patch submitted - please review at your convenience

On Tue, Sep 4, 2012 at 5:24 PM, Jason Pell <[email protected]> wrote:
> I have created a jira issue, I will submit a patch asap and post to
> this group again at that time.
>
> https://issues.apache.org/jira/browse/CXF-4497
>
>
>
> On Tue, Sep 4, 2012 at 5:08 PM, Jason Pell <[email protected]> wrote:
>> Ok - so that JAXBDataBinding createJAXBContextAndSchemas method is where
>> the additional classes that are not referenced directly are added to the 
>> schema.
>>
>> The line:
>>
>> JAXBContextCache.scanPackages(classes);
>>
>> is where I think my trouble starts.  I don't believe I have an issue
>> with sub classes or
>> xsi:type stuff, so I am wondering if I can add a configuration
>> parameter to not scan
>> the package for additional classes, but rely instead on the
>> walkReferences in JAXBContextInitializer
>>
>>
>> I am working on a small patch to JAXBDataBinding to optionally not
>> call the scanPackages method.  I will raise a
>> jira, hopefully it will be acceptable.
>>
>>
>> On Tue, Sep 4, 2012 at 11:51 AM, Jason Pell <[email protected]> wrote:
>>> Hi,
>>>
>>> I have a jaxb library of classes which are shared between services.  I
>>> would like to exclude any types from the WSDL that are not directly
>>> referenced by an operation in a particular service.   Where might I
>>> start to submit a patch to support this?  I am guessing the easiest
>>> solution would be to somehow update the ServiceWSDLBuilder to
>>> determine whether a type is referenced and rewrite the SchemaInfo
>>> objects?
>>>
>>> I know the obvious answer is to remove the jaxb elements (or at least
>>> the annotations from these objects, and remove from jaxb.index or
>>> ObjectFactory) but we can't do that, we instead want to generate a
>>> WSDL with a subset of the types.
>>>
>>> I will start looking at how we might update the existing WSDL
>>> generation functionality to support this and if its acceptable submit
>>> a patch that can have the functionality enabled via configuration.
>>>
>>> Any pointers or existing code I could use would be much appreciated.
>>>
>>> Thanks
>>> Jason

Reply via email to