Re: [YANG] mandatory & default

Andy Bierman writes:
>We strongly disagree on the interpretation of RFC 4741
>and whether a node with a default value really exists or not.

Are you saying that 4741 forces defaults to be emitted in <get-config>?

>It is very fragile SW design to assume the agent is using the exact default
>that was tracked down in the documentation.

Here we strongly disagree.  Adding noise to configs makes them
more fragile, not less so.

In using the term "tracked down", are you missing that putting these
values in the YANG module (or any metadata/schema) makes them more
than just documentation?  An application should be able to move
from a node in the data tree to the YANG definition of this node
with ease.  Finding the default value (and the data type, range,
and all the other constraints on the data values) should be no
problem for the client application.

>It is up to the operator whether knowing that a node
>is actually supported, and knowing the exact value set by the agent,
>is a waste of space or important debugging information.

You should publish the with-defaults capability as a draft.  I
do see this as having value, but disagree with making it the
default behavior.  If there's a way to turn it on, operators
that want it on can turn it on in their client software.

Thanks,
 Phil

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