Juergen Schoenwaelder <[email protected]> wrote:
> On Thu, Apr 16, 2015 at 09:15:47AM +0200, Juergen Schoenwaelder wrote:
> > On Thu, Apr 16, 2015 at 09:02:32AM +0200, Martin Bjorklund wrote:
> > > Juergen Schoenwaelder <[email protected]> wrote:
> > > > I think some thought should also be given to simplify the references.
> > > > Right now, in order to refer to a termination point, I have to use a
> > > > triple (network-ref, node-ref, tp-ref). Have you considered using an
> > > > instance-identifier with require-instance instead? One option would be
> > > > to define proper typedefs such as
> > > > 
> > > >    typedef tp-ref {
> > > >      type instance-identifier { require-instance }
> > > >      description
> > > >        "An instance identifier refering to a termination point."
> > > >    }
> > > > 
> > > > and then things can be collapsed. One would likely do the same for
> > > > other model elements, that is introduce typdefs for node-ref, link-ref
> > > > network-ref. This will make the tree much more compact.
> > > 
> > > There are two drawbacks with this scheme.  First, the model is less
> > > precise.  An instance-identifier can refer to anything, and there is
> > > not formal constraint in the model to restrict it to a termination
> > > point.
> > 
> > Hence you wrap it in a typedef with an appropriate description clause.
> > And by using the typedef, the model actually becomes more readable.
> > 
> > > Second, even if the data model becomes more compact, the
> > > instance data with an instance-identifier is more verbose and maybe
> > > harder to read for an operator (less "direct").
> > 
> > I love to hear from operators that they find
> > 
> >   /nt:network[foo]/nt:node[bar]/nt:termination-point[baz]
> > 
> > more complicated than this:
> > 
> >   
> > <nt:network-ref>foo</nt:network-ref><nt:node-ref>bar</nt:node-ref><nt:termination-point>baz</nt:termination-point>
> >
> 
> Oops, I got it wrong:
> 
>     
> /nt:network[network-id="foo"]/nt:node[node-id="bar"]/nt:termination-point[tp-id="baz"]
> 
>     
> <nt:network-ref>foo</nt:network-ref><nt:node-ref>bar</nt:node-ref><nt:tp-ref>baz</nt:tp-ref>
> 
> Anyway, if I count characters, the instance-identifier is still shorter.
> 
> /js
> 
> PS: If the argument is that operators can't understand instance identifier,
>     then NACM is horribly operationally broken. ;-)

No the argument is not that operators can't understand instance
identifiers.  The argument is that leafrefs are more precise and might
be easier to read and write than i-is.


/martin

_______________________________________________
i2rs mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/i2rs

Reply via email to