Re: [netmod] [core] js review of draft-ietf-core-yang-cbor-12

[Juergen Schoenwaelder <[e-mail]>]

On Tue, Apr 07, 2020 at 10:55:28PM +0200, Carsten Bormann wrote:
> On 2020-04-07, at 21:47, Juergen Schoenwaelder <[e-mail]> wrote:
> > 
> > For me, the question is whether this ID defines SIDs and the other ID
> > details how they are managed or this ID imports the definition of SIDs
> > from the other document, which also details how they are managed.
> > Right now, it seem something in between, i.e., it is not clear what
> > the dependencies between these specifications is.
> 
> Thank you for that feedback.
> 
> So I think we need to be a bit more radical/precise:
> 
> -YANG-CBOR should explain that SIDs are 63-bit unsigned integers and how the delta encoding in the CBOR map keys works.  I.e., define the concept, but not the number space management.
> 
> -sid does the latter.
> 
> The document that defines the media types (comi) gets to bind YANG-CBOR, for these media types, to the specific SID concept defined in -sid.
> 
> -YANG-CBOR can still refer to -sid as a preferred way to manage the number space.
> But you can implement -YANG-CBOR without understanding -sid, so this is not a normative reference.
> Any other document using -YANG-CBOR can refer to -sid as well (or, exceptionally, to something specific for that usage).
>

This makes sense, but I prefer to have the media types defined in
YANG-CBOR. If I use CBOR with RESTCONF, I love to depend on YANG-CBOR
only and not in addition on COMI just for the media type definition.
The serialization format is defined in YANG-CBOR, hence it makes sense
to me to define the corresponding media-type(s) there as well.

Note, I did not manage to review COMI due to a lack of time. So I am
digging into the type definitions now. RESTCONF (RFC 8040) defines:

application/yang-data+json
application/yang-data+xml

COMI defines:

application/yang-data+cbor
application/yang-identifiers+cbor
application/yang-instances+cbor

It seems that yang-data+cbor really is yang-data+cbor+sid and it seems
there is no media type for yang-data+cbor+names (i.e., the CBOR
encoding that uses names instead of SIDs as keys). The other two media
types yang-identifiers+cbor and yang-instances+cbor seem to be COMI
specific. Hence, me preference would be to define something like

application/yang-data+cbor+sid
application/yang-data+cbor+name

in YANG-CBOR and to leave the other two

application/yang-identifiers+cbor
application/yang-instances+cbor

for COMI. There is a certain interest to stream binary encoded YANG
data and having a self-contained YANG-CBOR document would be desirable
to minimize dependencies and maintenance headache down the road.

/js

-- 
Juergen Schoenwaelder           Jacobs University Bremen gGmbH
Phone: [phone]         Campus Ring 1 | 28759 Bremen | Germany
Fax:   [phone]         <https://www.jacobs-university.de/>