Saašha, I do think the three-level suggestion has merit and is worth considering further.
The spec could be represented as:
Level 0: a profile of 8601
Level 1: first-level extensions
Level 2: second level extensions
And to claim conformance, you must at least support level 1 (support for level 2 includes support for level 1).
Level 0 would be the 100 and 200 features.
For level 1, I suggest:
- uncertain/approximate excluding internal.
- intervals, excluding those with uncertain/approximate and temporal expressions, but including open and unknown.
- masking with "u"
- Lists (one of a set, all of a set)
- internal uncertain/approximate
- temporal expressions
- long year
- masking with "x"
Please comment. I will hold off on further BNF changes pending some agreement on this.
> -----Original Message-----
> From: Discussion of the Developing Date/Time Standards
> [mailto:[log in to unmask]] On Behalf Of Bruce D'Arcus
> Sent: Wednesday, May 25, 2011 9:07 AM
> To: [log in to unmask]
> Subject: Re: [DATETIME] A three level suggestion
> On Wed, May 25, 2011 at 8:38 AM, Saašha Metsärantala <[log in to unmask]>
> > Hello!
> > I wonder what you think about the following suggestion.
> > Keeping in mind that EDTF is thought of as
> > "both a profile of and extension to ISO 8601"
> > according to
> > http://www.loc.gov/standards/datetime/spec.html
> > we could skip "reinventing the wheel", define the first EDTF level as
> > a profile of ISO 8601 and just add some constraints on ISO 8601 to
> > build the first level of EDTF. This could make both the BNF and the
> > coming regexes easier to write, just carving away what we do not want
> > Thereafter, we could have a second level thought of as an extension
> > the first level. Thus, we could use the BNF just to add features to
> > the first level. I'm particularly thinking of lists, "x", longYears,
> > seasons and temporal expressions. There would not be any "uncertain,
> > approximate, unspecified" here. Well, ... "temporal expressions" and
> > seasons may contain a kind of approximation, but I suggest to place
> > them in the second level anyway.
> - where would intervals go?
> - not clear why 'x' is here and not below?
> > Thereafter, we could have a third level thought of as an extension of
> > the second level. Thus, we could use the BNF just to add features to
> > the second level. I'm particularly thinking of "?", "~" and "u".
> > we would introduce "uncertain, approximate, unspecified".
> > Comments are welcome!
> > Regards!
> > Saašha,