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 XQTS lacks coverage in several known areas(and probably unknown too), and here is a known one which I haven't written tests for locally: As of this writing there have been a lot of bugs reported on tests that aren't static typing safe. These tests haven't been reported previously despite static typing implementors have been reported other bugs, suggesting that their implementations aren't conformant is this area. I believe an improvement for this lack of coverage is to find all reports kindly reported by tim@cbcl.co.uk and write static typing tests that are similar, but intentionally cases type errors. So, this report can be addressed whenever a pickup-heat is done for XQTS, and new tests are added. That implementors seem to implement static typing so widely, really makes me wonder how useful spec'ing static typing is. (Another area where XQTS could use stronger coverage is regular expressions.) Frans
Such XQTS tests need to take into account possible Static Typing Extensions (http://www.w3.org/TR/xquery/#id-static-extensions). Consider for example http://www.w3.org/Bugs/Public/show_bug.cgi?id=3686. Implementing the Formal Semantics by the book makes the following expression fail: fn:matches(xs:string("This is a characte"), xs:string("This is a characte")) However, an implementation may infer statically that both arguments are guaranteed to be of type xs:string. As such a static typing implementation is not required to report a failure here.
Yes, forgot about that. It really sucks(pardon my french) that static typing errors aren't deterministic. No interoperability here. This could have been avoided by disallowing Static Typing Extensions that accept expressions which without the extension would yield a type error. Frans
I'm going to mark this as an enhancement and we can consider it after we publish XQTS 1.0.1.
I'm withdrawing this. We lack coverage in many areas, and at least I don't have the measures for fixing them. Maybe we'll stumble over more tests or methods to produce them.