Proposed changes to aut-num policy syntax in RPSLng
Andrei Robachevsky has proposed the following changes to the syntax for policy expressions in RPSLng based on feedback from the list and a desire to simplify the syntax for the case where v4/v6 and/or unicast/multicast policy are converged. 1) the absence of safi in an expression means that both unicast and multicast are defined. I.e., afi ipv4 == afi ipv4.unicast,ipv4.multicast 2) The afi specification will be optional in the mp-* attributes in the aut-num class. If no afi is specified, the policy shall be expanded to include all afi and safi types, namely -- afi ipv4.unicast,ipv4.multicast,ipv6.unicast,ipv6.multicast 3) introduce a pseudo afi with the value of "any". I.e., afi any.unicast == afi ipv4.unicast,ipv6.unicast afi any == afi ipv4.unicast,ipv6.unicast,ipv4.multicast,ipv6.multicast Looking forward to your feedback. Regards, Larry
I have not yet received any comments on these changes. I've gone ahead and produced a new Internet Draft based on the text below. It's currently available at http://www.radb.net/rpslng-03.html and http://www.radb.net/rpslng-03.txt The IETF Secretariat is not accepting new drafts until March 1. -Larry On Fri, 2004-02-20 at 16:15, Larry J. Blunk wrote:
Andrei Robachevsky has proposed the following changes to the syntax for policy expressions in RPSLng based on feedback from the list and a desire to simplify the syntax for the case where v4/v6 and/or unicast/multicast policy are converged.
1) the absence of safi in an expression means that both unicast and multicast are defined. I.e.,
afi ipv4 == afi ipv4.unicast,ipv4.multicast
2) The afi specification will be optional in the mp-* attributes in the aut-num class. If no afi is specified, the policy shall be expanded to include all afi and safi types, namely --
afi ipv4.unicast,ipv4.multicast,ipv6.unicast,ipv6.multicast
3) introduce a pseudo afi with the value of "any". I.e.,
afi any.unicast == afi ipv4.unicast,ipv6.unicast afi any == afi ipv4.unicast,ipv6.unicast,ipv4.multicast,ipv6.multicast
Looking forward to your feedback.
Regards, Larry
Hi, I want to check if there exist RPSL -> XML converter utility/program/scripts written by somebody. if no, are there any specific guidelines or things I need to be know about if I'm to write one myself? Related are there plans to develop RPSL futher in the future and to create compatibility with CRISP/IRIS (again this would be RPRL <> XML conversion scheme question I guess). --- William Leibzon Elan Networks william@elan.net
william(at)elan.net wrote:
Hi,
I want to check if there exist RPSL -> XML converter utility/program/scripts written by somebody. if no, are there any specific guidelines or things I need to be know about if I'm to write one myself?
Related are there plans to develop RPSL futher in the future and to create compatibility with CRISP/IRIS (again this would be RPRL <> XML conversion scheme question I guess).
Not that I know of, but it does make sense. My take on it: - the domain name people have a serious problem with Whois, because it doesn't meet their needs - the addressing people have a serious problem with Whois, because it doesn't meet their needs - the routing people are just fine :) CRISP will offer referrals (not needed for most routing registries, because they are highly mirrored), standard schema (not needed because routing is all done with RPSL), authentication (not needed because RPSL is typically public), and standard query mechanisms (not needed because there are only a few real RPSL server implementations). I think eventually CRISP will have to address routing policy, but I see it as a lower priority than for dreg and areg folks. -- Shane Kerr RIPE NCC
participants (3)
-
Larry J. Blunk -
Shane Kerr -
william(at)elan.net