Margaret Wasserman wrote:

"I don't know. A year ago, I would have said "no", but with documents on the IESG agenda like draft-black-snmp-uri-08.txt, I am taking a new view of URIs. This document defines a URI syntax that can be used for access to SNMP objects, and one of the use cases includes having a local SNMP manager that is accessed through a URI -- the URI then being translated into an SNMP request that is sent to the agent via SNMP.

So, URIs can, effectively, be used as a programmatic interface to other applications running on the local system. In that sort of case, I think we very well might need to include scoped addresses in URIs. And, IMO, it would be better to have a standard way to do this than to live in a world where folks just insert an unencoded % and some parsers pass it through cleanly, others escape it for you and still others return an error -- which is how the current behaviour of URI parsers has been described in this thread."

I agree that URIs with literal addresses and zoneids could be made to work within the local system.


But, since the zoneid is a local id with no global meaning, having such URIs leak outside the local system would potentially cause less things to work.
For example,
machine A assigns zoneid 1 to the one ethernet and zoneid 2 to
another ethernet for the like-local zones


        machine B has only on NIC and assigs zoneid 1 to it.

        The NIC on B is on the Ethernet which attaches to the zoneid 2
        NIC on A.

Without zonids in the URL, B can handle a URL with a literal link-local IPv6 address generated by A, because it only has one link-local interface which happens to connect to A.

But if you where to put zoneids in the URLs, then in this case B would see a URL with a literal link-local IPv6 address *and* zoneid=2. Would B then conclude that it can't talk to it because the zoneid doesn't match?


*If* there is utility of such URLs on the local system, wouldn't we need to say that a zoneid in a URL should be ignored except if the IP address is indeed an IP address of the host itself?


    Erik



--------------------------------------------------------------------
IETF IPv6 working group mailing list
[EMAIL PROTECTED]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to