On Tue, Oct 4, 2022 at 2:15 PM Randy Presuhn < [email protected]> wrote:
> Hi - > > On 2022-10-04 12:15 PM, Jürgen Schönwälder wrote: > ... > > I am hoping for a technically sound solution that provides predictable > > behaviour of independently developed tools. Some "hints" that may be > > used or ignored at the discretion of an implementation do not really > > meet that bar. > ... > > There are conflicting goals at work here. I think we all know this, > but sometimes we need to remind ourselves: > > (1) wanting to maximize the likelihood that a system assembled from > various bits and pieces will do what is intended. This is where > developers' minds are at most of the time, and as long as the versions > of the bits and pieces play nicely together, developers don't (want to) > think about it. > > (2) wanting to know (and control) *exactly* what bits and pieces went > into a particular configuration. This is not just a concern for the > folks generating a shippable "build" of a product, but also for > operations folks who want to keep the number of systems configurations > they have to deal with in some sort of bounds. > > The difference between the two is that the former is a mesh of > many-to-many potential compatibility relationships, some of which my be > speculative (e.g. versions that do not exist yet) while the latter > is a concrete list of known implemented components. > > We conflate YANG Conformance with goal (2). We assumed (incorrectly) in RFC 6020 that YANG does not need any form of explicit conformance statements. The conformance will be self-describing. A YANG module is used in a server within a YANG Library. All the modules, revisions, features, and deviations are known. But this represents the "server capabilities", not the conformance requirements. We have not made much progress in defining useful conformance information that can be applied beyond a specific server implementation. (Goal (1)) YANG Packages can help by moving the operational focus to large functional blocks instead of 100s of YANG modules. Both have value, but they're quite different beasts. > > Randy > > Andy > _______________________________________________ > netmod mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/netmod >
_______________________________________________ netmod mailing list [email protected] https://www.ietf.org/mailman/listinfo/netmod
