RE: date in user's timezone

> You are missing the point!  By allowing multiple timezone specifications,
> the same date/time can be specified in multiple ways.  The canonical
> representation seeks to aviod this by fixing on a single timezone.
> We happened to pick UTC but any other would have been just as good as
> long as you had to specify the date/time using just that one timezone.

Ashok,

You fellas are a barrel of monkeys. I will regret it when this thread
finally comes to an end!

I think I get the point you are making. I'm pointing out the opposite
problem:

 ==> Values that have no canonical representation! <==

We don't have to solve this problem using timezones, but I think it is the
obvious and intuitive solution.

First of all: Are we reading different specs? Mine is dated 24 October. It
says (3.3.27.1) "The lexical representation for date is the reduced (right
truncated) lexical representation for timePeriod: CCYY-MM-DD."

If that is not what yours says, I stand corrected. Please give me a link to
the new one.

Otherwise,

If 3.3.27.1 is supposed to include time-of-day components then I object to
the way it is worded and to the sample text-pattern. If time-of-day
components are allowed, then what does "right-truncated" mean? And why does
the sample text-pattern omit them? Adding time-of-day components as the fix
for the problem would further confuse the situation because the definition
clearly says that the period begins at midnight on the specified day. The
second paragraph of D.2 also makes it clear that the time-of-day components
are not present in a date, month, year, or century.

Incidentally, I believe it is wrong in said paragraph not to mention
recurringDay and recurringDate. They have lexical representations based on
that of date and do not allow for time-of-day components.

I appreciate that there must be just one canonical way of representing, as
an object of type "date", the 24-hour period that begins and ends at
midnight Pacific Daylight Time and spans the day commonly known as 26
October 2000. I would just like to know what that one canonical way is. This
ought to be a simple question, but I say it cannot be done within the spec.

Try it, please! As I read 3.3.27.1 and .2, the answer must be of the form

    dddd-dd-ddZ

where each 'd' is a decimal digit. Please end this thread by picking eight
digits that yield the proper meaning and including them in the next post!

Graham

Received on Friday, 27 October 2000 19:29:59 UTC