W3C home > Mailing lists > Public > www-xml-schema-comments@w3.org > January to March 2001

Problem in canonical representation of "time" type.

From: Kohsuke KAWAGUCHI <kohsuke.kawaguchi@eng.sun.com>
Date: Mon, 19 Mar 2001 16:37:50 -0800
To: www-xml-schema-comments@w3.org
Message-Id: <20010319162335.6A65.KOHSUKE.KAWAGUCHI@eng.sun.com>

I found another problem this time in "time" type.

What is the canonical representation of "08:00:00+09:00" (morning in
Japan)?  It should be something like "-01:00:00Z" (11PM of the previous
day) but this is not in the value space of "time" type.

According to the spec, this will be "23:00:00Z" (by the rule that unspecified
fields are removed after addition.)

Although It looks harmless at the first glance, this is very problematic.

If you are living in Japan, you may want to do the following:

<restriction base="time">
  <!-- our business hours -->
  <minInclusive value="08:00:00+09:00" />
  <maxInclusive value="17:00:00+09:00" />

I would like to prove that this restriction violates the schema

Let us normalize these two values into the canonical representation.
This shouldn't change the meaning, and in fact most of datatype
validators will do this internally anyway.

The result of normalization will be

<restriction base="time">
  <!-- our business hours -->
  <minInclusive value="23:00:00Z" />
  <maxInclusive value="08:00:00Z" />

This violates "minInclusive<=maxInclusive" constraint, which is defined
in 4.3.7.

So I think this is a problem of the spec.

E-Mail: k-kawa@bigfoot.com
Received on Monday, 19 March 2001 19:37:47 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:49:54 UTC