Re: [netmod] [Netmod-ver-dt] Adding a feature is it BC or NBC ? [was: RE: YANG packages draft - now ready for review please]

[Balázs Lengyel <[e-mail]>]

Hello Rob,

I agree: so introducing the “not” operator into if-feature IMHO was a mistake. It can have its uses, but it is very dangerous.

regards Balazs

(Sending it to the full netmod group.)

 

From: Rob Wilton (rwilton) <[e-mail]> 
Sent: 2019. október 11., péntek 12:50
To: Balázs Lengyel <[e-mail]>; Reshad Rahman (rrahman) <[e-mail]>; [e-mail]
Subject: RE: [Netmod-ver-dt] Adding a feature is it BC or NBC ? [was: RE: YANG packages draft - now ready for review please]

 

Hi Balazs,

 

This is an interesting example.

 

It doesn’t just break packages, but breaks YANG conformance more generally, particularly if the leaf was also marked as mandatory.  I.e. whether a server implements the radius feature will determine whether or not the configuration is valid.

 

Basically, I think that this is an unwise use of if-feature.  Generally, I think that the principle behind features is that they represent functionality that is optional to support, and probably should not be used in this way to remove nodes from the underlying model.  I think that this is probably a bug/flaw in YANG.

 

Thanks,
Rob

 

 

 

From: Netmod-ver-dt <[e-mail] <mailto:[e-mail]> > On Behalf Of Balázs Lengyel
Sent: 10 October 2019 23:57
To: Rob Wilton (rwilton) <[e-mail] <mailto:[e-mail]> >; Reshad Rahman (rrahman) <[e-mail] <mailto:[e-mail]> >; [e-mail] <mailto:[e-mail]> 
Subject: [Netmod-ver-dt] Adding a feature is it BC or NBC ? [was: RE: YANG packages draft - now ready for review please]

 

Hello,

Adding a new feature MAY or MAY NOT be backward compatible. Think about the following YANG:

 

feature radius {}

 

leaf xxxx {

   if-feature “not radius”;

}

 

So if I add radius it actually removes the leaf xxx. NBC.

regards Balazs

 

Regards Balazs

 

From: Netmod-ver-dt <[e-mail] <mailto:[e-mail]> > On Behalf Of Rob Wilton (rwilton)
Sent: 2019. október 10., csütörtök 13:17
To: Reshad Rahman (rrahman) <[e-mail] <mailto:[e-mail]> >; [e-mail] <mailto:[e-mail]> 
Subject: Re: [Netmod-ver-dt] YANG packages draft - now ready for review please

 

Hi Reshad,

 

Thanks for the comments.

 

I’ve fixed all of the typos.

 

For the Abstract:

 

Old:

               This document defines YANG packages, a versioned

               organizational structure holding a set of related YANG

               modules, that can be used to simplify the conformance and

               sharing of YANG schema.  It describes how YANG instance data

               documents are used to define YANG packages, and how the YANG

               library information published by a server can be augmented

               with packaging related information.

 

Proposed:

 

               This document defines YANG packages, a versioned organizational

               structure holding a set of related YANG modules, that collectively

               define a YANG schema.  It describes how YANG instance data documents are

               used to define YANG packages, and how the YANG library information

               published by a server can be augmented with packaging related

               information.

 

For 5.2.1.1

Re: “Should this list also include/state “any NBC changes to a module in the package”?”

 

Ah, I see.  This is what the following current text of the second paragraph was meant to mean

 

Old:

                   <t>Changing a package import to select a package version that is

                   non-backwards-compatible to the prior package version, or removing a

                   previously imported package.</t>

                   <t>Changing a module import to select a module revision that is

                   non-backwards-compatible to the prior module revision, or removing a

                   previously implemented module.</t>

                   <t>Removing a previously supported feature.</t>

                   <t>Adding, changing, or removing a deviation that is considered a

                  non-backwards-compatible change to the affected data node in the

                   schema associated with the prior package version.</t>

 

Perhaps the following text would be more clear:

 

                   <t>Changing an 'imported-packages' list entry to select a package

                   version that is non-backwards-compatible to the prior package

                   version, or removing a previously imported package.</t>

                   <t>Changing a 'modules' or 'import-only-modules' list entry to

                   select a module revision that is non-backwards-compatible to the

                   prior module revision, or removing a previously implemented

                   module.</t>

                   <t>Removing a feature from the 'supported-feature' leaf-list.</t>

                   <t>Adding, changing, or removing a deviation that is considered a

                   non-backwards-compatible change to the affected data node in the

                   schema associated with the prior package version.</t>

 

Re 5.5.1 “MUST” instead of “SHOULD” for uniqueness.

Probably we should discuss today, but my default position is that I think that I will change this.

 

I think that my question was that how do you enforce global uniqueness ….

 

Thanks,

Rob

 

 

From: Reshad Rahman (rrahman) <[e-mail] <mailto:[e-mail]> > 
Sent: 10 October 2019 03:57
To: Rob Wilton (rwilton) <[e-mail] <mailto:[e-mail]> >; [e-mail] <mailto:[e-mail]> 
Subject: Re: [Netmod-ver-dt] YANG packages draft - now ready for review please

 

Hi Rob,

 

Looks good, but TBH I haven’t gone over the minute details of the examples etc. Some comments. 

 

Abstract

 

The sentence with “simplify the conformance and sharing of YANG schema” doesn’t jive well to me. I think I know what you mean (we probably discussed this previously), I don’t have better phrasing to propose. Only suggestion I have is to consider shortening the abstract.

 

Terminology

 

s/oraganization/organization/

 

Introduction

 

s/appendicies/appendices/?

 

Objectives

 

s/were a client/where a client/

 

YANG Package definition

 

s/oraganization/organization/

s/collectively define/collectively defines/?

3rd paragraph starting with “Each version of a YANG package defines…”. Do this in bullet form instead?

 

5.1 Package definition rules

 

s/decendent/descendent/

s/supercedes/supersedes/

 

5.2.1

Leaf is ‘nbc-changes’ (not ‘nbc-change’)

 

5.2.1.1.  Non-Backwards-compatible changes

Should this list also include/state “any NBC changes to a module in the package”?

 

5.3

 

s/libary/library/

 

5.5.1

 

Why SHOULD is used for uniqueness? Why not MUST?

s/publically/publicly/?

 

6

s/pacakge/package/

 

Regards,

Reshad.

 

From: Netmod-ver-dt <netmod-ver-dt-bounces@ietf..org <mailto:[e-mail]> > on behalf of "Rob Wilton (rwilton)" <[e-mail] <mailto:[e-mail]> >
Date: Monday, October 7, 2019 at 9:58 AM
To: "[e-mail] <mailto:[e-mail]> " <[e-mail] <mailto:[e-mail]> >
Subject: [Netmod-ver-dt] YANG packages draft - now ready for review please

 

Hi,

 

I think that this document has had sufficient updates that it is ready for review.

 

The latest version is available at per the commit information above, or attached.

 

I appreciate that it is quite short notice, but if it is possible for folks to review and provide comments before Thursday’s meeting that probably be helpful.

 

Thanks,

Rob