related to ISSUE-181

From: Little, Chris [mailto:chris.little@metoffice.gov.uk]
Sent: Thursday, 13 April, 2017 04:02
To: p.barnaghi@surrey.ac.uk; Cox, Simon (L&W, Clayton) <Simon.Cox@csiro.au>
Cc: phila@w3.org
Subject: RE: Wide review - help needed

Hi Payam, Simon,


I a   I am not sure whether the durations mentioned (yearMonthDuration, dayTimeDuration) solve the right problems. I need to delve deeper, but they seem tied to the Gregorian calendar minus leap seconds, and I am not sure that they address durations in times encompassing different transitions from the Julian to Gregorian calendars (multifarious dates between 1588 and 1923).



I       I think that the dates selected for testing durations (1696 and 1903) are related to the Metonic cycle for calculating the days of the week in any given year.


The new XSD durations certainly do not help with other calendars, such as Mayan or Martian.



M   Also, my naïve interpretation of the regular expressions is that lexical values like P1Y13M (=P25M =P2Y1M) are allowed. I am not sure whether that this is good or bad.



P     Perhaps the issue of “why using the XSD datatypes for dates, times, and durations is not (good) enough” can be addressed by adding one sentence to end of Section 3.2:



“      “For example, ISO8601 style notation is often used, but ignoring leap seconds, which are explicitly mandated by the international standard, or a schema is defined in which date strings like 2017-99-99 validate correctly.”



C       Chris



    PS apologies but my  left hand  margins have gone peculiar.