Re: [netmod] ?= ?==?utf-8?q? mandatory choice with non-presence container cas

On Tue, 2019-06-25 at 13:49 +0000, Rob Wilton (rwilton) wrote:
> > -----Original Message-----
> > From: netmod <netmod-bounces@ietf.org> On Behalf Of Ladislav Lhotka
> > Sent: 25 June 2019 14:14
> > To: netmod@ietf.org
> > Subject: Re: [netmod] ?= ?==?utf-8?q? mandatory choice with non-presence
> > container cas
> > 
> > On Tue, 2019-06-25 at 13:59 +0200, Martin Bjorklund wrote:
> > > Michal Vaško <mvasko@cesnet.cz> wrote:
> > > > Hi Rob,
> > > > actually, I have used model with the container TOP just for
> > > > simplification, I have encountered the issue while implementing
> > > > ietf-ssh-server model from its current draft. I have created the
> > > > container "users" [1] without any "user" list instances. Now, you
> > > > may argue that this is still not a valid use-case because there are
> > > > no users but I only tried to satisfy the condition.
> > > 
> > > Yes, I think that this list (user) should have a "min-elements 1".  I
> > > think that matches the i ntent.
> > 
> > Interestingly, the "users" container actually becomes a P-container: its
> > presence indicates that the corresponding case is selected.
> 
> I don't think that this makes it a P-container.

Not formally, according to the current rules, but effectively "the container
itself carries some meaning" (sec. 7.5.1).

> 
> 
> 
>  It might make
> > sense for an admin to select this case even before any users are
> > configured.
> 
> Sure, the "users" container could have been marked as having presence in the
> YANG model.
> 
> 
> > This example also exposes the drawback of the XML representation - it
> > cannot distinguish between an empty list and nothing. In JSON, the
> > problems of this thread could potentially be circumvented by configuring
> > 
> > "users" : {
> >     "user" : [
> >     ]
> > }
> 
> I don't think that an empty list "exists" in a configuration datastore, i.e. I

I am not sure about this, it possibly depends on an implementation.

Lada

> don't think that it should impart any meaning, in that regard is seems
> somewhat like an NP-container. 

> 
> Thanks,
> Rob
> 
> 
> > Lada
> > 
> > > /martin
> > > 
> > > 
> > > 
> > > > There are some users
> > > > on the system but they are generated into the configuration
> > > > on-demand when operational data is requested.
> > > > 
> > > > Regards,
> > > > Michal
> > > > 
> > > > [1]
> > > > https://tools.ietf.org/html/draft-ietf-netconf-ssh-client-server-14#
> > > > page-22
> > > > 
> > > > On Tuesday, June 25, 2019 11:08 CEST, "Rob Wilton (rwilton)"
> > > > <rwilton@cisco.com> wrote:
> > > > 
> > > > > Hi Michal,
> > > > > 
> > > > > It is not the printing of the data that makes it valid/invalid.
> > > > > 
> > > > > I don't think that your input data was ever valid, because
> > > > > "container C" doesn't satisfy the mandatory statement because it
> > > > > isn't a real data node in the tree - it is instantiated when
> > > > > required and may be deleted when it is no longer required.
> > > > > 
> > > > > I.e. your model has been designed such that it can never be
> > satisfied.
> > > > > 
> > > > > If your model was instead:
> > > > > 
> > > > > container TOP {
> > > > >   leaf L {
> > > > >     type empty;
> > > > >   }
> > > > >   choice A {
> > > > >     mandatory true;
> > > > >     container C {
> > > > >       leaf L2 {
> > > > >         type empty;
> > > > >       }
> > > > >     }
> > > > >   }
> > > > > }
> > > > > 
> > > > > 
> > > > > Then this data is valid:
> > > > > 
> > > > > <TOP>
> > > > >   <L/>
> > > > >   <C>
> > > > >    <L2/>
> > > > >   </C>
> > > > > </TOP>
> > > > > 
> > > > > 
> > > > > But this data is not:
> > > > > 
> > > > > <TOP>
> > > > >   <L/>
> > > > > </TOP>
> > > > > 
> > > > > 
> > > > > Nor is this, which is directly equivalent to the one above,
> > > > > because the <C/> container doesn't really exist if it doesn't have
> > > > > a child node present.
> > > > > 
> > > > > <TOP>
> > > > >   <L/>
> > > > >   <C/>
> > > > > </TOP>
> > > > > 
> > > > > Thanks,
> > > > > Rob
> > > > > 
> > > > > 
> > > > > > -----Original Message-----
> > > > > > From: Michal Vaško <mvasko@cesnet.cz>
> > > > > > Sent: 24 June 2019 18:15
> > > > > > To: Andy Bierman <andy@yumaworks.com>
> > > > > > Cc: Rob Wilton (rwilton) <rwilton@cisco.com>; netmod
> > > > > > <netmod@ietf.org>
> > > > > > Subject: Re: [netmod] ?= mandatory choice with non-presence
> > > > > > container cas
> > > > > > 
> > > > > > Hi Andy,
> > > > > > 
> > > > > > On Monday, June 24, 2019 19:11 CEST, Andy Bierman
> > > > > > <andy@yumaworks.com>
> > > > > > wrote:
> > > > > > 
> > > > > > > On Mon, Jun 24, 2019 at 10:01 AM Michal Vaško
> > > > > > > <mvasko@cesnet.cz>
> > > > > > > wrote:
> > > > > > > 
> > > > > > > > Hi Rob,
> > > > > > > > I think there is a problem in the RFC because using only
> > > > > > > > allowed steps I got invalid data from initially valid data.
> > > > > > > > That cannot be
> > > > > > correct.
> > > > > > > No.  See sec. 7.5.7
> > > > > > > 
> > > > > > >    If a non-presence container does not have any child nodes,
> > the
> > > > > > >    container may or may not be present in the XML encoding.
> > > > > > > 
> > > > > > > 
> > > > > > > Just because your retrieval does not contain the NP-container,
> > > > > > > that does not mean the NP-container was not present in the
> > > > > > > server for the mandatory-stmt validation.
> > > > > > 
> > > > > > I agree, but these valid data were correctly printed into
> > > > > > invalid data. I do not think printing is allowed to change the
> > > > > > validity of data.
> > > > > > 
> > > > > > Michal
> > > > > > 
> > > > > > > Regards,
> > > > > > > > Michal
> > > > > > > > 
> > > > > > > > 
> > > > > > > Andy
> > > > > > > 
> > > > > > > 
> > > > > > > > On Monday, June 24, 2019 18:52 CEST, "Rob Wilton (rwilton)"
> > > > > > > > < rwilton@cisco.com> wrote:
> > > > > > > > 
> > > > > > > > > Hi Michal,
> > > > > > > > > 
> > > > > > > > > My thoughts:
> > > > > > > > > 
> > > > > > > > > According to 7.5.1:
> > > > > > > > > 
> > > > > > > > >    In the first style, the container has no meaning of its
> > > > > > > > > own,
> > > > > > existing
> > > > > > > > >    only to contain child nodes.  In particular, the
> > > > > > > > > presence of the
> > > > > > > > >    container node with no child nodes is semantically
> > > > > > > > > equivalent to
> > > > > > the
> > > > > > > > >    absence of the container node.  YANG calls this style a
> > > > > > > > > "non-
> > > > > > presence
> > > > > > > > >    container".  This is the default style.
> > > > > > > > > 
> > > > > > > > > Hence your request (because the NP container does not have
> > > > > > > > > any
> > > > > > > > > children)
> > > > > > > > is equivalent to:
> > > > > > > > >  <TOP>
> > > > > > > > >    <L/>
> > > > > > > > >  </TOP>
> > > > > > > > > 
> > > > > > > > > which fails the "mandatory" check.
> > > > > > > > > 
> > > > > > > > > Thanks,
> > > > > > > > > Rob
> > > > > > > > > 
> > > > > > > > > 
> > > > > > > > > > -----Original Message-----
> > > > > > > > > > From: netmod <netmod-bounces@ietf.org> On Behalf Of
> > > > > > > > > > Michal Vaško
> > > > > > > > > > Sent: 24 June 2019 17:39
> > > > > > > > > > To: netmod <netmod@ietf.org>
> > > > > > > > > > Subject: [netmod] mandatory choice with non-presence
> > > > > > > > > > container case
> > > > > > > > > > 
> > > > > > > > > > Hi,
> > > > > > > > > > I have encountered a situation that I think is not
> > > > > > > > > > covered by RFC
> > > > > > > > 7950. My
> > > > > > > > > > specific use-case was as follows.
> > > > > > > > > > 
> > > > > > > > > > model:
> > > > > > > > > > 
> > > > > > > > > > container TOP {
> > > > > > > > > >   leaf L {
> > > > > > > > > >     type empty;
> > > > > > > > > >   }
> > > > > > > > > >   choice A {
> > > > > > > > > >     mandatory true;
> > > > > > > > > >     container C;
> > > > > > > > > >   }
> > > > > > > > > > }
> > > > > > > > > > 
> > > > > > > > > > data:
> > > > > > > > > > 
> > > > > > > > > > <TOP>
> > > > > > > > > >   <L/>
> > > > > > > > > >   <C/>
> > > > > > > > > > </TOP>
> > > > > > > > > > 
> > > > > > > > > > Parsing was successful, but printing these data back to
> > > > > > > > > > XML
> > > > > > produced:
> > > > > > > > > > <TOP>
> > > > > > > > > >   <L/>
> > > > > > > > > > </TOP>
> > > > > > > > > > 
> > > > > > > > > > and parsing this correctly failed with missing mandatory
> > choice.
> > > > > > > > According
> > > > > > > > > > to section 7.5.7 [1], I think the C container could be
> > > > > > > > > > omitted but the whole situation does not seem correct.
> > > > > > > > > > Thank you for any
> > > > > > input.
> > > > > > > > > > Regards,
> > > > > > > > > > Michal
> > > > > > > > > > 
> > > > > > > > > > [1] https://tools.ietf.org/html/rfc7950#section-7.5.7
> > > > > > > > > > 
> > > > > > > > > > _______________________________________________
> > > > > > > > > > netmod mailing list
> > > > > > > > > > netmod@ietf.org
> > > > > > > > > > https://www.ietf.org/mailman/listinfo/netmod
> > > > > > > > 
> > > > > > > > _______________________________________________
> > > > > > > > netmod mailing list
> > > > > > > > netmod@ietf.org
> > > > > > > > https://www.ietf.org/mailman/listinfo/netmod
> > > > > > > > 
> > > > 
> > > > 
> > > > _______________________________________________
> > > > netmod mailing list
> > > > netmod@ietf.org
> > > > https://www.ietf.org/mailman/listinfo/netmod
> > > 
> > > _______________________________________________
> > > netmod mailing list
> > > netmod@ietf.org
> > > https://www.ietf.org/mailman/listinfo/netmod
> > --
> > Ladislav Lhotka
> > Head, CZ.NIC Labs
> > PGP Key ID: 0xB8F92B08A9F76C67
> > 
> > _______________________________________________
> > netmod mailing list
> > netmod@ietf.org
> > https://www.ietf.org/mailman/listinfo/netmod
-- 
Ladislav Lhotka
Head, CZ.NIC Labs
PGP Key ID: 0xB8F92B08A9F76C67