Comments on XML Schema 1.1 part 2 WD of 2006-01-16

The attached are comments developed by John Cowan,
a member of the XML Core WG.  The WG discussed these
comments, and we are in agreement for the most part,
though there was some dissent in some areas.  However,
overall, we felt it would be more helpful for us to
send these comments in now.

Paul Grosso

-----Original Message from John Cowan-----

Subject: Comments on XML Schema 1.1 part 2 WD of 2006-01-16


I have the following comments on XML Schema 1.1 part 2 WD of 2006-01-16:

1) In sections 3.3.8.1 and 3.3.9.1, it is said that a time without a
timezone is incomparable to a time with a timezone within a 28-hour
period, since the range of possible timezones is from GMT+1400 to
GMT-1400.  This has only been true since 1995, however, when the island
nation of Kiribati changed two of its three timezones so that the entire
nation would be on the same day at the same time.  If the plan to
abolish
leap seconds comes to fruition, we may see further adjustments in time
zone in many countries.  I suggest therefore that more liberal bounds
be set, such as GMT+2400 to GMT-2400.

2) In section 3.4.3, references are made to [RFC 3066] or its
successor(s)
in the IETF Standards Track.  This is language that we introduced into
the XML Rec, and is now copied into various places including here.
Unfortunately (and I take much of the blame for this) it is wrong and
has always been wrong.

RFC 3066 is not on the Standards Track, nor was its predecessor RFC
1766.
The right way to refer to the evolving IETF language tagging rules is as
[IETF BCP 47].  This is a "logical name" which is always bound to the
current language-tag RFC, whatever it may be.  We should likewise fix
XML 1.0, XML 1.1, and possibly other Recs that we control.  This is
particularly important now that RFC 3066 has been superseded, though
its successor has not yet been published or assigned an RFC number.

3) I urge the Schema WG to adopt the proposed rules by which BCE year
numbers are 0 for 1 BCE, -1 for 2 BCE, and so on.  The 1.0 rules do
not have a year 0, and use -1 for -1 BCE, -2 for -2 BCE, and so on.
While superficially clearer, the 1.0 model is irritating when doing
year arithmetic.

4) J.1.1: for "they non-identical" read "they are non-identical".

5) We should urge the Schema WG to adopt Henry Thompson's approach to
XML 1.1 validation; that is, the meaning of the types Name and NCName
and of regular expressions including any of \i, \I, \c, or \C depends
on the version number of the XML document being validated.  The version
number of the schema is of course irrelevant.

Received on Wednesday, 22 February 2006 22:16:56 UTC