Re: [YANG] keyref

Martin Bjorklund wrote:
> Andy Bierman <ietf@andybierman.com> wrote:
>> Ex 2) data type for 'address' is in the leaf clause
>>
>>        leaf address {
>>            type yang:ip-address;
>>            leafref "../../interface[name = $this/../ifname]/address/ip";
>>        }
>>
>> Win-win scenario: works for humans, and works for tools.
> 
> Do you want this new 'leafref' statement to have the same semantics as
> the keyref type?
> 
> 

I'm not 100% sure what the semantics are.
I do not agree that the set of instances needs to be limited to
the current values in use in the leafref target.

There are 2 important corner-cases (2 sides of the same one):

1) pre-provisioning
    creating an entry that points to an instance that represents HW
    that has not been installed yet.  The config sits harmlessly, waiting until the
    hardware is installed to become active.

2) hot-swap cards
    As Juergen pointed out with the old SNMP "interface to Chicago" problem,
    the DM writer does not know if the config for a board that was
    just removed should be saved or deleted -- only the operator knows that.

I want the 'real' type to be used in the type field
and the optional leafref-stmt is available to describe whatever
relationship the DM writer wants.  The description clause
SHOULD identify this relationship in plain text as well.

>> One reason I was confused -- a bug in the example on p. 94-95:
> 
> Ok, fixed.
> 
> 
> /martin
> 
> 

Andy

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