Michal Vasko <[email protected]> wrote:
> Hi Andy,
>
> thanks for the clarification but it definitely could have been worded
> in a clearer way. Both the 'conceptual data node' and 'rest of the
> nodes' hint that an actual node must be the target.
But it says:
[...] the first node in the absolute-schema-nodeid string
identifies the YANG data structure to augment
I think that is pretty clear.
/martin
>
> Regards,
> Michal
>
> On 2. 10. 2025 18:28, Andy Bierman wrote:
> >
> >
> > On Wed, Oct 1, 2025 at 11:51 PM Michal Vasko
> > <[email protected]> wrote:
> >
> > Hi,
> >
> > I would just like to include the text that I think restricts this
> > use of the augment-structure. It is in the description of this
> > extension in RFC 8791 <https://datatracker.ietf.org/doc/html/rfc8791>.
> >
> > The mandatory 'path' parameter value identifies the YANG
> > conceptual data node that is being augmented and is
> > represented as an absolute-schema-nodeid string, where the
> > first node in the absolute-schema-nodeid string identifies the
> > YANG data structure to augment, and the rest of the nodes in
> > the string identifies the node within the YANG structure to
> > augment.
> >
> > It clearly mentions referencing a "conceptual data node", not a
> > 'structure' extension. I think this is the reason we have not
> > supported it in 'libyang'. Note that I do not mind adding support
> > for this in libyang (we have already implemented it) but the
> > wording is definitely misleading if the intention was for
> > 'structure' to be allowed to be augmented directly. Maybe even
> > deserves an errata.
> >
> >
> >
> > It does not say the following nodes are required.
> >
> > and the rest of the nodes
> > INTENDED:
> > and the rest of the nodes (if any)
> > It is true that the old RFC 8040 'yang-data' extension did require a
> > container in it, but that was changed in RFC 8791.
> > This was done in part to make groupings work better e.g. data node
> > /foo/A
> > grouping stuff { leaf A { type string; }
> > }
> > rc:yang-data foo {
> > container foo {
> > uses stuff;
> > }
> > }
> > sx:structure foo { uses stuff; }
> > container foo { uses stuff; }
> >
> > Regards,
> > Michal
> >
> >
> >
> > Andy
> >
> > On 2. 10. 2025 8:38, Phú Liên wrote:
> >>
> >>
> >> ---------- Forwarded message ---------
> >> From: *Phú Liên* <[email protected]>
> >> Date: Thu, Oct 2, 2025 at 1:36 PM
> >> Subject: Re: [netmod] Question about using augment-structure to
> >> augment the root node of a Yang data structure
> >> To: Andy Bierman <[email protected]>
> >>
> >>
> >> Hi Andy,
> >>
> >> Thank you for your confirmation. I can now request a bug fix for
> >> Yanglint.
> >>
> >> Best regards,
> >> Phu
> >>
> >> On Thu, Oct 2, 2025 at 12:02 AM Andy Bierman <[email protected]>
> >> wrote:
> >>
> >>
> >>
> >> On Wed, Oct 1, 2025 at 12:56 AM Phú Liên
> >> <[email protected]> wrote:
> >>
> >> Hi RFC 8791 authors,
> >>
> >> I have a question regarding YANG data structure
> >> extensions (RFC 8791) and the use of augment-structure.
> >> Specifically, is it possible to augment the root of a
> >> YANG data structure?
> >>
> >> For example, with the YANG model below, yanglint
> >> validates successfully if the last "sx:augment-structure
> >> /foo:mystruct" statement is commented out. However, if
> >> that statement is included, yanglint reports an error:
> >>
> >> * Yang model
> >>
> >> ```
> >> module foo {
> >> namespace "foo";
> >> prefix foo;
> >>
> >> import ietf-yang-structure-ext { prefix sx; }
> >>
> >> sx:structure mystruct {
> >> container mycontainer {
> >> leaf leaf1 {
> >> type string;
> >> }
> >> }
> >> }
> >>
> >> sx:augment-structure /foo:mystruct/foo:mycontainer {
> >> leaf leaf2 {
> >> type string;
> >> }
> >> }
> >>
> >> // Augment to the root of the structure, so `leaf3`
> >> will be a sibling of `mycontainer`.
> >> sx:augment-structure /foo:mystruct {
> >> leaf leaf3 {
> >> type string;
> >> }
> >> }
> >> }
> >> ```
> >>
> >> * Yanglint error
> >>
> >> ```
> >> libyang err : Augment extension target node
> >> "/foo:mystruct" from module "foo" was not found.
> >>
> >> (/foo:{extension='sx:augment-structure'}/{augment='/foo:mystruct'})
> >> ```
> >>
> >> Moreover, if we cannot use augment-structure to add
> >> augments to the root of a structure, could you please
> >> suggest how to make the following leafref path valid in
> >> the YANG model below?
> >>
> >>
> >>
> >> I do not see any text in RFC 8791 that prohibits augmenting
> >> the structure.
> >> A valid path argument is provided to the extension.
> >>
> >> sx:augment-structure /foo:mystruct
> >>
> >> The structure 'mystruct' should be treated like a top-level
> >> NP container.
> >> It shares the same namespace as real data-def-stmts.
> >>
> >>
> >> Andy
> >>
> >> * a.yang
> >>
> >> ```
> >> module a {
> >> namespace "urn:a";
> >> prefix a;
> >>
> >> import b {
> >> prefix b;
> >> }
> >> import ietf-yang-structure-ext {
> >> prefix sx;
> >> }
> >>
> >> grouping leafrefs {
> >> leaf leaf1 {
> >> type leafref {
> >> path "/a:top/a:leaf2";
> >> }
> >> }
> >> }
> >>
> >> grouping leaves {
> >> leaf leaf2 {
> >> type string;
> >> }
> >> }
> >>
> >> container top {
> >> config false;
> >> uses leaves;
> >> }
> >>
> >>
> >> // Augments 'a:leaf1' into '/b:mystruct/b:mycontainer'.
> >> This augmented
> >> // 'a:leaf1' will be a sibling of 'b:myleaf'.
> >> //
> >> // XPath within a structure can only reference nodes
> >> inside the same
> >> // structure. Therefore, '/a:top/a:leaf2' must also be
> >> augmented into
> >> // 'b:mystruct' to satisfy the leafref path.
> >> sx:augment-structure "/b:mystruct/b:mycontainer" {
> >> uses leafrefs;
> >> }
> >>
> >> // This 'augment-structure' attempts to satisfy the
> >> leafref path by augmenting
> >> // '/a:top/a:leaf2' into the root of '/b:mystruct'.
> >> However, yanglint reports
> >> // an error because we are augmenting to the root of a
> >> structure.
> >> sx:augment-structure "/b:mystruct" {
> >> container top {
> >> uses leaves;
> >> }
> >> }
> >>
> >> // This 'augment-structure' is valid in yanglint but
> >> does not satisfy the
> >> // leafref path because it augments '/b:top/a:leaf2',
> >> not '/a:top/a:leaf2'.
> >> sx:augment-structure "/b:mystruct/b:top" {
> >> uses leaves;
> >> }
> >> }
> >> ```
> >>
> >> * b.yang
> >>
> >> ```
> >> module b {
> >> namespace "urn:b";
> >> prefix b;
> >>
> >> import ietf-yang-structure-ext {
> >> prefix sx;
> >> }
> >>
> >> sx:structure mystruct {
> >> container mycontainer {
> >> leaf myleaf {
> >> type string;
> >> }
> >> }
> >> container top { }
> >> }
> >> }
> >> ```
> >>
> >> Yanglint reports the following error for the second
> >> augment-structure statement:
> >>
> >> ```
> >> libyang err : Augment extension target node "/b:mystruct"
> >> from module "a" was not found.
> >> (/a:{extension='sx:augment-structure'}/{augment='/b:mystruct'})
> >> ```
> >>
> >> And the following error for the third augment-structure
> >> statement:
> >>
> >> ```
> >> libyang err : Not found node "top" in path.
> >> (/b:mycontainer/a:leaf1)
> >> ```
> >>
> >> By the way, we found that this use of augment-structure
> >> is claimed to be supported in the following links:
> >>
> >> *
> >>
> >> https://docs.yumaworks.com/en/latest/yangauto/define-abstract-yang.html
> >> *
> >>
> >> https://dr2lopez.github.io/yang-provenance/draft-lopez-opsawg-yang-provenance.html
> >>
> >> BR/Phu
> >> _______________________________________________
> >> netmod mailing list -- [email protected]
> >> To unsubscribe send an email to [email protected]
> >>
> >>
> >> _______________________________________________
> >> netmod mailing list [email protected]
> >> To unsubscribe send an email [email protected]
> > _______________________________________________
> > netmod mailing list -- [email protected]
> > To unsubscribe send an email to [email protected]
> >
> >
> > _______________________________________________
> > netmod mailing list [email protected]
> > To unsubscribe send an email [email protected]
_______________________________________________
netmod mailing list -- [email protected]
To unsubscribe send an email to [email protected]
