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 datatypes for recurring calendar events (specifically gMonthDay, gDay and gMonth) are all recorded in appendix A with: <hfp:hasProperty name="cardinality" value="countably infinite"/> However it appears that there are at most 366 days in a year, and at most 59? timezones or maybe 290000 if we take an extremist view of the possible timezones. (Alternatively, the lexical form is of finite length and comes from a finite vocabulary, hence there are finitely many different lexical forms). Is this an oversight? See: http://lists.w3.org/Archives/Public/www-xml-schema-comments/2002OctDec/0068.html
Discussed at the 2003-11-20 telecon. Agreed to classify as error w/erratum. Agreed to change countably infinite to finite. Ashok to draft text.
And whenever we change cardinality to finite, we also need to change bounded to "true".
Proposed fix: (1) in the pseudo schema document with declarations for the types gYearMonth, gYear, gMonthDay, gDay, and gMonth (currently labeled "primitives.nxsd", but that name is subject to change), change <hfp:hasProperty name="bounded" value="false"/> <hfp:hasProperty name="cardinality" value="countably infinite"/> to <hfp:hasProperty name="bounded" value="true"/> <hfp:hasProperty name="cardinality" value="finite"/> This should automatically repair the descriptions in sections 3.3.11.3, 3.3.12.3, etc., and in appendix F. (2) Reply to Jeremy Caroll, who raised the issue long ago, thanking him for the comment and explaining that in 1.0 we were confused by the fact that each value of these types itself denotes a countably infinite set (e.g. the infinite set of periods of time labeled January in the Gregorian calendar). I am changing the 'version' field from 1.0/1.1 both to 1.1 only, and spawning a twin issue for the 1.0 corrigendum: bug 2668 is the 1.0 twin.
Correction: gYearMonth is not finite, for the same reason that gYear is not finite. So it should NOT be changed. It's only the three types which don't have year information in them: gMonth, gDay, and gMonthDay. Sorry for the slip.
On its telcon of 13 January, the WG adopted the proposal in comment 3, as amended in comment 4 and with the instruction to the editor to get it right this time. The change was integrated into the status-quo document on 13 January 2006,