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 regular expression syntax test cases in matches.re.xml test if \p{block} matches every character in that unicode code block, even characters that are non-assigned, e.g. ԰ for Armenian or ޿ for Thaana. It does not make much sense to test for characters that do not actually exist.
The fact that a test "does not make much sense" seems irrelevant to me. The question is, what does the spec say that an implementation should do with a given query? As far as I can see, the codepoints 1328 and 1983 are legal codepoints in XML, and they are defined in the Unicode database (specifically, the blocks.txt file) to be within the range of a particular block, and should therefore match \p{IsBlock} for that block.
The WG decision that the test is valid: The key rationale is to decouple our spec from Unicode: using an unassigned character may be (perhaps is) a violation of the agreement between data source and data sink, if they have agreed to use a particular version of Unicode. But it is not (and should not be) a violation of our specs. Unicode may assign the character next month or next year, and it should be possible to use the character then, in a conforming implementation. That means it must be possible *now* to use it in a conforming implementation.