Thank you for the information and proposals.
I’ve talked with Marc about the thinking during the ieee meeting and will further keep align with 802.1 on this.
For 802.3, the modification is better to be done before September meeting if possible : ). To facilitate that, we can discuss the modifications in the following adhoc calls.
Could you update the ethernet-interface/-legacy modules next month?
发件人: Robert Wilton [mailto:rwilton@xxxxxxxxx]
收件人: Zhuangyan (Yan); STDS-802-3-YANG@xxxxxxxxxxxxxxxxx
主题: Re: [802.3_YANG] Changes to 802.3 modules in light of IETF actions
ietf-interfaces will be revised to be NMDA compliant.
In particular I expect:
- config false nodes will be coped into the "interfaces" tree.
- the entire "interfaces-state" tree will be marked as deprecated.
There should hopefully be a new draft in the next month, and I will be pushing for this to move to WG adoption and last call quickly.
For our models:
I propose that we should:
(1) update the main models to only augment from "interfaces" and not "interfaces-state".
(2) also publish (in an appendix) a temporary "ethernet-state" module that is entirely config false and only augments "interfaces-state".
However, before we do this, we might want to discuss/confirm with 802.1 what our plans are.
On 20/07/2017 10:22, Zhuangyan (Yan) wrote:
Thanks for bringing this up.
Yes, IETF is encouraging modules to move to NMDA-style as stated in drafts below. We’ve discussed during the previous call when IETF brought this idea up, but no conclusion that
time. Given the progress in IETF, it would be good to get back to it. My suggestion is to make the modification before we go to D2.0.
The thing is, the ietf-interface module itself which most of our modules augments is not transferred to NMDA-style yet. Also, Robert is working on tools that can automatically change
the previous modules to NMDA-style ones which can ease our work. Otherwise, we can do it manually as well.
All (but mostly to Rob W),
It is my understanding that IETF recently took a decision to change how NETCONF/RESTCONF handles datastores for YANG models. New models no longer separate configuration and state trees and existing models should deprecate
state trees and move their contents to a combined configuration/state tree (see links copied below)
Do we need to make any modifications to our modules in light of the recent actions of IETF to combine configuration and state trees?
Here are relevant drafts:
* https://www.ietf.org/proceedings/99/slides/slides-99-netmod-sessa-nmda-guidelines-01.pdf (*)
FutureWei Technologies Inc.
Director, Access R&D