Re: [Jcardcal] Genart LC review: draft-ietf-jcardcal-jcal-09

[Peter Saint-Andre <[e-mail]>]

On 3/20/14, 4:34 PM, Pete Resnick wrote:
> On 3/20/14 5:20 PM, Robert Sparks wrote:
>>>> (This may be more than a nit): In the ABNF in section 3.6.5, where is
>>>> the implementer supposed to go to find the definition of 'zone'? (Or
>>>> the other production names?) I think _this_ chunk of ABNF (as opposed
>>>> to that compiled in the appendix) is intended to be normative, yes?
>>>> FWIW, it's not reflected in Appendix B.
>>> Indeed, there are some shortcomings here. For the
>>> year/month/day/hour/minute/second production names I can add an explicit
>>> reference to ISO.8601.2004 Section 2.2. The zone designator is only
>>> marginally mentioned in ISO.8601.2004 Section 4.3.2.
>>>
>>> The exact date format is not reflected in Appendix B because the
>>> Appendix does not intend to capture the structure of each different
>>> property type.
>>>
>>> As the two jcardcal drafts are my first rfc documents, I am not quite
>>> sure when its a good idea to make the ABNF normative. I could probably
>>> declare it informative by virtue of the reference to ISO 8601 and
>>> RFC5545. The ABNF was added in draft 08, in alignment with the changes
>>> we've made in jcard (rfc7095). There was a lot of discussion on the
>>> jcardcal list on the date formats and it became clear that the specific
>>> variations of ISO 8601 2001 and 2004 used in iCalendar and vCard make it
>>> hard to grasp. By explicitly mentioning the date format I wanted to
>>> counteract.
>>>
>>> I'd appreciate some feedback on how to further handle this issue.
>> Please work with your AD on this one.
>> It's a little detail, not a big problem, but we do need to be clear
>> about what the grammar actually is.
>
> Hmm....I wonder why neither RFC 5545 nor this document reference RFC
> 3339 instead of ISO 8601? That would get you all of the ABNF you need.

That's a good point. RFC 3339 is, IMHO, nicely clear and tightly scoped 
compared to ISO 8601 (which leaves quite a few options open). However, 
even when referencing RFC 3339 it's important to document all the 
details about various options, such as use of non-UTC time zones and 
fractional seconds.

Peter