Re: [netconf] WGLC: draft-ietf-netconf-https-notif-06

[Kent Watsen <[e-mail]>]

NETCONF,

      The just posted -07 addresses issues raised thus far.
      https://tools.ietf.org/html/draft-ietf-netconf-https-notif-07 <https://tools.ietf.org/html/draft-ietf-netconf-https-notif-07>

Henning, Eric, Reshad,

      Please re-review as your comments are important to us.

Everyone else,

      The hasn’t been many reviews as yet.  Please review this
      version, as it is significantly better (ready for publication?)
      than the prevision version.

K.



> On Feb 1, 2021, at 11:33 AM, Reshad Rahman <[e-mail]> wrote:
> 
> Good with me.
>  
> Regards,
> Reshad.
>  
> From: Kent Watsen <[e-mail]>
> Date: Sunday, January 31, 2021 at 7:54 AM
> To: Reshad Rahman <[e-mail]>
> Cc: "[e-mail]" <[e-mail]>, "[e-mail]" <[e-mail]>
> Subject: Re: [netconf] WGLC: draft-ietf-netconf-https-notif-06
>  
>  
>> I've reviewed this document and I think the document will be ready once the changes planned by the authors (regarding notification-messages) are made.
>  
> Thanks, but a couple more problems:
>  
> 1) the URLs are bad.  Specifically, the example in the “Learning Receiver Capabilities” section uses a nonsensical URL (‘/‘).   The authors propose to instead use a sub-resource to the user-configured “path” (e.g., GET /some/path/capabilities) and move the POST URL to another sub-resource (e.g., /some/path/relay-notification).  [see below for more on this]
>  
> 2) the media-types are bad: “application/yang-data+[xml/json]” is used for notifications, which is wrong since they are not YANG-defined, and the custom media-types for capabilities (application/ietf-https-notif-cap+[xml/json]) is a way overkill.  The authors propose to use “application/[xml/json]” for both cases.
>  
> 
> 
>> I have a question on the following YANG description. Section 4.1 mentions 'path-absolute' for the URI but the description below says "Relative URI...". Should this description be clarified/corrected or am I missing something?
>> 
>>             augment "transport/tls/tls/http-client-parameters" {
>>               leaf path {
>>                 type string;
>>                 description
>>                   "Relative URI to the target resource.";
>>               }
>>               description
>>                 "Augmentation to add a path to the target resource.";
>>             }
>  
> Agreed.  How about this?
>  
>         leaf path {
>           type string;
>           mandatory true;
>           description
>             "URI prefix to the target resources.  Under this
>              path the receiver must support both the 'capabilities'
>              and 'relay-notification' resource targets, as described
>              in RFC XXXX.";
>         }
> 
> So, if path==“/som/path”, capability-discovery would be to "/some/path/capabilities” and notifications would be sent to “/some/path/relay-notification”.
>  
> Thoughts?
>  
>  
>> Formatting nit:
>>            Trust anchors (i.e. CA certs) that are used to authenticat\
>>  e
>  
> Fixed, but the other example cannot be fixed because the “xmlns” strings alone are too long, and XML doesn’t internally support folding, and the “string” type doesn’t allows for whitespace (include ‘\n’) both at the beginning and end of strings (recall my request for rfc6991-bis to define a “token” type that would enable the parser to discard any found, so our examples could more often be manually-folded).
>  
> K.