----- Original Message ----- From: "Juergen Schoenwaelder" <j.schoenwaelder@xxxxxxxxxxxxxxxxxxxx> Sent: Friday, March 11, 2016 1:07 PM > On Fri, Mar 11, 2016 at 11:15:28AM +0000, tom p. wrote: > > Lada, Robert > > > > The other angle from which this might be approached is that the I-D > > already says > > > > " Using the "type" statement, a type is specified for the annotation > > value according to the same rules as for YANG "leaf" type. " > > > > while rfc6020bis says > > > > " The "leaf" statement is used to define a scalar variable of a > > particular built-in or derived type." > > > > so if you know your YANG off by heart, then you will know that > > annotations must be scalar. I agree that the text needs to be clearer. > > Perhaps, > > OLD > > " o annotations are scalar values and cannot be further structured;" > > NEW > > "Annotations obey the same rules as for a YANG "leaf" type [rfc6020bis > > s.7.6] and so are limited to scalar variables." > > There is no 'leaf type' in YANG. YANG has leaf nodes in the schema > tree. An annotation is not a node in the schema tree. Perhaps > something like this: Juergen Well, I know, but I was quoting directly from yang-metadata-04 s.3, namely "Using the "type" statement, a type is specified for the annotation value according to the same rules as for YANG "leaf" type. " which is why I gave a reference to s7.6 of RFC6020bis rather than s.7.4. Perhaps change s.3 in addition to your change OLD Using the "type" statement, a type is specified for the annotation value according to the same rules as for YANG "leaf" type. NEW Using the "type" statement, a type is specified for the annotation value according to the same rules as for the type of a YANG "leaf"[RFC6020bis s.7.6]. I do think that that mention of leaf is helpful - as you say, the WG agreed to this restriction as opposed to allowing more complex annotations and referencing "leaf" for me makes that clearer. Tom Petch > An annotation carries a single value. The type substatement, which > must be present, takes as an argument the name of an existing > built-in or derived type and the value of the annotation must match > this type. See Section 7.4 of [RFC6020bis] for details. > > /js >