As of the examples at #002, a zone-offset may be appended to a time, such as
2004-01-01T10:10:10+05:00
According to the BNF, such a construct is valid, but the day or monthDay
is optional. The BNF allows both
2004-01T10:10:10+05:00
and
2004T10:10:10+05:00
which is ... "undocumented" in the specification. My guess is that this is
not what we want, is it? At level zero, I suggest to rewrite dateAndTime as:
dateAndTime = yearMonthDay "T" time
But, what do we want at level one and level two? Do we want to be able to
include zone-offsets in lists and intervals?
It is also impossible to specify a zone-offset without an expressed,
certain and non-approximate "second precision". More exactly, all these
are excluded
2004-01-01T10:10:(10)?+05:00
2004-01-01T10:uu:uu+05:00
2004-01-01T(10)~+05:00
etc. according to both the specification and the BNF. Maybe that is not
exactly what we want ...
