Re: [yang-doctors] Question about absolute and relative paths in YANG

[Martin Björklund <[e-mail]>]

Hi,

Italo Busi <[e-mail]> wrote:
> Any feedbacks/comments about this question?
> 
> Thanks, Italo
> 
> From: Italo Busi
> Sent: giovedì 23 settembre 2021 18:56
> To: [e-mail]
> Cc: Zhenghaomian <[e-mail]>om>; 'Aihua Guo'
> <[e-mail]>
> Subject: Question about absolute and relative paths in YANG
> 
> Hi YANG doctors,
> 
> I am quite active within the IETF CCAMP and TEAS WGs on YANG models
> for optical transport networks
> 
> We are having some discussions about the use of absolute and relative
> paths in some of the drafts we are developing for IETF
> 
> I have been told that YANG doctors prefer absolute paths to relative
> paths

I don't recognize this preference.  Also I don't know if you refer to
XPath paths in must/when or something else.  I assume you mean XPath
paths. 

IMO, a short relative path is often easier to read and understand, and
easier to get right.  In some cases it truly doesn't matter, and in
some other cases you must use one and not the other.

In the examples below, I'd use a relative path.


/martin



> We think it makes a lot of sense to use absolute paths whenever
> possible but there are some cases where we are not sure whether and
> how we could use absolute paths
> 
> We would appreciate your opinions and suggestions about how to address
> these cases
> 
> One case could be one of the augment statements defined in RFC8795 (TE
> Topology model):
> 
>   augment "/nw:networks/nw:network/nw:node/tet:te/"
>         + "tet:te-node-attributes/tet:connectivity-matrices/"
>         + "tet:label-restrictions/tet:label-restriction" {
>     when "../../../../../../nw:network-types/tet:te-topology/"
>        + "example-topology" {
> 
> The intention is to say that this augmentation applies only to a
> specific network-types of this network instance. Using an absolute
> path the code could become:
> 
>   augment "/nw:networks/nw:network/nw:node/tet:te/"
>         + "tet:te-node-attributes/tet:connectivity-matrices/"
>         + "tet:label-restrictions/tet:label-restriction" {
> when
> "/nw:networks/nw:network[nw:network-id=current()../../../../../../nw:network-id]/"
>    + "nw:network-types/tet:te-topology/"
>        + "example-topology" {
> 
> It seems that using the absolute path is moving the issue on requiring
> a relative path to get the value of the network-id of this network
> 
> Is there an alternative solution to avoid using relative path in this
> case?
> 
> Another case we have considered is the definition of a grouping with a
> leaf that references another node within the same network. In this
> case, using a relative path or an absolution path embedding a relative
> path required to get the network-id of this network, would make quite
> difficult to re-use this grouping in multiple places within the YANG
> tree
> 
> One work-around could be to use the refine statement, such as:
> 
> grouping node-reference {
>   leaf network-ref {
>     type leafref {
>       path "/nw:networks/nw:network/nw:network-id"
>     }
>   }
> 
>   leaf node-ref {
>     type leafref {
>       path "deref(../network-ref)/../nw:node/nw:nw:node-id"
>     }
>   }
> }
> 
> augment "/nw:networks/nw:network/nw:node/tet:te/"
>       + "tet:te-node-attributes/tet:connectivity-matrices/"
>       + "tet:label-restrictions/tet:label-restriction" {
>   uses node-reference {
>     refine network-ref {
>       default current()/../../../../../../network-id;
>       must current() = current()/../../../../../../network-id;
>    }
>   }
> 
> This work-around seems a bit weird but at least it should the need to
> use relative paths within the grouping but keep the requirement to use
> the relative paths in the refine statements
> 
> Is there a better solution to avoid using relative path within the
> grouping?
> 
> Thanks in advance for your help
> 
> Italo
>