Re: [YANG] new pyang errors

Phil Shafer wrote:
> Andy Bierman writes:
>> The current draft clearly says that 'mandatory true' refers to the
>> configuration database, not to whether the manager must provide
>> an explicit value.
>> I agree with the interpretation in the current draft.
> 
> Well, you agree with your interpretation of the draft ;^)
> 
> The issue isn't whether a leaf with a default has to appear
> in any particular PDU, but whether a leaf with a default
> "exists" until some{one,thing} creates it.  If leafs (and
> non-presence containers) are automagically created "in the
> configuration database", then configurations become big
> and verbose, which is the opposite of what I want.  So
> I don't want leafs with defaults to be considered "created".
> 
> And when I read the draft with that prespective, it agrees
> with me.  So the draft is underspecified.  What outcome do
> we want and what do need to add to the draft to ensure there
> is only one possible reading of our spec?
> 

It is really up to the DM writer whether they want agent-created
default leafs or not.  I do not care about containers as much,
although the principle is the same.

There is no auto-magic.
The logic is very simple when a leaf value is not provided by the manager:

Q1: Is the leaf mandatory?

   No:  processing is done, node is not instantiated in the agent

  Yes:  Q2: Is there a default-stmt for the leaf, or if not, is there
            a default-stmt for the typedef used by the leaf?

          No: missing mandatory data error

         Yes: use the default value to instantiate the leaf node

> Thanks,
>  Phil
> 
> 

Andy

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