Re: [netmod] IETF91 NETMOD agenda ?

On Thu, Nov 6, 2014 at 2:47 AM, t.petch <ietfc@btconnect.com> wrote:
> ----- Original Message -----
> From: "Ladislav Lhotka" <lhotka@nic.cz>
> To: "Andy Bierman" <andy@yumaworks.com>; "Benoit Claise"
> <bclaise@cisco.com>
> Cc: <netmod@ietf.org>; <rtg-ads@tools.ietf.org>
> Sent: Wednesday, November 05, 2014 8:40 AM
> Subject: Re: [netmod] IETF91 NETMOD agenda ?
>
>
>> Hi,
>>
>> I agree with Andy. I think we already have a LOT to do with the
> existing
>> WG items, and YANG 1.1 in particular, and we should devote most of the
>> WG meeting time to them.
>
> Disagree
>
> What you do now with YANG 1.1 will not hit the streets for a while; the
> charter says submit to the IESG in March 2015 and I have never known
> something like this be early.  And then it will be a while before the
> implementations catch up and then more before the designers of models
> get there.
>
> Benoit just listed some 70 YANG I-Ds and it is safe to assume that
> within those will be many misunderstandings, bad design choices and
> plain errors, and once they make it into RFC, they will be with us for
> years, if not decades, reminding us of how not to do things.
>

So the NETMOD WG should not work on its main chartered item because
there are 70 unchartered drafts that happen to contain YANG modules?
That doesn't seem very focused.

The bad YANG will get fixed.  I remember lots of MIB modules were submitted
that did not even compile. This is not a work item for the NETMOD WG.
For the modules that actually make it to WG draft, they will get
fixed before they are published.

> The priority here and now should be
>
>> 2. Get advice about YANG-related aspects.
>
> in the shape of guidelines (or advice to the YANG Directorate) of what
> to do and how to do it.  Even if the Directorate do pick up the worst
> practices, it would be still more productive for them not to be made in
> the first place, by the protocol experts in the other WGs who are
> writing the models.
>
> Yes, YANG 1.1 will fix or forestall some of this but for most, it will
> be too late.

You think the IETF is going to work through the 70 draft queue by March?
It's not that hard to write a reasonable YANG module.  It is much harder
to get all the vendors to agree on the protocol knobs in the module.

>
> I speak from a few decades of experience with SNMP (a much simpler data
> modelling language:-).
>
> Tom Petch

Andy

>                                                      Offloading this
> work to interim telcos is IMO
>> not good.
>>
>> In my view, authors of new YANG modules face three challenges:
>>
>> 1. Organize a group of domain experts to work on the substance of the
>> module (scope, configuration logic, dealing with the legacy of
> existing
>> CLIs etc.).
>>
>> 2. Get advice about YANG-related aspects.
>>
>> 3. Find a home for the module so that it can eventually become an RFC.
>>
>> For the most part, #1 and #3 should be done outside NETMOD in an
>> appropriate IETF area. It involves convincing domain experts that the
>> work is worth doing in the first place, and then find enough support
> and
>> capacity to carry it out. But this is I guess IETF business as
>> usual. NETMOD WG cannot act as a shortcut for this hard part. And
>> judging from the recently announced guidelines, I think the routing
> area
>> got it pretty much right.
>>
>> There might be modules that need to be done in the NETMOD WG, e.g. if
>> they span multiple areas, but they should be an exception.
>>
>> As for #2, I think it is a task for YANG doctors. In this case, I'd be
>> willing to accept stricter rules and more deterministic workflow so
> that
>> module autors receive feedback in a timely manner.
>>
>> Lada
>>
>> Andy Bierman <andy@yumaworks.com> writes:
>>
>> >>
>> >
>> > It is not NIMBY. The YANG experts are not all SYSLOG experts.
>> > Getting the right people in the room is important (you said that ;-)
>> > Making a few SYSLOG experts sit through 140 minutes of
>> > unrelated material may not be the best option for them either.
>> >
>> > It is unfortunate that the SYSLOG WG has decided the protocol
>> > is complete and no further work is needed.  I don't agree IETF
>> > process should be a good reason to choose NETMOD WG.
>> > This draft is already chartered and I am not trying to stop
>> > the SYSLOG work.  I don't want it to start a trend though.
>> >
>>
>> --
>> Ladislav Lhotka, CZ.NIC Labs
>> PGP Key ID: E74E8C0C
>>
>> _______________________________________________
>> netmod mailing list
>> netmod@ietf.org
>> https://www.ietf.org/mailman/listinfo/netmod
>