topic for this WG, reply-to set. ]

there can be no argument of the widespread adoption of
cheshire-dnsext-multicastdns in implementations, applications, and
userbase. given that, are there political landminds associated with
trying to updating the draft (at minimum, it's expired) and working with
DNSEXT to pick it up as a WG item? cheshire-dnsext-dns-sd as well?

time to go read the history in the archives, i guess.

-- 
- bill fumerola / billf@FreeBSD.org