Hi,

On Sat, Jul 30, 2022 at 1:54 PM Michael Richardson <[email protected]>
wrote:

>
> I had a few (well.. two) hallway conversations about RFC9164 (IPv4/IPv6)
> tags
> for CBOR this week.
>
> Specifically... in large YANG described dumps (such as a BGP FIB table) it
> becomes critical not spend so many bytes on some fundamental datatypes when
> there are hundreds of thousands of entries.
>
> "ietf-yang-types" and "ietf-inet-types"
> Also a need for AS numbers, MAC addresses in optimized formats.
>
> I think that we concluded that there are three ways to do this.
>
> 1) a hack in a document when I use yang:ipv6-address I wrote some text
>    that says one should use CBOR foo.  The result is very specific to that
>    use, and it is never encoded in the YANG. [not great]
>
> 2) some kind of extension where when I *use* yang:ipv6-address that I say
>    that it should be encoded in CBOR using tag <foo>.
>    Also specific to that document/module, but the YANG knows.
>    Russ speculated on the ways in which one could use the right yang
>    extensions to do this without new syntax.
>
>
IMO (2) is the best approach
  - Extensions can be added inline or from a different module using
"deviate add"
  - could use 2 approaches: (1) define a new type and (2) use an existing
type + extension
  - Other tools (like code generators and maybe SID file generator)
    can use the extension

Something like:

    leaf foo {
        type inet:ipv6-address;
        ext:cbor-type cbor:bin-ipv6-address;
    }

Assumes that some module like ietf-cbor-types.yang is published with
appropriate
replacement data types.


3) some revision to rfc6991 that would include statements about how to
> encode
>    the known things in CBOR.  Then we just have to include a new enough
> YANG
>    module that specifies things.
>
> It seems that draft-ietf-netmod-rfc6991-bis-13 is in progress, although I
> don't seem to follow the list to know what's up with.
> (I thought I subscribed, but maybe it was all via IMAP)
>
> Is there some interest in doing this?
> I guess the better question is, what are the objections?
>
>
Strongly support.



> --
> Michael Richardson <[email protected]>, Sandelman Software Works
>  -= IPv6 IoT consulting =-
>
>
>
Andy


>
> _______________________________________________
> netmod mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/netmod
>
_______________________________________________
netmod mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/netmod

Reply via email to