Re: [dnsext] we need help to make names the same, was draft-yao-dnsext-identical-resolution-02 comment

[Tony Finch <[e-mail]>]

On Thu, 17 Feb 2011, Mark Andrews wrote:
>
> This is where if CNAME's were being properly handled you wouldn't have a
> issue.  Give the service the indirection support and aliasing support.
> HTTP uses the alias record for indirection which is just wrong.

It's worth noting that email moved away from doing client-side CNAME-based
redirection.

But you are right that it is better to have protocol support.

SMTP has vestigial client redirects (251 and 551 server responses), though
no-one ever implemented them. Many sites like to rewrite message headers
to canonicalize email addresses based on some idea of friendliness e.g.
[e-mail] -> [e-mail]. I generally think this is
pointless since (in the absence of 251 and 551 support) it doesn't have
any useful effect on the address books of people who are using the
non-canonical aliases. However since email is mostly open-loop, the
complexity of a site's aliases don't need to be exposed to their
correspondents.

XMPP presence handling is a more awkward example: last time I looked the
protocol doesn't have a way to redirect a presence subscription request
from one JID to another (e.g. from an alias to the JID where you actually
advertise your presence). The only way to handle aliases is on the client
side as a multiplicity of separate accounts. It matters much more to your
IM buddies that they know the JID(s) you currently use.

Tony.
-- 
f.anthony.n.finch  <[e-mail]>  http://dotat.at/
Viking, North Utsire, South Utsire, Northeast Forties: Easterly or
southeasterly 5 to 7, decreasing 4 at times in North Utsire and Viking. Rough
or very rough, but moderate in North Utsire. Rain or snow. Moderate or poor.
_______________________________________________
dnsext mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/dnsext