Kent, Thanks for the pushback on eid 8784 — you're right, it should be a Reject.
My HDU reflected a concern that the example was incomplete, but your point about holistic validation addresses it. YANG validation fires on every change to the intended configuration and evaluates the datastore in full, without reference to prior state. That means when a leafref target is instantiated, normal validation automatically re-evaluates all outstanding leafref constraints, including those embedded in union types. There is no asymmetric monitoring problem in a correct implementation — both directions are handled by the same pass. Your observation about the sentence's placement also holds: it sits in a paragraph addressing schema-node concerns but speaks to data-node behavior. That's a source of confusion, but confusion in an example is an issue with writing-quality, something yang-next should fix, not a normative gap that an erratum can or should fix. On Errata 7021, we agree — Reject. Thanks. > On Jun 26, 2026, at 9:51 AM, Kent Watsen <[email protected]> wrote: > > Hi Mahesh, any thoughts on these? > > I'm trying to apply all verified/HDU errata to the yang2 doc before the > window closes... > > Kent > > >> On Jun 8, 2026, at 10:52 AM, Kent Watsen <[email protected]> wrote: >> >> Hi Mahesh - thanks for looking at this so quickly! >> >> >>>> https://errata.rfc-editor.org/eid7021 >>> >>> Reject: >> >> Agreed. Also note that YANG-next Issue #75 >> <https://github.com/netmod-wg/yang-next/issues/75> seeks to remove the >> `revision-date` statement altogether. Rob's "yang-next-agreement" document >> concurs <https://github.com/netmod-wg/yang-next/issues/75>. >> >> >>>> https://errata.rfc-editor.org/eid8784 >>> >>> HDU: >> >> Actually, I'd consider Reject. >> >> YANG validation happens each time <intended> changes. The validation is >> conceptually holistic, and not sensitive to what the datastore had >> previously. >> >> I think that the RFC 7950 text is being misinterpreted. When writing "is >> removed", I think the text only means in the context of the example, and not >> specifically pointing to datastore per/post states. >> Besides, the paragraph the sentence is added to regards schema, not data, >> yet the sentence regards data. Even if it were an important clarification, >> it's on the wrong paragraph. >> >> >> >> Kent // contributor >> >> >> _______________________________________________ >> netmod mailing list -- [email protected] >> To unsubscribe send an email to [email protected] > Mahesh Jethanandani [email protected]
_______________________________________________ netmod mailing list -- [email protected] To unsubscribe send an email to [email protected]
