[
http://issues.apache.org/jira/browse/WODEN-97?page=comments#action_12454508 ]
John Kaputin commented on WODEN-97:
-----------------------------------
Keith,
all the bindings in MessageTest-2G refer to interfaces. I assume you mean
MessageTest-3G, whose binding named SafeEchoHTTPBinding is interface-less.
Firstly, I want to make sure I understand your requirement correctly. I think
the problem you've hit is that you currently have to use the Woden API
differently to determine binding information for faults (or operations)
depending on whether or not the interface has an explicit binding, but what you
would like is to use the same Woden programming model regardless of whether the
interface binding is explicit or derived. In other words, if it's derived you
would like Woden to make that transparent and give you a common programming
model for the binding information needed to build your concrete messages.
Currently, if the interface is explicitly bound you can just get the Binding
component from the Endpoint then navigate down the component API to the Binding
sub-components to retrieve their binding-specific extension properties and
their associated Interface sub-components. However, for a Service whose
Endpoint refers to a generic (interface-less) binding, you need to get the
Interface from the Service to get its faults or operations then get the Binding
from the Endpoint to obtain its binding extensions properties and then match
the Interface components and binding extension properties yourself to derive
the appropriate fault and operation binding information.
Here a couple of reasons why I think we should not change the existing
behaviour of the Binding.getBindingFaults() method.
A goal of the Woden project is to define an API that faithfully reflects the
WSDL Component model as defined by the spec, so the existing getter methods in
the component API map directly to the component properties defined by curly
braces {} in the spec and behave as described in the spec. For example,
Binding.getBindingFaults() represents the {binding faults} property of the
Binding component which according to the spec only contains BindingFaults for
<fault> child elements of the <binding> element.
Also, note what the spec says in Part 1, section 2.7.1:
"Conversely, a Binding component which omits any operation-specific binding
details and any fault binding details MAY omit specifying an interface. Binding
components that do not specify an interface MAY be used to specify
operation-independent binding details for Service components with different
interfaces. That is, such Binding components are reusable across one or more
interfaces."
So even if we did change Binding.getBindingFaults() to accomodate derived fault
binding information, the potential one to many mapping from Binding to
Interface via Endpoint and Service could make the getBindingFaults() method
impractical. Remember that a single Binding instance exists for a <binding>
element, even if it is referenced by multiple endpoints.
I guess the existing Woden programming model conforms to the intention of the
WSDL 2.0 spec. That is, it's up to the concrete message builder to access the
required information from the appropriate part of the WSDL component model, so
this information is supposed to be extracted differently from the WSDL
component model for explicit interface binding compared to generic binding.
If the Woden API was to make this transparent then I don't think accomodating
derived binding information within the Binding API is the correct approach. If
all you want is easy access to the Interface components then Arthur's
suggestion of convenience methods on Endpoint would suffice. If actually you
want some transparent mapping of Interface components to binding extension
properties, then possibly some derived binding helper class retrieved via
Endpoint would be a suitable approach, but that might still not give you a
common programming model (e.g. if it is not appropriate for this derived
binding to be treated as a subtype of Binding). I need to give this more
thought. Any comments are welcome.
> When a interface is not referenced from a binding component the
> getInterfaceFault() method returns null
> -------------------------------------------------------------------------------------------------------
>
> Key: WODEN-97
> URL: http://issues.apache.org/jira/browse/WODEN-97
> Project: Woden
> Issue Type: Bug
> Reporter: Keith Godwin Chapman
>
> Hi,
> When I use the WSDL of Message-Test-2G and try to get the interfaceFault for
> a fault thats in a binding I get a null object. The reason is that the
> binding does not specify a interface. Should woden resolve the fact that the
> interface referenced by this binding is specified in the service component
> and return the correct InterfaceFault.
> In my opinion it would be better is woden returns the correct Interface Fault
> by resolving the reference from the service component.
> Thanks,
> Keith.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://issues.apache.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
