This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
In the Microsoft DataTypes tests, test anyURI_a001_1336 appears to assume that 1111111111111111111http://foo/bar is a valid instance of xs:anyURI. I believe that it isn't. The specification states: The ·lexical space· of anyURI is finite-length character sequences which, when the algorithm defined in Section 5.4 of [XML Linking Language] is applied to them, result in strings which are legal URIs according to [RFC 2396], as amended by [RFC 2732]. The XLink escaping algorithm has no effect on this string. The string cannot be interpreted as an absolute URI because the first component of an absolute URI (the scheme) must start with an alpha; and it cannot be interpreted as a relative URI because the part before the first "/" contains a colon. Similarly, other strings that the tests consider to be valid anyURI values, but which I think are invalid, include: test URI(s) anyURI_a003 :a b: anyURI_b004 % anyURI_b006 //
Agreed that the expected outcome of this is invalid should be. We are following up with the WG to determine the process of updating the test suite.
Agreed that these are not legal instances of xs:anyURI under 1.0, though they become legal in 1.1. Noted however that there are significant problems in the 1.0 spec concerning xs:anyURI: * it references RFC 2396 rather than 3986 * it's not clear about the status of relative URI [references], fragment identifiers, etc. (We clearly intend to allow them, but the RFC itself is so vague in its terminology that it's unclear whether the rules as written do allow them.) Perhaps we should mark these particular tests as invalid, but add a note saying the rules are fuzzy. Decision: mark the expected result as "invalid".