Re: [YANG] keyref

Martin Bjorklund wrote:
> Phil Shafer <phil@juniper.net> wrote:
>>> I already made a proposal -- instead of a builtin type called keyref
>>> there would be an optional clause for a leaf of leaf-list called
>>> leafref that has the path-stmt from the keyref as its argument.
>> Your proposal called this a "documentation clause", which implies
>> something targeted to humans and not usable by software.  This
>> would be bad, but I would live with a distinct statement for the
>> ref target that is a formal target.
> 
> Why is formal redundancy good?  If you're using a leafref, the type is
> well known; it is the type if the target node.  But you still have to
> specify it again.  I assume that it would be an error to say that the
> leaf is of type int32, but the leafref points to a string leaf.   This
> also brings up the question when two types are considered to be equal.
> 

The type is not 'well known'.

There is a cryptic Xpath expression that contains many prefixes and names.
The DM reader first has to understand the Xpath to pick the right one,
then go to the top of the file and scan all the import statements to find
the module for that prefix, then find that module, then find the correct
leaf object within it.  If the desired leaf came from a uses expansion, then
the object will not be where the Xpath says it is, so all the groupings
for all the uses-stmts need to be tracked down and checked.
(If the nested leaf is called 'name', or something else generic, good
luck finding it with grep.)

After all this detective work, the leaf is found, and the DM writer finds
the name of the data type for the original leaf.

Or the type field in the original leaf could specify this data type
and an additional leafref clause can be used by a machine to
track down the additional details.

IMO this is not formal redundancy from the POV of the DM reader.
Maybe it is to a YANG compiler, but the DM reader is top
of the food chain here.

> 
> /martin
> 
> 

Andy

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