Re: [netmod] system-creatable

["Randy Presuhn" <[e-mail]>]

Hi -

> From: "Martin Bjorklund" <[e-mail]>
> To: <[e-mail]>
> Sent: Monday, October 12, 2009 9:33 AM
> Subject: [netmod] system-creatable
...
> Here is proposed text for a new statement "system-creatable"
> (a.k.a. "assigned-by system").
> 
> 
> 7.6.6.  The leaf's system-creatable Statement
> 
>    The "system-creatable" statement, which is optional, specifies
>    whether the server is permitted to create instances of a leaf.  Such
>    leafs are used when the device needs to record persistent values that
>    cannot be known until the device validates the configuration.  The
>    argument is one of the strings "true" or "false".  If not specified,
>    the default is "false".
> 
>    If "system-creatable" is "true", the server MAY create an instance of

Replace "MAY" with "MUST".  Otherwise, interoperability will suffer.

>    the leaf in the data tree, if it doesn't exist, in the following
>    cases.  The behavior depends on the leaf's closest ancestor node in
>    the schema tree which is not a non-presence container:
> 
>    o  If no such ancestor exists in the schema tree, the server MAY

Replace "MAY" with "MUST".

>       create an instance of the leaf if it doesn't exist.
> 
>    o  Otherwise, if this ancestor is a case node, the server MAY create

Replace "MAY" with "MUST".

>       an instance of the leaf in the data tree if any node from the case
>       exists in the data tree, or if the case node is the choice's
>       default case, and no nodes from any other case exist in the data
>       tree.
> 
>    o  Otherwise, the server MAY create an instance of the leaf in the
>       data tree if the ancestor node exists in the data tree.

Replace "MAY" with "MUST".

As I see it, the only difference between system-creatable and "default"
is that a default value is explicitly given in the model, and is consequently
in some sense redundant when given in response to a retrieval operation.

Randy