Minutes for NETMOD at interim-2015-netmod-16
|Meeting Minutes||Network Modeling (netmod) WG|
|Title||Minutes for NETMOD at interim-2015-netmod-16|
|Other versions||plain text|
============================================================= NETCONF Data Modeling Language WG (netmod) 19th YANG 1.1 Virtual Interim Monday, August 31st, 2015, 17:00-18:00 CEST Minutes Juergen Schoenwaelder ============================================================= * Participants DB = Dean Bogdanovic JS = Juergen Schoenwaelder KW = Kent Watsen LL = Ladislav Lhotka MB = Martin Bjorklund MJ = Mahesh Jethanandani * Definition of 'data tree' One option is to define 'data tree' to capture the choice currently embedded in the xpath expression evaluation section. This will make the xpath section way simpler. It is unclear whether this also brings simplification at other places. Is it a goal to remove any reference to <running/> in normative text? Note that RESTCONF does not have <running/> but instead a unified datastore (but it seems this got lately changed again). Observation: YANG does not define terms like 'datastore' or 'configuration datastore' or <running/>. RFC 6020 is silently assuming the import of these terms from RFC 6241. MB: I suggest to import the necessary terms. JS: I agree that key terms should be defined in the document or the definition should be imported. In an ideal world, we would have an architecture document defining architectural concepts and then YANG and protocol specifications would import from there. We can only slowly move to an ideal world in the IETF. KW: Should we refer to an 'intended' configuration datastore instead of the running configuration datastore? LL: Perhaps 'active' configuration with 'active' meaning 'running' on a traditional NETCONF server? KW: running + ephemeral => intended => applied MB: YANG 1.0 clearly uses running. If we create a new term, it must resolve to <running/> in NETCONF. LL: The issue for me is config false node refering to a config true node. With possible future extensions of NETCONF (or RESTCONF), referring to <running/> may not be useful anymore. MB: The notifications in the SMIv2 mapping to YANG make use of references to <running/> datastore content. JS: But this is accomplished via leafrefs, not via xpath expressions. Resolution: We keep the choice text in xpath section. We will replace <running/> with 'running configuration datastore'. What about data tree? It seems to make sense to expand its definition but we need to figure out where the expansion would not be OK. LL volunteers to take a look. MB will do this as well.