If we fixed the firewall,
if we fixed the RPC protocol thereto,
if we fixed the local DNS APIs and all the documentation thereto,
if we fixed the config file parser in the DNS server,
if we fixed the GUI DNS record editor in the DNS server,
if we fixed the active directory parser in the DNS server,
and if this then got deployed broadly enough (the issue is not just
servers, but workstations as well: we've prototyped CallerID in Outlook
Express, for example, and there are scenarios in which that
implementation vehicle is compelling)
then yes, we might have a world where DNS RR types were extensible.
Bob
PS: I'm confused about your XML comment; can you elaborate?
-----Original Message-----
From: Eric A. Hall [mailto:ehall(_at_)ehsco(_dot_)com]
Sent: Friday, May 21, 2004 10:28 AM
To: Bob Atkinson
Cc: Arnt Gulbrandsen; Yakov Shafranovich; IETF MARID WG
Subject: Re: suggested new RRtype experiment
If you'd fix your RPC-DNS gateway so that it allowed for new types,
you
wouldn't have these kinds of problems and could fully enjoy this and
all
other DNS efforts much easier.
It's becoming a bit clearer to me why you stick whole XML documents
into
CID records anyway.
On 5/21/2004 12:08 PM, Bob Atkinson wrote:
Thank you for your kind, polite, and constructive words. Now that
you've
framed your point of view in this way I value and understand it in a
way
that I had heretofore not appreciated.
Please also remember that this is not simply a DNS client issue, but
also a DNS server issue: Windows DNS server lacks extensible RR type
support as well, as I have been mentioning.
Bob
-----Original Message-----
From: Eric A. Hall [mailto:ehall(_at_)ehsco(_dot_)com]
Sent: Friday, May 21, 2004 10:06 AM
To: Bob Atkinson
Cc: Arnt Gulbrandsen; Yakov Shafranovich; IETF MARID WG
Subject: Re: suggested new RRtype experiment
I have a hard time believing that your network can't support:
routers
firwalls
print servers
macs
*nix
or any other system/device that speaks regular DNS over UDP 53.
If your network is really that borked then I'm not sure its worth
considering as a mainstream dependancy. It sounds like you guys have
chosen to effectively disable DNS, which means that we're talking
about
site-specific architecture issues.
--
Eric A. Hall
http://www.ehsco.com/
Internet Core Protocols
http://www.oreilly.com/catalog/coreprot/
--
Eric A. Hall
http://www.ehsco.com/
Internet Core Protocols
http://www.oreilly.com/catalog/coreprot/