Thanks for the detailed explanation, Ray. I thought calling EDTF a
"profile" of 8601 just meant that redundant ways to represent the same
information would be removed; I didn't realize the scope of what could
be represented might be reduced. But I still have some questions.
First, you said that no one spoke up for retaining fractional seconds,
but the first URL you give is a message from Edward Zimmermann in which
he says: "8601 uses hh:mm:ss.fff where fff are milliseconds.. Why not
where ss are rational seconds. e.g. ss including decimal fractions
without restriction, e.g. allowing for times such as 12:21:44.5 and
12:59:59.1001". So he _did_ speak up for fractional seconds! In fact,
he asked for any number of decimal places -- which is actually what
8601 allows, not just milliseconds (i.e., exactly 3 decimal places),
according to two descriptions of it I have.
And second, I have the impression that EDTF Level 0 can express
virtually everything 8601 can express _except_ fractional seconds
(though with fewer ways to express a lot of things). If I'm right, it
seems like a real mistake to remove fractional seconds in EDTF Level 0.
Especially since Levels 1 and 2 don't add them back in! Of course, as
you say, there will be a follow-up opportunity to propose more
features, but even so.
On Thu, 14 Feb 2013 15:54:02 -0500, Ray Denenberg <[log in to unmask]> wrote:
> Hi Don - Yes, maximum precision is one second. It wasn't always so, early
> drafts supported fraction of second in the same manner as ISO 8601.
> It was never the intention of this effort to fully support 8601. It tries to
> be a profile/extension of/to ISO 8601, As it says in the abstract:
> "ISO 8601 describes a large number of date/time formats. On one hand some of
> these formats are redundant and/or not very useful; to reduce the scope for
> error and the complexity of software, it seems worthwhile to restrict the
> supported formats to a smaller set. On the other hand, there are a number of
> date and time format conventions in common use that are not included in ISO
> 8601; it seems worthwhile to normalize these."
> And so, early drafts had quite a lot of features that did not make the cut.
> We went through a process of questioning each feature, and if nobody spoke
> up for a feature, it was removed.
> (For "fraction of a second", see
> 9F159E636F2A03&Y=rden%40loc.gov where I asked
> "For time, is hour:minute:second sufficient, or do we need to support
> fraction of a second?"
> Nobody responded, so it was removed.
> That's not to say that fraction of second is not a useful feature, it simply
> means that nobody who has participated in this round of the development of
> the spec found it useful. There will be a follow-up opportunity to propose
> additional features.
>> -----Original Message-----
>> From: Discussion of the Developing Date/Time Standards
>> [mailto:[log in to unmask]] On Behalf Of Byrd, Donald A.
>> Sent: Thursday, February 07, 2013 11:41 AM
>> To: [log in to unmask]
>> Subject: [DATETIME] Precision better than one second
>> I may be blind, but as far as I can tell from both the text and the BNF,
>> EDTF's maximum precision is one second. OK, That's considerably more
>> limited than ISO 8601, which allows unlimited precision! For example,
>> "13:10:30,7" is 30.7 sec. after 13:10, with precision of 1/10 second. And
>> some people seem actually to be using 8601 with millisecond resolution --
>> for example, see the Joda.org website:
> http://joda-time.sourceforge.net/cal_iso.html . I assume that this is
>> either my mistake or a mistake in the EDTF proposal? Surely it's not a
>> idea to add any further limitations to 8601.
Woodrow Wilson Indiana Teaching Fellow
Adjunct Associate Professor of Informatics
Visiting Scientist, Research Technologies
Indiana University Bloomington