W3C home > Mailing lists > Public > www-xml-schema-comments@w3.org > October to December 1999

More comments on time related datatypes

From: Arnold, Curt <Curt.Arnold@hyprotech.com>
Date: Mon, 8 Nov 1999 15:08:15 -0700
Message-ID: <61DAD58E8F4ED211AC8400A0C9B46873415490@THOR>
To: "'www-xml-schema-comments@w3.org'" <www-xml-schema-comments@w3.org>
3.2.4.1 time zone offset

In the second paragraph, you say time zone offset is + or - followed by
hhmm.  The example uses hh:mm.  Either would be acceptible ISO usage,
however for consistency with using the extended form else where I would
recommend hh:mm.

3.2.5.1 timeDuration lexical representation

b) The current ISO 8601 section 5.5.3.2.1 is ambiguous whether a P is
required before a time period.  It could be implied from section 5.5.2 which
says that a P should preceed a duration.  Section 5.5.4.2.2 in the draft
revision is explicit that a P is to preceed that alternative format.  The
example would then be P0001-02-03T10:30:00
D.3.2 What would be the interpretation of a negative timeInstant (years
BC?).  
What is the interpretation of a recurringInterval with a negative duration?
I think you must constrain the period of recurringInterval to non-negative
durations or explicitly state that a negative sign is ignored (or switch to
the primary ISO representation of durations).
Applications that want to adhere to ISO8601 can create derived datatype's
with minInclusive of 0 to prevent the occurrence of negative dates.
D.3.3 Lexical Representation for Time Period
Time Period was introduced as a concept in 3.2.5.1 but was not implemented
as a type.  I think you meant this section to be titled "Lexical
Representation for timeDuration".  If you prepend the P, then the
timeDuration format proposed would be the alternative form for timeDurations
in section 5.5.3.2.1 of the current ISO draft.  
Why was month == 30 days concept added.  If you do this, then there is not a
way to create a data type that represents a specific day in each month
(since <period>0000-0100</period> would be interpreted as
<period>0000-0030</period>.  If you do this, why not year==365 days, century
== 36500 days.  
The alternative form in IS0 8601 does have problems if you wanted to say a
duration of 300 days.  Though it seems to save a little parsing problems, I
think that we would save a lot of long term problems by sticking with the
primary ISO representation of time periods.  How much more complex could
validating periods against PnYnMnDTnHnMnS be?
p.s. There is a misspelling of digit as dgit in the description of C in
section D.1.
Received on Monday, 8 November 1999 17:11:11 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 23:08:45 UTC