YANG Versioning Weekly Call Minutes - 2021-02-09 Agenda bashing: - drive to updated YANG Versioning and Semver drafts (deadline is ~2 weeks) - Balazs & T1 (state & NBC) - whitespace - 3GPP / ORAN update
Updates to YANG Versioning: - T3: filenames & chars: [Reshad] update draft & send to Weekly Call Group - T2: [Rob] to do a pull request with draft text & to NETMOD list - Import by revision: impact of changing import statements: did we post back to the WG? Said it was BC (but you can mark it NBC if you wanted to) - NBC/BC for config false - Prio 2: [Reshad] see if any other github issues are closed and could be folded in [Balazs] NBC/BC for Config false: - reduce value space for optional or mandatory nodes is BC. - increasing value space (within the type) for optional or mandatory nodes is BC. - type change is NBC - optional->mandatory is BC. mandatory->optional is NBC. - systems can be more tolerant to state changes, no validation on outgoing state - trying to accomodate what most clients can handle (otherwise too many changes will be flagged NBC, too much noise) - be strict with NBC in config, but less strict in state - draft document text, pass snippets to the WG [Jason] Put this under definition of NBC marker: - changes can be marked NBC any time if the author thinks there is a compat. issue that impacts most clients [Jan] Whitespace: - publisher (in the act of publishing) would always give a new version # - what if a user (non-owner) of a module adds annotations (e.g. yang extensions, comments), etc (e.g. aids for GUI generation). What is the version/label of that derived result? It is a different version or even a different module (different entity than the original owner). - can you republish module with same date making some whitespace changes. No. You must give a new version. - publish with linux then republish with dos: different versions. - how to define/describe publish ? - tools may say that 2 versions have no changes (if the tool ignores whitespace and line endings). But that's acceptable. Jason
_______________________________________________ netmod mailing list [email protected] https://www.ietf.org/mailman/listinfo/netmod
