Re: [core] ? WG adoption of draft-veillette-core-yang-cbor-mapping-00

[Andy Bierman <[e-mail]>]

On Thu, Apr 21, 2016 at 2:37 PM, Juergen Schoenwaelder <
[e-mail]> wrote:

> On Thu, Apr 21, 2016 at 09:00:29PM +0000, Michel Veillette wrote:
> > Hi Juergen
> >
> > The extract I have provided is from a publically available document, see:
> > https://www.iab.org/wp-content/IAB-uploads/2016/03/ipso-paper.pdf
> >
> > My remark about the LWM2M modeling language was just to highlight that
> such language don't necessary imply code generation.
> >
>
> YANG does not imply code generation either. A YANG module defines a
> contract, and in the light of a protocol a programmatic interface.
> There is nothing in YANG that requires code generation.
>
> It just happens that people often tend to automate things if they find
> themself repeatedly writing similar code. It is at the end a question
> of how much data a device exposes and what kind of developers you are
> dealing with and whether you have a product that is expected to evolve
> over years or something designed to be sold and thrown away
> afterwards. ;-)
>
> /js
>
> PS: I have just implemented a YANG data model 'manually' because I had
>     reasons to not depend on tool chains (the target environment are
>     OpenWrt type of devices). But still I took advantage of having
>     YANG tools available to validate test cases so that I can be
>     reasonably sure my manually written code is a good match of the
>     contract.
>
>

There are commercial and open-source toolchains that handle the NETCONF/YANG
interaction model and most of the data model details in the protocol stack,
instead of pushing that
work out to the model instrumentation.  This can save a lot of code and
makes sure
the client or server has consistent behavior.

But you are making an important point.
The expected use CoMI/CoOL co-authors have discussed is client firmware
written
to work with specific YANG modules/revisions.  The client needs to retrieve
enough
server state to know if the server supports the same module revisions, and
then just assumes
the API contract will be followed.  It does not need any YANG parser or
code automation.


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

Andy


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