Re: [YANG] default values

Andy Bierman <ietf@andybierman.com> wrote:
> Martin Bjorklund wrote:
> > Hi,
> >
> > Andy Bierman <ietf@andybierman.com> wrote:
> >   
> >> A trivial example is a boolean knob (or presence container in YANG)
> >> that represents some service or feature, and the WG agrees it is 'off' 
> >> by default.
> >>     
> >
> > This is an interesting example, but currently YANG does not support
> > this use case, i.e. marking a presence container as present by
> > default.  Is that something we should add?
> >   
> 
> What does it mean wrt/ the child nodes of the container
> if the default present flag is set to true?
         ^^^^^^^^^^^^^^^^^^^^
What is that?

> Isn't the container created by default if it has any child nodes with
> defaults?  If this flag=FALSE, does that change?

A normal structural container (with no presence statement) behaves
like that, except that it seems to be a meta-physical question whether
the container really is created or not ;-)

> This new feature would allow the container to be created even if
> it has no child nodes with defaults?

Currently, it cannot be defined in the data model if a presence
container gets created by default or not.  So if you do:

  container foo {
    presence "turns on the foo protocol";

    leaf bar {
       type int32;
       default 42;
    }
  }

Then it is not defined if foo exists or not (when it's parent gets
created).  If it exists, foo/bar has the value of 42 by default.

At least we should clarify this in the draft, e.g. by saying that a
presence container is not created by default.  That means that the
manager knows which behavior to expect.

Or we could add a new statement to control this.

> Can this be turned on/off with a container refinement statement (in a 
> uses-stmt)?

If we add such a statement, yes I think it could be refined, just like
defaults for leafs.

/martin

_______________________________________________
YANG mailing list
YANG@ietf.org
https://www1.ietf.org/mailman/listinfo/yang