This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
The order relation on date is specified as: "If date values are considered as periods of time, the order relation on date values is the order relation on their starting instants." This makes sense to me. The order relation on time is specified as: "The order relation on time values is the Order relation on dateTime (3.2.7.3) using an arbitrary date." This also makes sense to me. Now consider the order relation on eg gMonthDay: "If gMonthDay values are considered as periods of time, the order relation on gMonthDay values is the order relation on their starting instants." I don't think this is quite right. A gMonthDay is not a single period of time but a recurring period. The dateTime ordering relation compares two specific instants of time. Thus in order to turn a gMonthDay into a specific instant of time, you need to use an arbitrary year (just as with time you need to use an arbitrary date). However, I'm guessing --02-29 is allowed as a gMonthDay; if so, the year is not an arbitrary year but an arbitrary leap year. So the spec should say something like: "If gMonthDay values are considered as periods of time using an arbitrary leap year, ..." Similarly, for gMonth and gDay. See: http://lists.w3.org/Archives/Public/www-xml-schema-comments/2002AprJun/0045.html
Discussed at May 31 telcon. Resolved to classify as error w/erratum. Proposed text: http://lists.w3.org/Archives/Member/w3c-xml-schema-ig/2002Jun/0011.html Text approved at June 14 telecon: http://lists.w3.org/Archives/Member/w3c-xml-schema-ig/2002Jul/0009.html Erratum E2-26 added.