Re: [Netconf] [netmod] date-and-time vs. dateTime

[Andy Bierman <andy at netconfcentral.com>]

Juergen Schoenwaelder wrote:
> On Tue, Jun 09, 2009 at 05:28:22PM +0200, Andy Bierman wrote:
>  
> > The subject line says date-and-time and that is the issue here.
> > Remove it, change it, or change ALL of the NETCONF content
> > that uses dateTime to perfectly align with the YANG language.
> >
> > I was hoping we were going to avoid the same sort of compartmental
> > politics that plagued EOS/SMIng.  I guess not.
>
> YANG started late and some data models were started before and this
> explains the situation (and this particularly applies to RFC 5277,
> published in July 2008). I do not see compartmental politics - just
> the IETF working as it is.
>
> Concerning RFC 5277, I like to quote section 2.2.1:
>
>    Parameters:
>
>       eventTime
>
>          The time the event was generated by the event source.  This
>          parameter is of type dateTime and compliant to [RFC3339].
>          Implementations must support time zones.

This text is why I proposed an RFC Errata to fix the problem.
I do not think the NETCONF WG understood the differences
at the time.  However, the XSD clearly uses dateTime as
the data type, not string.

RFC 5277 does not say what to do with startTime and/or stopTime
with negative dateTime. Is -2010-06-09T12:00:00Z before 'now'?
(What were the XSD designers thinking!?!)

So problem solved.  The text you quoted above is correct
and the XSD is wrong, and an RFC Erratta note is needed
for RFC 5277 after all. From now on, all time values
in NETCONF/YANG MUST be date-and-time.



> This says eventTime is compliant to RFC3339 and thus I believe it is
> actually compliant with date-and-time (but perhaps the WG was not
> aware of the corner case differences between XSD dateTime and RFC3339
> ietf-yang-types is pointing out).
>
> The more interesting question for me is whether eventTime is part of
> the content or part of the protocol. For me, it seems to be more part
> of the protocol as it is hardwired to the <notification> message but
> I understand that this piece of the design can be debated.

eventTime is another problem, but at least it is
read-only and the agent generates only canonical
format, so the differences never show up.


> /js

Andy