From: Rintze Zelle
Thanks for the additional corrections, they have been applied.
>> "For year-month, month may instead be "season" (i.e. 21, 22, 23, or
> >> 24)
> however, in that case the entire year-season combination is qualified;
> neither may be qualified individually."
> Does this mean that in the example "2011-23~", year and season are
> approximate? The current explanation, "Approximate season (Autumn 2011)",
> suggests that only the season is approximate.
The intent is that a season never stands alone, thus by "season" we mean a
year-season combination. For example, "spring", the season, has no meaning
in isolation (in the context of this spec). So the expression "Approximate
season" implicitly means "Approximate year-season combination".
> "the character 'u' may be used in place of one or more digits"
> I have the same comment here I had at 5.2.2: the current description is
> unclear about whether or not a single "u" can replace multiple digits
Ok, I've clarified this.
> Despite what's said in "6. Sort Order", shouldn't dates in square brackets
> always be sorted (ascending, oldest to newest)?
No, there isn't any such requirement.
....Compare [1760-01, 1760-12..]
> to [1760-12, 1760-01..] (the latter is valid according to the spec)
Yes, at this stage of the development of this spec, it's valid. You could
bring this issue up during the next phase.
> The term "scientific notation" might be more precise than "exponential
Well it's not really scientific notation that we're prescribing, which we
see as a special case of exponential notation - roughly, scientific notation
is normalized exponential notation - though admittedly these definitions
tend to get mixed up. We did have some discussion of this: scientific
notation requires a decimal point with one digit preceding and the remaining
significant digits following. You'll note in the BNF that there is no
decimal point, just an integer and exponent. You can see some of the
discussion on this, go to http://listserv.loc.gov/cgi-bin/wa?S1=datetime and
search on "scientific notation". One of the arguments was to avoid
constructs such as 1.2345e3 which would evaluate to 1234.5, which is not a
valid year (and this is used only for years).
> >> "Seasons may be sorted with Spring < Summer < Autumn < Winter. Thus
> "2011-21" should sort before "2011-23", but applications may choose to
> "2011-21" and "2011-10" either way"
> I find the use of both "may" and "should" confusing. Which one is it?
As it says in section 2, these are defined in RFC 2119
> Also, reading the whole spec it is unclear to me whether time strings can
> be used with any type of date (e.g. uncertain/approximate/unspecific
> or dates with only year or month precision). The BNF allows time strings
> after any date string.
No (I don't think so). "date" is defined in level 0. It does not include
uncertain, etc. Those things are defined only in levels 1 and 2.
"dateAndTime" is also defined in level 0. But "dateAndTime" is never
referenced in levels 1 and 2.