On 02/05/17 10:25, Martin Bjorklund wrote: > Robert Varga <[email protected]> wrote: >> On 20/04/17 18:35, Andy Bierman wrote: >>> >>> Yes -- it looks correct. >>> The structure is defined in RFC 6020: >>> https://tools.ietf.org/html/rfc6020#section-5.6.4 >>> >> >> Hello Andy, >> >> One thing that is not quite clear in RFC6020 due to not being clear what >> 'supported module' means. > > Right; specifically the difference between "implement" and "support > for import", which is clarified in RFC 7950 and RFC 7895. > >> Should deviations be applied in case a module is advertised, but is not >> mentioned in a deviations parameter like this: > > No. > >> <hello xmlns="urn:ietf:params:xml:ns:netconf:base:1.0"> >> <capability> >> http://example.com/syslog?module=syslog >> </capability> >> <capability> >> http://example.com/my-deviations?module=my-devs >> </capability> >> </hello> >> >> If not, does this imply that any deviation statement targeting an >> external module, which does not have a matching deviations parameter >> should silently be ignored? > > Yes. Just like a grouping can be defined without ever being used. > > For example, a device might put all its deviations in a > single module, but then list it in the deviations parameter for a > subset of all potential modules.
Thanks Martin, this does make sense, but there seems to be little implied about this in the text of https://tools.ietf.org/html/rfc7950#section-7.20.3. Furthermore https://tools.ietf.org/html/rfc7950#section-7.20.3.3 is a bit misleading, as it states: A server would advertise both modules "example-base" and "example-deviations". Based on your reply, it is not sufficient to advertise both modules as 'implement', but rather for 'deviation' statements to be effective, 'example-deviations' needs to also be present in example-base's advertised deviation list. Maybe 7.20.3 could use a paragraph describing interplay with advertisement and the text in 7.20.3.3 could be made more explicit? Regards, Robert
signature.asc
Description: OpenPGP digital signature
_______________________________________________ netmod mailing list [email protected] https://www.ietf.org/mailman/listinfo/netmod
