On Mon, 6 Jun 2011 15:53:51 -0400, Ray Denenberg, Library of Congress wrote
> From: Bruce D'Arcus
> > One way to slice the levels is how different feature diverge from the
> > base. For example, a standard range of "2002-02/2002-03" is not
> > supported by Level 0, but it's components are.
> That's a disturbing observation as I suppose I didn't realize that 8601
does not support "2002-02/2002-03". Thus features 208-211, which are
represented as being 8601 conformant, aren't.
I only have a copy of the 2000 draft.. but..
220.127.116.11 defines these truncated dates.
5.5.1 gives pattern a) as an interval: start/end where start and end are
Since the terms "truncated representation" is used in this section there is
no reason to assume that one their intent was not to allow for the use
truncated representations for start and end. While 5.4.1 is specified for
date it was I think an oversight.
I think we should "fix" this oversight and allow for start/end where start
and end are ISO8601 dates (including truncated representations) in any
minimal level--- I will assume that most people's implementations of 8601
have been doing this the whole time anyway.
> ISO 8601 is a horribly frustrating document to try to make sense out. It
makes no common sense to me why 8601 wouldn't want to support "2002-02/2002-
03"; I can see that it doesn't, but only by the lack of an example to the
contrary. (At least as far as I can tell - and given 8601's general lack of
examples, deciphering it based on [UTF-8?]it’s examples is not a
No. It clearly says in 18.104.22.168 that intervals defined with start/end use
5.4.1. It does not explicitly, however, disallow 5.4.2. (reduced
representations) but does not explicitly allow them either. I see this as an
oversight. Should reduced representations not have been desired they would,
I think, have been explicitly disallowed.
> But I digress. If "2002-02/2002-03" is not a valid 8601 string while
both "2002-02" and "2002-03" are, then I agree that it does make sense to
cast "2002-02/2002-03" into level 1 rather than level 2. On the other hand,
that seems to be the only feature that would fit this criterion and all else
would be cast into level 2, which would make level 1 rather slim.
Edward C. Zimmermann, NONMONOTONIC LAB