Re: [netmod] draft-clacla-netmod-yang-model-update-00.txt : Re: [RTG-DIR] handling module incompatibility => handling module transition

[Benoit Claise <[e-mail]>]

Replying to Jürgen, Andy, and Rob in this email.

Yes. Semantic versioning is a step in the right direction. Specifically 
for native/generated models (yes, those exist into the wild). And for 
"standard" models too. See 
https://tools.ietf.org/html/draft-clacla-netmod-yang-model-update-01#section-2.1

We'll have to come to the notion of release bundle, for sure.
The "leaf tree-type" (*) is a release bundle type of metadata anyway
I was hoping that the IETF would be releasing YANG modules in bundle, 
but looking at the rate, it seems more drop by drop. Anyway, all those 
routing modules should be working together. The YANG impact analysis 
should help in checking this. See 
https://www.yangcatalog.org/yang-search/impact_analysis.php?modules[]=ietf-routing&recurse=0&rfcs=1&show_subm=1&show_dir=both
Btw, the "dependent" and "dependencies" are also metadata in the 
yangcatalog.org

Just saying that YANG modules "are designed and known to work together 
into meaningful packages" is interesting.
However, in the end, operators combine/test YANG modules to deploy 
services. This is Rob"s feature bundle described below. Since this 
notion of feature bundle is service specific, for service composition, 
in the end, the feature bundle is a service-specific release bundle, so 
operator specific, so orchestrator/controller specific.

> I agree that semantic versioning is only part of the solution.
> In OpenConfig versioning we have the concept of release bundles that 
> have a semver, these contain modules that are known to work together - 
> and are the base for compliance descriptions.
> The individual modules semver has been useful to mark where there are 
> backwards incompatible changes in an individual module.
>
> On top of release bundles, we also have feature bundles which describe 
> a particular implementation's requirements for the modules. A feature 
> bundle is a description of paths within a release bundle.
>
> We're continuing to gain experience with how well this works, but 
> certainly I don't see the need for the bundles alleviating the need 
> for the semantic versions for modules.
+1.

It's a matter of determining which problem we want to tackle now.

(*) https://datatracker.ietf.org/doc/draft-clacla-netmod-model-catalog/

leaf tree-type {
         type enumeration {
           enum "split" {
             description
               "This module uses a split config/operational state layout.";
           }
           enum "nmda-compatible" {
             description
               "This module is compatible with the Network Management Datastores
                Architecture (NMDA) and combines config and operational state nodes.";
           }
           enum "transitional-extra" {
             description
               "This module is derived as a '-state' module to allow for transitioning
                to a full NMDA-compliant tree structure.";
           }
           enum "openconfig" {
             description
               "This module uses the Openconfig data element layout.";
           }
           enum "unclassified" {
             description
               "This module does not belong to any category or can't be determined.";
           }
           enum "not-applicable" {
             description
               "This module is not applicable. For example, because the YANG module only contains typedefs, groupings, or is a submodule";
           }
         }
         description
           "The type of data element tree used by the module as it relates to the
            Network Management Datastores Architecture.";
         reference "draft-dsdt-nmda-guidelines Guidelines for YANG Module Authors (NMDA)";
       }


Regards, Benoit
>
> On Fri, 3 Nov 2017 at 11:05 Andy Bierman <[e-mail] 
> <mailto:[e-mail]>> wrote:
>
>     On Fri, Nov 3, 2017 at 10:48 AM, Juergen Schoenwaelder
>     <[e-mail]
>     <mailto:[e-mail]>> wrote:
>
>         My take here is that structured version numbers do only partially
>         solve the problem. Andy's work years ago on packages offers in
>         my view
>         a superior foundation for a solution. Once we can bundle
>         modules that
>         are designed and known to work together into meaningful
>         packages, then
>         it may be possible to relax some of the strict RFC 7950 update
>         rules.
>
>         Once the NETMOD WG gets the work on NMDA "completed", I
>         believe "YANG
>         packages" are a worthwhile target to work on. There is a need
>         to get
>         more structure into the module space, not just additional
>         structured
>         version numbers.
>
>
>     I agree ;-)
>
>     It is nice to have a way to know current implementations will probably
>     break if they upgrade to the new version. It is even nicer if the APIs
>     are stable and don't break existing code.
>
>     It is not encouraging that the IETF cannot produce stable YANG modules
>     published in RFCs.  We expect I-Ds to ignore the YANG update rules,
>     but not RFC versions.
>
>     Since the semantic versioning is only per-module, it is not
>     usefull for
>     determining if module foo will work with module bar.  If it is OK
>     to break backward-compatibility then it will become increasingly
>     difficult to just use the latest version of a module. Real
>     interoperability
>     at the granularity of modules will become more difficult. YANG
>     packages
>     can dramatically reduce the number of API permutations.
>
>
>         /js
>
>
>     Andy
>
>
>         On Fri, Nov 03, 2017 at 06:30:52PM +0100, Benoit Claise wrote:
>         > Dear all,
>         >
>         > Let me present this draft
>         >
>         https://datatracker.ietf.org/doc/draft-clacla-netmod-yang-model-update/
>         >
>         >                     New YANG Module Update Procedure
>         >  draft-clacla-netmod-yang-model-update-01
>         >
>         > Abstract
>         >
>         >    This document specifies a new YANG module update
>         procedure in case of
>         >    non backward-compatible changes, as an alternative
>         proposal to the
>         >    YANG 1.1 specifications.  This document updates RFC 7950.
>         >
>         >
>         > Problem statement:
>         > Changing a YANG module name each time there is a non
>         backward compatible
>         > change (as RFC7950 requires) adds a lot of complexity to
>         automation, from an
>         > import and service composition point of view.
>         >
>         > Solution:
>         > We need a different mechanism. The solution in the draft is
>         based on the
>         > semantic versioning YANG extension: it was proposed
>         openconfig in the past
>         > and is currently used by the openconfig YANG modules
>         >
>         > Note: there might other solutions, such as new YANG
>         keywords, but at this
>         > point in time, it's important to recognize that we need to
>         change the way we
>         > produce YANG modules at the IETF. Let's discuss on the list
>         and during the
>         > NETMOD meeting.
>         >
>         > Regards, Benoit.
>         > > On 10/12/2017 3:30 PM, Benoit Claise wrote:
>         > > > Hi Lou,
>         > > > >
>         > > > > So circling back to the original question: what do we
>         do about
>         > > > > the non backward-compatible module being defined in
>         rfc8049bis?
>         > > > >
>         > > > > While being sympathetic to many of the comments made
>         below as
>         > > > > well as the "do over" concept, I find the comments about
>         > > > > adhering to the rules of 7950 compelling - which leads to
>         > > > > renaming the module in the bis to ietf-l3vpn-svc-2.
>         > > > >
>         > > > > It would be good to ensure that this is the consensus
>         of the
>         > > > > group before asking the authors make this change.
>         > > > >
>         > > > Since this draft is AD sponsored, I'll evaluate the
>         consensus on
>         > > > RFC8049bis.
>         > > > Moving to ietf-l3vpn-svc-2 is the easy path not to break
>         backward
>         > > > compatibility. However, since ietf-l3vpn-svc is
>         unimplementable (it
>         > > > has broken XPATH expressions, so a compliant
>         implementation is
>         > > > impossible), so technically, ietf-l3vpn-svc does not
>         even exist.
>         > > See my message on this topic, as the IETF LC follow up.
>         > >
>         https://www.ietf.org/mail-archive/web/ietf-announce/current/maillist.html
>         > > If a follow up is required, I propose that we use a single
>         public email
>         > > thread: the [e-mail] <mailto:[e-mail]>
>         > >
>         > > Regards, Benoit
>         > > >
>         > > > What NETMOD should focus on is closing on the NMDA
>         transition: the
>         > > > ietf-routing versus ietf-routing-2 issue.
>         > > > Way bigger impact in terms of dependent YANG modules
>         > > >
>         > > > Regards, Benoit (as OPS AD)
>         > > > See below.
>         > > > >
>         > > > > This change course doesn't solve the versioning issue
>         discussed
>         > > > > below, but this is not a new issue it's just the first
>         time
>         > > > > we'll actually executing the steps envisioned as part
>         of the
>         > > > > rules laid out in yang. My personal take away is that
>         means that
>         > > > > we should immediately start work on an extension
>         defining along
>         > > > > the lines of  ' *_obsolete|update_*' mentioned below.
>         > > > >
>         > > > I believe that option 1 is the more pragmatic and
>         complete solution.
>         > > > option 2 is just half a step in the right direction.
>         > > > I believe we should discuss this topic in Singapore.
>         > > >
>         > > > Regards, Benoit (as individual contributor)
>         > > > >
>         > > > > Lou
>         > > > >
>         > > > > On October 8, 2017 10:59:15 AM Benoit Claise
>         <[e-mail] <mailto:[e-mail]>> wrote:
>         > > > >
>         > > > > > Dear all,
>         > > > > >
>         > > > > > Focusing on draft-wu-l3sm-rfc8049bis, the big
>         problem is:
>         > > > > > RFC8049 is broken. The small problem is: trying to
>         maintain
>         > > > > > backward compatibility.
>         > > > > > draft-wu-l3sm-rfc8049bis has rightly focused on the
>         big problem.
>         > > > > >
>         > > > > > Let me extend the scope of this email thread from
>         "handling
>         > > > > > module incompatibility" to "handling module
>         incompatibility
>         > > > > > and NMDA transition".
>         > > > > > As I mentioned in the past (see "semver.org
>         <http://semver.org> comparison of
>         > > > > > two YANG modules" email in NETMOD), I believe the
>         IETF will
>         > > > > > have to change its way of working in terms of backward
>         > > > > > compatibility. See also the email "ietf-routing or
>         > > > > > ietf-routing-2? module naming convention for NMDA
>         > > > > > transition. Re: [netmod] upcoming adoptions" in NETMOD.
>         > > > > >
>         > > > > > However, we will have to tackle this discussion one
>         day or the other:
>         > > > > > - we need _an automatic way_ to make the link
>         between the
>         > > > > > YANG module in RFC8049 and the YANG module in
>         > > > > > draft-wu-l3sm-rfc8049bis, or any non backward compatible
>         > > > > > YANG modules.
>         > > > > > - we need _an automatic way_ to make the link
>         between the
>         > > > > > YANG module in RFC8022 and the YANG module in
>         > > > > > draft-acee-netmod-rfc8022bis, or any new YANG module
>         names
>         > > > > > used for NMDA transition.
>         > > > > > Note: actually, we face two different problems.
>         > > > > > draft-wu-l3sm-rfc8049bis might be declared backward
>         > > > > > incompatible with RFC8049, while RFC8022bis is backward
>         > > > > > compatible with RFC8022. The RFC8022bis went for a
>         new YANG
>         > > > > > module name ietf-routing-2 to avoid to document the
>         -state
>         > > > > > tree (as deprecated), based on the argument that
>         > > > > > ietf-routing is not yet implemented.
>         > > > > >
>         > > > > > Which solutions do we have from here?
>         > > > > > #1. We keep the same module name and express that
>         the YANG
>         > > > > > module in draft-wu-l3sm-rfc8049bis is not backward
>         > > > > > compatible with the RFC8049 one. This is the
>         openconfig way.
>         > > > > > See draft-clacla-netmod-model-catalog (and
>         > > > > > draft-openconfig-netmod-model-catalog before)
>         > > > > >
>         > > > > >        // extension statements
>         > > > > >           extension openconfig-version {
>         > > > > >             argument "semver" {
>         > > > > >               yin-element false;
>         > > > > >             }
>         > > > > >             description
>         > > > > >               "The OpenConfig version number for the
>         module. This is
>         > > > > >               expressed as a semantic version number
>         of the form:
>         > > > > >                 x.y.z
>         > > > > >                where:
>         > > > > >                 * x corresponds to the major version,
>         > > > > >                 * y corresponds to a minor version,
>         > > > > >                 * z corresponds to a patch version.
>         > > > > >               This version corresponds to the model
>         file within which it is
>         > > > > >               defined, and does not cover the whole
>         set of OpenConfig models.
>         > > > > >               Where several modules are used to
>         build up a single block of
>         > > > > >               functionality, the same module version
>         is specified across each
>         > > > > >               file that makes up the module.
>         > > > > >
>         > > > > >               A major version number of 0 indicates
>         that this model is still
>         > > > > >               in development (whether within
>         OpenConfig or with industry
>         > > > > >               partners), and is potentially subject
>         to change.
>         > > > > >
>         > > > > >               Following a release of major version
>         1, all modules will
>         > > > > >               increment major revision number where
>         backwards incompatible
>         > > > > >               changes to the model are made.
>         > > > > >
>         > > > > >               The minor version is changed when
>         features are added to the
>         > > > > >               model that do not impact current
>         clients use of the model.
>         > > > > >
>         > > > > >               The patch-level version is incremented
>         when non-feature changes
>         > > > > >               (such as bugfixes or clarifications to
>         human-readable
>         > > > > >               descriptions that do not impact model
>         functionality) are made
>         > > > > >               that maintain backwards compatibility.
>         > > > > >
>         > > > > >               The version number is stored in the
>         module meta-data.";
>         > > > > >           }
>         > > > > >
>         > > > > > Similarly, we always keep the same YANG module name
>         in case
>         > > > > > of NMDA transition. So ietf-routing-2 should be
>         changed back
>         > > > > > to ietf-routing.
>         > > > > > The email thread "[Rtg-dt-yang-arch] ietf-routing or
>         > > > > > ietf-routing-2? module naming convention for NMDA
>         > > > > > transition. Re: [netmod] upcoming adoptions" seems
>         to go in
>         > > > > > that direction.
>         > > > > >
>         > > > > > #2. Or we have a different module name, let's say
>         > > > > > ietf-l3vpn-svc-2 or ietf-routing-2 but then, how do
>         we make
>         > > > > > the link with the previous module?
>         > > > > > Then ...  What? We create extension that will link the
>         > > > > > draft-wu-l3sm-rfc8049bis YANG module to the RFC8049 YANG
>         > > > > > module? Same principle as #1, but just more complex.
>         > > > > > Or we have a new YANG keyword (this implies YANG 2.0)
>         > > > > >
>         > > > > >     <CODE BEGINS>file"[e-mail]"
>         > > > > >     module ietf-l3vpn-svc-2 {
>         > > > > >       yang-version 1.1;
>         > > > > >       namespace
>         "urn:ietf:params:xml:ns:yang:ietf-l3vpn-svc";
>         > > > > >       prefix l3vpn-svc;
>         > > > > >       *_obsolete|update _*ietf-l3vpn-svc@2017-01-2
>         > > > > >
>         > > > > > And whose responsibility is this to warn/push all
>         authors of
>         > > > > > ietf-routing YANG modules to move to ietf-routing-2? (*)
>         > > > > >
>         > > > > > The following are non solution IMO:
>         > > > > > - Going from the draft-wu-l3sm-rfc8049bis YANG
>         _module _to
>         > > > > > the draft-wu-l3sm-rfc8049bis _document _to lookup
>         the IETF
>         > > > > > "obsolete" flag in order to understand that the
>         RFC8049 YANG
>         > > > > > module is obsolete is not an automatic solution.
>         > > > > > - Using the yangcatalog.org <http://yangcatalog.org>
>         might be a solution as we track
>         > > > > > the derived semantic, but this is just an offline trick.
>         > > > > > This is not what I call "automatic way"
>         > > > > > - Using the YANG module description field might be
>         > > > > > interesting, but again this is not an "automatic way":
>         > > > > >
>         > > > > >       description
>         > > > > >        "This YANG module defines a generic service
>         configuration
>         > > > > >         model for Layer 3 VPNs. This model is common
>         across all
>         > > > > >         vendor implementations. This obsoletes the
>         RFC8049 YANG
>         > > > > >         module, ietf-l3vpn-svc@2017-01-2";
>         > > > > >       revision 2017-09-14 {
>         > > > > >        description
>         > > > > >     "First revision ofRFC8049
>         <https://tools.ietf.org/html/rfc8049>.";
>         > > > > >        reference
>         > > > > >         "RFC xxxx: YANG Data Model for L3VPN Service
>         Delivery";
>         > > > > >
>         > > > > >
>         > > > > > In conclusion, I believe openconfig got this right
>         and that
>         > > > > > solution #1 is the solution to go ... while waiting
>         for a
>         > > > > > new YANG keyword in YANG 2.0
>         > > > > >
>         > > > > > (*) If you want to change the module from
>         ietf-routing to
>         > > > > > ietf-routing-2, then you should follow with all
>         authors of
>         > > > > > dependent modules to make sure they transition to
>         > > > > > ietf-routing-2
>         > > > > > In the yangcatalog.org <http://yangcatalog.org>,
>         because I needed the information as
>         > > > > > OPS AD, we created a small script to get that
>         authors list
>         > > > > > for IETF drafts. For the ietf-routing, this produces the
>         > > > > > following
>         > > > > > {
>         > > > > >     "output": {
>         > > > > >         "author-email": [
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>",
>         > > > > > "[e-mail]
>         <mailto:[e-mail]>"
>         > > > > >         ]
>         > > > > >     }
>         > > > > > }
>         > > > > >
>         > > > > > Fortunately, we only deal with IETF dependent YANG
>         modules
>         > > > > > in case of the ietf-routing. That's an easier case.
>         > > > > > If we would change ietf-interfaces to
>         ietf-interfaces-2, we
>         > > > > > would have an cross SDO issue ... Btw, there are no
>         > > > > > automatic ways to extract the authors of YANG
>         modules from
>         > > > > > different SDOs: it's only a metadata that that the
>         different
>         > > > > > SDOs should insert in the yangcatalog. So we would
>         have to
>         > > > > > rely on liaisons or direct emails, assuming we know the
>         > > > > > authors. Time consuming, believe me.
>         > > > > >
>         > > > > > Regards, Benoit
>         > > > > > > Hi,
>         > > > > > >
>         > > > > > >      As part of the my Routing Directorate review of
>         > > > > > > draft-wu-l3sm-rfc8049bis I noted that there were
>         several incompatible
>         > > > > > > changes being made to the ietf-l3vpn-svc module
>         without changing the
>         > > > > > > name.  I raised this with the YANG doctors and
>         others involved with the
>         > > > > > > draft and it surfaced some topics which really
>         should be discussed here
>         > > > > > > in NetMod.
>         > > > > > >
>         > > > > > > The background (as explained off-list by others,
>         so I hope I have it
>         > > > > > > right)  is that one of the YANG Doctors noted that
>         RFC8049 was broken
>         > > > > > > and could not be implemented as defined, and
>         therefore a fix was
>         > > > > > > needed.  L3SM has concluded so the fix is in the
>         individual draft
>         > > > > > > draft-wu-l3sm-rfc8049bis.  Since the rfc8049
>         version of ietf-l3vpn-svc
>         > > > > > > module could not be implemented, the feeling by
>         the YANG Dr was that
>         > > > > > > even though the new module is incompatible with
>         the original definition
>         > > > > > > the module the rule defined in Section 11 of YANG
>         1.1 (or section 10 of
>         > > > > > > RFC 6020) didn't have to be followed and the same
>         name could be used.
>         > > > > > >
>         > > > > > > In the subsequent discussion with the YANG Drs.,
>         the general discussion
>         > > > > > > was heading down the path of using a new module
>         name, and thereby not
>         > > > > > > violating YANG module update rules.  This lead us
>         back to the a similar
>         > > > > > > discussion we've been having in the context of
>         8022bis: how best to
>         > > > > > > indicate that a whole module is being obsoleted. 
>         RFCs do this by adding
>         > > > > > > 'metadata' to the headers, e.g., "Obsoletes:
>         8049", but this doesn't
>         > > > > > > help YANG tooling.  For 8022, we have one approach
>         - publishing an
>         > > > > > > updated rev of the original module marking all
>         nodes as deprecated - but
>         > > > > > > that doesn't really make sense for rfc8049bis.
>         > > > > > >
>         > > > > > > So the discussion for the WG is:
>         > > > > > >
>         > > > > > > How do we handle incompatible module changes,
>         notably when one module
>         > > > > > > 'obsoletes' another module --  from both the
>         process and tooling
>         > > > > > > perspectives?
>         > > > > > >
>         > > > > > > I know Benoit plans to bring in some
>         thoughts/proposals, perhaps there
>         > > > > > > are others.
>         > > > > > >
>         > > > > > > Cheers,
>         > > > > > >
>         > > > > > > Lou
>         > > > > > >
>         > > > > > > (as contributor/reviewer)
>         > > > > > >
>         > > > > > >
>         > > > > > >
>         > > > > > >
>         > > >
>         > >
>         >
>
>         > _______________________________________________
>         > netmod mailing list
>         > [e-mail] <mailto:[e-mail]>
>         > https://www.ietf.org/mailman/listinfo/netmod
>
>
>         --
>         Juergen Schoenwaelder           Jacobs University Bremen gGmbH
>         Phone: [phone] <tel:+49%20421%202003587>    Campus
>         Ring 1 | 28759 Bremen | Germany
>         Fax: [phone] <tel:+49%20421%202003103>  
>          <http://www.jacobs-university.de/>
>
>         _______________________________________________
>         netmod mailing list
>         [e-mail] <mailto:[e-mail]>
>         https://www.ietf.org/mailman/listinfo/netmod
>
>     _______________________________________________
>     netmod mailing list
>     [e-mail] <mailto:[e-mail]>
>     https://www.ietf.org/mailman/listinfo/netmod
>