Re: [netmod] 答复: (no subject)

[Martin Bjorklund <[e-mail]>]

Zhuzhiguo <[e-mail]> wrote:
> Hi Martin,
> Thank you for your reply.
> 
> If module-A and modul-B are 1.0 modules, are they should be presented
> in <hello>?

Yes, but note that if B is present in <hello>, a client will think
that all nodes in B are implemented.  This is why yang-library has the
'conformance' leaf.


/martin

> -----邮件原件-----
> 发件人: Martin Bjorklund [mailto:[e-mail]] 
> 发送时间: 2017年11月7日 20:08
> 收件人: Zhuzhiguo <[e-mail]>
> 抄送: [e-mail]; Liquan (Liquan, NOS CSD) <[e-mail]>
> 主题: Re: [netmod] (no subject)
> 
> Zhuzhiguo <[e-mail]> wrote:
> > Hi,
> > 
> > I have one question about how to implement YANG module that import 
> > other-module which is NOT be implemented?
> > 
> > For example, module-A import module-B, but any nodes that depend on B 
> > are not-supported
> > 
> > Module A {
> >    import module-B;
> > }
> > 
> > There are two way to mark module-B is not really in use:
> > 
> > Option-1: refer conformance-type in RFC 7895 (ietf-yang-library), mark
> > module-A as "implement", module-B is "import"
> 
> This is correct.  But if A augments some node in B, you have to
> implement B as well.
> -----////all these augments are deviated
> 
> > This way seems work, but some teammates think it may not comply with 
> > RFC.
> 
> Why wouldn't it?
> ------////client can get module-B, but can't do any operation on module-B
> 
> > And we also argue about what module-B should be presented in <hello>
> > 
> > <capabilities>
> >    <capability>module-A</capability>
> >    <capability>module-B</capability> ===>should any "import" module MUST
> >    be sent by server to client also?
> > <capabilities>
> 
> Assuming A and B are YANG 1.1 modules, they should NOT be listed in
> <hello> - yang-library is used instead.
> > Option-2: mark module-B as "implement" also, but mark all-nodes as 
> > deviated This way seems work also, but it will cause NETCONF-client 
> > and NETCONF-server load module that have NO node that can be accessed
> 
> There's no reason to do this.
> 
> 
> 
> /martin