Re: [netmod] [Technical Errata Reported] RFC7950 (5663)

[Martin Bjorklund <[e-mail]>]

Ebben Aries <[e-mail]> wrote:
> On Mar 19 23:27 PM, Robert Varga wrote:
> > On 19/03/2019 19:12, Rob Wilton (rwilton) wrote:
> > > Hi Ebben,
> > > 
> > > I've always taken the ABNF to list the definitive sub-statements that are allowed for the various deviate "add", "replace", or "delete" options.  Perhaps the RFC could state this more explicitly.  Perhaps raise an issue on the YANG Next issue tracker to clarify this (https://github.com/netmod-wg/yang-next/issues) and it might get discussed tomorrow.
> > 
> > I agree.
> > 
> > Proposed statements are simple cases, for which 'deviate replace' can be
> > used to specify the correct value -- for example remove 'min-elements'
> > by replacing it with 'min-elements 0'.
> 
> Sure - my point was rather that in either case we have an issue.  The
> table of substatements in 7.20.3.2 is either not accurate or we modify
> the grammar to match.

I agree that the document needs clarification, and the yang-next issue
will take care of that.  The document needs a clarification that the
refers to the grammar, or perhaps different substatement tables for
add/replace/delete.

Meanwhile, I think that this errata should be rejected.


/martin




> 'deviate replace' can be used to 'reverse' these
> substatements much like a 'delete' would as you point out but the
> wording in this section should state this - I'll raise an issue on the
> tracker
> 
> FWIW - pyang does not honor the grammar and allows for a mandatory
> substatement of 'deviate delete' while yanglint appears to follow the
> ABNF strictly
>