Re: [netmod] [Technical Errata Reported] RFC7950 (6031)

["Sterne, Jason (Nokia - CA/Ottawa)" <[e-mail]>]

That seems like a reasonable approach to me.
Jason

From: netmod <[e-mail]> On Behalf Of Rob Wilton (rwilton)
Sent: Friday, April 24, 2020 3:34 PM
To: Kent Watsen <[e-mail]>; [e-mail]
Subject: Re: [netmod] [Technical Errata Reported] RFC7950 (6031)

Hi Kent,

Thanks for creating the issue.

I think that errata falls under section 7 of https://www.ietf.org/about/groups/iesg/statements/processing-rfc-errata/, and could be classified as “Hold for Document Update”.  I.e. “Changes that modify the working of a protocol to something that might be different from the intended consensus when the document was approved should be either Hold for Document Update or Rejected. Deciding between these two depends on judgment. Changes that are clearly modifications to the intended consensus, or involve large textual changes, should be Rejected. In unclear situations, small changes can be Hold for Document Update.”

I think that the consensus of the long term fix (e.g. in YANG 1.2) is that “require-instance” should be allowed under typedefs that refined types that allow it.

Pragmatically, I think that we can mark this errata is a “Hold for Document Update”, with the accompanying errata notes (derived from Radek’s comments) changed to:

“The document does not specify whether the “require-instance” keyword is allowed in typedef refinements derived from the “leafref” or “instance-identifier” base types, but it is anticipated that a future revision of YANG would allow this.   It is suggested that modules using YANG language versions 1 [RFC 6020] and 1.1 [RFC 7950] avoid using this construct, YANG module validation tools flag a warning if this construct is used, but implementations allow this if possible.”

Does anyone object to this course of action (or wishes to refine my errata notes)?

Regards,
Rob


From: Kent Watsen <[e-mail]<mailto:[e-mail]>>
Sent: 23 April 2020 17:59
To: Andy Bierman <[e-mail]<mailto:[e-mail]>>
Cc: Radek Krejci <[e-mail]<mailto:[e-mail]>>; Juergen Schoenwaelder <[e-mail]<mailto:[e-mail]>>; Martin Björklund <[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>; Rob Wilton (rwilton) <[e-mail]<mailto:[e-mail]>>
Subject: Re: [netmod] [Technical Errata Reported] RFC7950 (6031)

The consensus seems to be that:
  - the errata should be rejected
        - Rob, do you agree?
  - YANG-next should fix it later
        - I created https://github.com/netmod-wg/yang-next/issues/104
  - implementations should try to do the right thing now
        - Radek’s suggestion below LGTM!


Tallies:
   - for reject: Andy, Martin, Juergen, and Kent
   - for accept: Radek, and Balazs
   - unclear: Lada, Rob, and Jason


Kent // as co-chair


On Apr 14, 2020, at 10:35 AM, Andy Bierman <[e-mail]<mailto:[e-mail]>> wrote:

Hi,

I agree with Juergen that this errata should be rejected and the issue resolved in yang-next.
No IETF module should use this construct. It is easy to convert to an equivalent form that is not under dispute.


Andy


On Tue, Apr 14, 2020 at 6:40 AM Radek Krejci <[e-mail]<mailto:[e-mail]>> wrote:
Hi,
Dne 09. 04. 20 v 17:26 Kent Watsen napsal(a):


On Apr 6, 2020, at 3:42 AM, Juergen Schoenwaelder <[e-mail]<mailto:[e-mail]>> wrote:

The definition I found in RFC 8639 is this:

       leaf stream {
         type stream-ref {
           require-instance false;
         }
         mandatory true;
         description
           "Indicates the event stream to be considered for
            this subscription.";
       }

This could be changed to:

       leaf stream {
         type leafref {
    path "/sn:streams/sn:stream/sn:name";
           require-instance false;
         }
         mandatory true;
         description
           "Indicates the event stream to be considered for
            this subscription.";
       }

I can confirm that `yanglint` validates the module cleanly after this change.



On Apr 6, 2020, at 7:38 AM, Martin Björklund <[e-mail]<mailto:[e-mail]>> wrote:

I think the correct fix is to change the text so that
"require-instance" is not classified as a restriction and keep the
default.

Agreed.


Also, I think that it would be easiest (for backwards
compatibility w/ existing models) to allow "require-inetance" to be
changed in derived types.

However, this cannot imo be done in an errata.

While I appreciate Radek and Michal’s perspective, I also think that is would be best for the community for `yanglint` to support this, as they are published modules doing it.


I don't feel as an expert for IETF processes, so I don't know if this issue can be solved in errata or not (and I'm not sure there is a consensus on this in mailing list). For the implementation, I would appreciate at least a consensus on a solution. So far I saw opinions to allow it, to disallow and also to make it implementation-specific (which means in fact to disallow from the authors perspective, since there can be a tool disallowing it and we are saying that such a tool is ok). So, there is no clear way for implementors, which means problems for interoperability - there will be always someone unhappy and so far I don't know what is the major opinion to go.

So far, I tend to allow it (accept by libyang), but print warning to warn authors about possible problems (some tool can refuse such a module). Is it ok?

Radek

As an aside, I feel that all modules should be tested against all available validation tools during the publication process, but to find issues in the modules and well as possibly improve the tools.

Sadly, I only have `yanglint` and `yangson` available to me.  I just checked for the “yang validator” project, but both www.yangvalidator.com<http://www.yangvalidator.com/> and https://www.yangcatalog.org/yangvalidator seem to be down.


Kent // contributor


_______________________________________________
netmod mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/netmod