Proposal for action 108

Action #108 is for me to come up with a more concrete proposal for WSDL
testing.  We're down to specifically two WSDL assertions, #3001, and #3002.

Here's what I'm noticing:
Test cases #6 & #7:
http://dev.w3.org/2008/ws/soapjms/testcases/testcases/testcases.html#test0006
http://dev.w3.org/2008/ws/soapjms/testcases/testcases/testcases.html#test0007

already have WSDL 1.1/2 documents in them indicating data.  We could
copy those test cases, and add some twists.  For example, a new test
based on #6 could add a URI in the wsdl11soap11:address @location
attribute, where the values in the URI are all different from the ones
expressed in the binding.  We can then assert that the values are the
ones from the URI.

To provide adequate testing for #3001, it seems like the possible
combinations include these:
URI trumps port (& endpoint in WSDL 2.0)
URI trumps service
URI trumps binding

port settings trump service
port settings trump binding

service settings trump binding

This is six additional tests for each of WSDL 1.1 & WSDL 2.0.  There are
other possibilities, but they seem like they're in the noise (for
example, URI, port, service, and binding all specified, URI still needs
to trump).

If we add one more set of tests - settings on the binding, then we've
also got coverage for #3002, as all of the above tests confirm that
settings are supported in the port/endpoint and service elements.

That means, with 16 new tests, we've got coverage for WSDL 1.1 & 2.0.

Sound right to everyone?

-Eric.

Received on Monday, 11 January 2010 23:16:08 UTC