Re: [netconf] restconf collections

On Thu, Oct 29, 2020 at 9:22 AM Martin Björklund <mbj+ietf@4668.se> wrote:

> Kent Watsen <kent+ietf@watsen.net> wrote:
> > Hi Martin,
> >
> > > This is not a good proposal, since the result is not a valid XML
> > > document.
> >
> > How is this important?
>
> I will be difficult to parse, since you can't just parse it with a
> standard XML parser.
>
>
+1

It is perplexing why interoperability with standard XML tools would be
forfeited
just to save the encoding length of 1 container  (about 30 bytes)


Andy


> > It’s not an HTTP-constraint, right?
>
> No.
>
> > and I
> > don’t think it is needed for message-framing either.
> >
> > I guess, as you mention below, the media-type is inappropriate (though
> > 11.3.1 isn’t as clear as one might hope), so maybe the response is
> > okay if only changing the media type?
>
> Sure, from a pure media type pow; but I still think it would be a
> mistake to send a concatenation of valid XML documents w/o any
> separators (and adding custom separators may work, but note how the
> original NETCONF XML encoding tried to do that, but failed...).
>
> > > An XML document has exactly one root element.
> > >
> > > Also, the media types used in the examples are misleading.  The
> > > example reply doesn't match the media type
> > > "application/yang-data+json" (same for xml).
> >
> > How isn’t the media-type okay for JSON, for the response in the
> > previous message?  It seems valid to me...
>
> You're right, it is valid.  My bad.  I forgot how clever the JSON
> encoding of a single list entry was ;-)
>
> > No dispute regarding the XML media-type.
> >
> >
> > > I suggest you define new media types and a "collection" root
> > > element; or keep the media types and define "collection" in a YANG
> > > module as:
> > >
> > >  anydata collection;
> >
> > Noted, but first would like to understand why simply changing the
> > media-type alone (while keeping the responses in the previous message)
> > isn’t sufficient, for both xml and json.
> >
> > Each media-type can define its own format so, e.g.,
> > “application/yang-data+xml-list" could be used to define a list of XML
> > documents, right?
> >
> > That said, I acknowledge that having a root-element may aid
> > xml-parsers, and thus perhaps worthwhile for that reason...just want
> > to ensure we understand the motivations accurately.
>
> Yes this is my concern.
>
>
> /martin
> _______________________________________________
> netconf mailing list
> netconf@ietf.org
> https://www.ietf.org/mailman/listinfo/netconf
>