#63 AssociationInvalidQuery should return fault

[Brett Allen]

Tested with March 10 daily build of the RI and OTM revision 2635.

Line #150: Association Invalid Query Fail
Expected - http://schemas.xmlsoap.org/ws/2004/09/enumeration/fault
Received - http://schemas.xmlsoap.org/ws/2004/09/enumeration/EnumerateResponse

[Thomas Weber]

Logged In: YES
user_id=1857294
Originator: NO

RI cannot check if Role is valid without association class. Perhaps we can change the OTM test in this sense.

[Thomas Weber]

Logged In: YES
user_id=1857294
Originator: NO

Same is true for WS-Protocol/AssociatedInvalidQuery test.

[Brett Allen]

Logged In: YES
user_id=2012760
Originator: YES

My understanding of this test is that the filter in the AssociationInvalidQuery test input should return instances of association classes that have "NotARole" as one of their key properties. Since no association class will have this as a key property there won't be any matches and DSP0227 R7.2-4 then recommends a CannotProcessFilter fault in this situation (the rule is focused on whether the Role matches any instances, rather than if it is or could be a valid Role for some class).

If I'm misinterpreting this please let me know where I've gone wrong. Thanks

[Thomas Weber]

Logged In: YES
user_id=1857294
Originator: NO

Sorry about the short note to this issue, I'll try to clarify:
First of all, both AssociatedInvalidQuery and AssociationInvalidQuery tests are valid to our oppinion. The problem clearly lies on RI side, where implementation is not possible without big efford. There is no way to check if the "NotARole" property is a key property of any unspecified association class. We think this is a conceptionally questionable requirement.
So what we propose is to change the OTM request in a way we can process in RI. In detail we would require an AssociationClassName to be specified for checking the Role name. This will not work without modifications to RI but is achievable with sensible efford.
Keeping this particular failure a known issue of RI could be an option as well.

[Brett Allen]

Logged In: YES
user_id=2012760
Originator: YES

Updated AssociatedInvalidQuery and AssociationInvalidQuery in OTM revision 2708. AssociatedInvalidQuery now passes. AssociationInvalidQuery is still having a problem but perhaps this is related to the known issue with AssociationRole.

[Navin Boppuri]

Logged In: YES
user_id=2002123
Originator: NO

Issue still exists in v1.0.6

[Navin Boppuri]

Logged In: YES
user_id=2002123
Originator: NO

Issue still exists in v1.0.6

[Navin Boppuri]

Logged In: YES
user_id=2002123
Originator: NO

Akash: Will investigate and update he status by EOD today

[Navin Boppuri]

Logged In: YES
user_id=2002123
Originator: NO

The latest OTM adds a Result Classname to the filter so the RI can check if the role is valid. RI still fails and does not return a fault. This may be need to fixed in the RI.