8955RE: [soapbuilders] Additional SOAP 1.2 tests redux
- Jan 2, 2003Bob,
I don't think there is any reason why xmlp-7, 8, and 10 should use RPC
convention instead of doc/lit, especially as the former seems to be
covered by 60.1 and 60.2 although I suspect that these may cause
problems as well? If so, could these test cases be handled by having two
WSDL documents that forced the fault to occur?
Thanks for the feedback,
Henrik Frystyk Nielsen
* * * * *
xmlp-7 (echoSenderFault), xmlp-8 (echoReceiverFault):
SOAP messages representing RPCs result in the invocation of a native
procedure in its normal runtime environment, outside the world of the
SOAP processor. It is impossible for such a procedure to cause the
generation of a SOAP fault other than indirectly. So at least in this
implementation, these tests can't be supported. It's a different
situation for doc/literal operations, it would be no problem, as an XML
processing app has access to the SOAP environment at runtime and can
generate any specific SOAP fault it wishes to. I hope this makes sense.
I wonder if any other implementations would have similar problems.
This is similar to the situation above. This is an RPC operation, and
the native procedure runs outside the SOAP environment, yet the test
requires that the procedure have access to infoset stuff (types) from
the original XML representation of the RPC. Of course the procedure is
called with the parameter data, but it has already been deserialized to
native types and XS type info is out of reach. Again, if it was a
doc/literal operation, no problem.
- << Previous post in topic Next post in topic >>