Re: [netmod] IP addresses with zone indices

[Ladislav Lhotka <[e-mail]>]

On Dec 10, 2012, at 6:14 PM, Andy Bierman <[e-mail]> wrote:

> 
> 
> On Mon, Dec 10, 2012 at 9:02 AM, Juergen Schoenwaelder <[e-mail]> wrote:
> On Mon, Dec 10, 2012 at 08:34:57AM -0800, Andy Bierman wrote:
> >
> > A: deploying standards is very expensive.  Temporary fixes last 10 - 20
> > years around here.
> > There is no such thing.  Get it right the first time or pay the price for
> > many years to come.
> >
> > B: There happens to be a 6021bis draft:
> > http://www.ietf.org/id/draft-schoenw-netmod-rfc6021-bis-00.txt
> > It adds 'hex-string' and 'uuid' to ietf-yang-types.yang.
> > (A 'Changes Since RFC 6021' section would make that more clear.)
> >
> > There are many useful typedefs in the common-types.yang module in
> > the proposed ACL draft.
> > http://www.ietf.org/id/draft-huang-netmod-acl-01.txt
> >
> > I don't know how practical it is to re-open 6021 every time the WG thinks of
> > a new typedef to add.  But it seems easier than chartering and publishing
> > new work,
> > and it is better for YANG developers to keep the standard types in a lot
> > of random modules.
> >
> > I don't agree with any of the 4 options above. I prefer:
> >
> > 5) work on 6021bis for a short time (1 - 2 months) and
> > publish reusable data types only in 1 place (6021bis).
> 
> The question is whether 1-2 months is realistic. It will likely be 6
> months in the WG. The devil is in the details and some of the details
> tend to only pop up at WG last calls. So are we fine with having the
> IP and routing data models sitting in the RFC queue for whatever time
> it takes to revise, approve, publish 6021 (in my calculation this is
> more likely 6 months plus IESG processing plus RFC queue time, so 8-10
> months)?
> 
> 
> OK -- it is not realistic if common-types.yang is considered,
> although IMO that module has some very useful datatypes
> based on other standards (so it's either right or it's wrong,
> but it isn't purple vs. dark blue). 
> 
> Adding 3 or 4 types (and 2 are done already) better not take 6 months.
> That would indicate a major process failure by the WG and its chairs.
> It's the translation of a complex operational model into a data model
> that takes all the time.  It 's the delay itself -- it promotes endless
> late comments and rewrites.  We need to learn how to do a Code Sprint
> for I-Ds.

The need for new types arises (naturally) during module development, often quite late in the process, so I think the only option in most cases is to define the type right in the module which needs it.

If such a type is found applicable in other modules, it might be useful to move it later to a type library. We need to figure out a reliable procedure for performing this migration, within the existing constraints of the IETF workflow. This procedure must not put ongoing development of modules on hold. 

Lada 

> 
> 
> /js
> 
> 
> Andy
> 
>  
> PS: My record so far has been RFC 4789 which took about 6 months from
>     initial I-D to RFC published. This was AD sponsored (no WG to
>     reach concensus) and a topic most people find kind of esoteric.
>     Hence I am questioning the 1-2 months unless we already limit the
>     scope of additions to consider for 6021bis.
> 
> --
> Juergen Schoenwaelder           Jacobs University Bremen gGmbH
> Phone: [phone]         Campus Ring 1, 28759 Bremen, Germany
> Fax:   [phone]         <http://www.jacobs-university.de/>
> 
> _______________________________________________
> netmod mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/netmod

--
Ladislav Lhotka, CZ.NIC Labs
PGP Key ID: E74E8C0C