From: Rintze Zelle
> Okay, so sorting the dates within square brackets seems to be impossible
> when there are dates with seasons around.
No. It says:
"When years, months, and seasons are mixed together in a list to be sorted, this specification does not address the sort order, ....."
That doesn't mean you can't do it, it means that it is outside the scope of the spec to say how to, that it is application-specific.
' ....that is, whether (for example) 2000 is before or after 2000-01, 2000-10, 2000-12, 2000-21, etc. An application may adopt whatever sorting algorithm fits its purpose, thus the value '2000', for purposes of sorting, could be considered to be before 2000-01-01, or after 2000-12-31, or before or after 2000-24, etc. Seasons should sort as Spring < Summer < Autumn < Winter (e.g. "2011-21" should sort before "2011-23"), but applications may choose to sort "2011-21" and "2011-10" either way, or consider them incomparable. '
> A) [1760-12, 1760-01..]
> Should this be read as
> 1) "January or December 1760, or some later month"
> 2 )"December 1760, or some month after January 1760" ?
The second. See below.
> B) [1760-21..]
> "Spring of 1760 or a later season"
> "Spring of 1760 or a later year"?
The only reasonable interpretation is the first, however I agree that this is not explicitly specified. I am not going to try to fix this, though, because it would cross the line from minor correction/clarification to substantive change. I will put it on the list of issues to be addressed in the next phase.
> C) [1760-06, 1760-21..]
> "June of 1760 or a later year"?
> D) [1760, 1760-2]
> "February of 1760 or later"
> "1760 or later"?
No. The first means, simply, "June of 1760 or Spring of 1760". The second, "the year 1760, or the month February of 1760". You can say that such examples are non-sensical, and maybe they are, but to disallow all nonsense examples would probably complicate the spec unnecessarily. (None of these examples is actually included in the spec, as far as I can see. Yes, they are legal, but the spec does try not to include nonsense examples.)
> I think the specification should either be clear about how to interpret the
> above examples, or simple do not allow them. If the specification stays as
> it is on the topic of sorting, the only way to use square brackets
> unambiguously seems to:
> a) disallow the joined use of year-season dates and the double-dot
> b) require that all dates are of the same precision when the double-dot is
> used With these two requirements, it becomes possible to unambiguously sort
> dates within square brackets and have the double-dot target the last and
> most recent date.
Well, again, I am not sure that the benefit of disallowing them outweighs the complexity of doing so, but this issue can be revisited in the next phase.