This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
A number of the extvardeclwithouttype tests fail with processors implementing static typing. Statically the external variables are typed as item()*, uisng these external variables in for example operators result in a XPTY0004 error. I see the following options: - make XPTY0004 an expected error (in addition to the current expected result) - add a constructor function when using the variable as operand. Here is the complete list: extvardeclwithouttype-10.xq extvardeclwithouttype-11.xq extvardeclwithouttype-12.xq extvardeclwithouttype-13.xq extvardeclwithouttype-14.xq extvardeclwithouttype-15.xq extvardeclwithouttype-16.xq extvardeclwithouttype-17.xq extvardeclwithouttype-20.xq extvardeclwithouttype-21.xq extvardeclwithouttype-22.xq Thanks, Marc
Hey Marc: Thanks for the comment. The first query on these tests follow the pattern let $var := 1 return $var) I changed part of the query to let $var := 1 return xs:sometype($var) The second query follows: declare variable $x as external; $x + $x I changed part of that query to: declare variable $x as xs:sometype external; Please close the bug if in agreement and when able to veryfy with next realease. Thanks, Carmelo
verified and looks ok
Hi, I think this one needs to be reopened. There seem to be more queries suffering. Note that none of these queries were reported before. Moving forward with our testing we found these additional ones. extvardeclwithouttype-2.xq extvardeclwithouttype-3.xq extvardeclwithouttype-4.xq extvardeclwithouttype-5.xq extvardeclwithouttype-6.xq extvardeclwithouttype-7.xq extvardeclwithouttype-8.xq extvardeclwithouttype-9.xq Thanks, Marc
Marc: Similar changed done although, Looking back at it, perhaps the thing to dod is just allow fo rthe extra outcome (or an error). Thanks, Carmelo
yes, adding an error code rather than the current apporach is fine with us. Actually, that's what we proposed in the initial report. Also, these tests are named "extvardeclwithouttype", as such adding a type declaration might well violate the intend of the test. As such I will not set the bug to fixed, I assume the Testing task Force will decide. Thanks, Marc
Marc: I think adding the extra eoor code as an outcome is more proper as that will be consistent with the test intentions. As such, I modified the tests and added the extra outcome to the catalog file. Thanks, Carmelo
Issue verified and indeed fixed! Closing. Thanks, Tom