javax.ws.xml.Holder not serializable

[Kurt T Stam-4]

Hi guys,

On the jUDDI project one of our APIs uses the javax.ws.xml.Holder class
(which contains jUDDI specific data classes). We have a requirement to
send the entire Holder class over RMI, but Holder is not Serializable. I
think this is a short coming of the implementation.

http://svn.apache.org/repos/asf/geronimo/specs/trunk/geronimo-jaxws_2.2_spec/src/main/java/javax/xml/ws/Holder.java

If you agree can this be fixed? Or do you believe that the spec dictates
this?

Thx,

--Kurt

kstam at apache.org

[Jarek Gawor-2]

Kurt,

I'm pretty sure that's what the spec mandates. I can test this out to
see if it breaks the TCK signature tests but even if it doesn't you
might still run into that problem on Java 6 or later since the
javax.xml.ws API are now provided by the JVM. However, I think you
could override these JVM API by placing the Geronimo spec jar into the
JVM's lib/endorsed directory.

I'll test this out and let you know.

Jarek

On Mon, Jun 29, 2009 at 7:56 PM, Kurt T Stam<kurt.stam@...> wrote:

> Hi guys,
>
> On the jUDDI project one of our APIs uses the javax.ws.xml.Holder class
> (which contains jUDDI specific data classes). We have a requirement to send
> the entire Holder class over RMI, but Holder is not Serializable. I think
> this is a short coming of the implementation.
>
> http://svn.apache.org/repos/asf/geronimo/specs/trunk/geronimo-jaxws_2.2_spec/src/main/java/javax/xml/ws/Holder.java
>
> If you agree can this be fixed? Or do you believe that the spec dictates
> this?
>
> Thx,
>
> --Kurt
>
> kstam at apache.org
>
>
>

[Kurt T Stam-4]

Thanks for going to test this Jarek.

It sounds like I would need to being this up with JCP-224 then.

--Kurt

Jarek Gawor wrote:

> Kurt,
>
> I'm pretty sure that's what the spec mandates. I can test this out to
> see if it breaks the TCK signature tests but even if it doesn't you
> might still run into that problem on Java 6 or later since the
> javax.xml.ws API are now provided by the JVM. However, I think you
> could override these JVM API by placing the Geronimo spec jar into the
> JVM's lib/endorsed directory.
>
> I'll test this out and let you know.
>
> Jarek
>
> On Mon, Jun 29, 2009 at 7:56 PM, Kurt T Stam<kurt.stam@...> wrote:
>  
>> Hi guys,
>>
>> On the jUDDI project one of our APIs uses the javax.ws.xml.Holder class
>> (which contains jUDDI specific data classes). We have a requirement to send
>> the entire Holder class over RMI, but Holder is not Serializable. I think
>> this is a short coming of the implementation.
>>
>> http://svn.apache.org/repos/asf/geronimo/specs/trunk/geronimo-jaxws_2.2_spec/src/main/java/javax/xml/ws/Holder.java
>>
>> If you agree can this be fixed? Or do you believe that the spec dictates
>> this?
>>
>> Thx,
>>
>> --Kurt
>>
>> kstam at apache.org
>>
>>
>>
>>    
>
>  

[Jarek Gawor-2]

Kurt,

The signature tests fail with this change.... so we can't add
Serializable to the Holder class.

Jarek

On Tue, Jun 30, 2009 at 12:00 AM, Jarek Gawor<jgawor@...> wrote:

> Kurt,
>
> I'm pretty sure that's what the spec mandates. I can test this out to
> see if it breaks the TCK signature tests but even if it doesn't you
> might still run into that problem on Java 6 or later since the
> javax.xml.ws API are now provided by the JVM. However, I think you
> could override these JVM API by placing the Geronimo spec jar into the
> JVM's lib/endorsed directory.
>
> I'll test this out and let you know.
>
> Jarek
>
> On Mon, Jun 29, 2009 at 7:56 PM, Kurt T Stam<kurt.stam@...> wrote:
>> Hi guys,
>>
>> On the jUDDI project one of our APIs uses the javax.ws.xml.Holder class
>> (which contains jUDDI specific data classes). We have a requirement to send
>> the entire Holder class over RMI, but Holder is not Serializable. I think
>> this is a short coming of the implementation.
>>
>> http://svn.apache.org/repos/asf/geronimo/specs/trunk/geronimo-jaxws_2.2_spec/src/main/java/javax/xml/ws/Holder.java
>>
>> If you agree can this be fixed? Or do you believe that the spec dictates
>> this?
>>
>> Thx,
>>
>> --Kurt
>>
>> kstam at apache.org
>>
>>
>>
>

[Kurt T Stam-4]

Thanks Jarek,

I'll take it up the JCP.

--Kurt

Jarek Gawor wrote:

> Kurt,
>
> The signature tests fail with this change.... so we can't add
> Serializable to the Holder class.
>
> Jarek
>
> On Tue, Jun 30, 2009 at 12:00 AM, Jarek Gawor<jgawor@...> wrote:
>  
>> Kurt,
>>
>> I'm pretty sure that's what the spec mandates. I can test this out to
>> see if it breaks the TCK signature tests but even if it doesn't you
>> might still run into that problem on Java 6 or later since the
>> javax.xml.ws API are now provided by the JVM. However, I think you
>> could override these JVM API by placing the Geronimo spec jar into the
>> JVM's lib/endorsed directory.
>>
>> I'll test this out and let you know.
>>
>> Jarek
>>
>> On Mon, Jun 29, 2009 at 7:56 PM, Kurt T Stam<kurt.stam@...> wrote:
>>    
>>> Hi guys,
>>>
>>> On the jUDDI project one of our APIs uses the javax.ws.xml.Holder class
>>> (which contains jUDDI specific data classes). We have a requirement to send
>>> the entire Holder class over RMI, but Holder is not Serializable. I think
>>> this is a short coming of the implementation.
>>>
>>> http://svn.apache.org/repos/asf/geronimo/specs/trunk/geronimo-jaxws_2.2_spec/src/main/java/javax/xml/ws/Holder.java
>>>
>>> If you agree can this be fixed? Or do you believe that the spec dictates
>>> this?
>>>
>>> Thx,
>>>
>>> --Kurt
>>>
>>> kstam at apache.org
>>>
>>>
>>>
>>>      
>
>  

[dkulp]

On Tue June 30 2009 11:38:35 am Kurt T Stam wrote:
> Thanks for going to test this Jarek.
>
> It sounds like I would need to being this up with JCP-224 then.

I'm pretty sure you won't like the answer.   :-)

Basically, the jcp224/jaxws folks (and the jaxb folks), at one point long ago,
specifically decided against making things "Serializable" since the
Serializable symantics are not honored by the XML serialization.   (example:
field marks transient would still be written to XML)

Anyway, for the most part, it was a conscious decision.   While that decision
may or may not apply to the Holder objects, it wouldn't surprise me at all if
they say something along those lines.

In anycase, it's definitely not something that would change for JAX-WS
2.0/2.1/2.2 so you would be looking at something at LEAST a year away,
probably more, which probably doesn't help you now.

Dan

> --Kurt
>
> Jarek Gawor wrote:
> > Kurt,
> >
> > I'm pretty sure that's what the spec mandates. I can test this out to
> > see if it breaks the TCK signature tests but even if it doesn't you
> > might still run into that problem on Java 6 or later since the
> > javax.xml.ws API are now provided by the JVM. However, I think you
> > could override these JVM API by placing the Geronimo spec jar into the
> > JVM's lib/endorsed directory.
> >
> > I'll test this out and let you know.
> >
> > Jarek
> >
> > On Mon, Jun 29, 2009 at 7:56 PM, Kurt T Stam<kurt.stam@...> wrote:
> >> Hi guys,
> >>
> >> On the jUDDI project one of our APIs uses the javax.ws.xml.Holder class
> >> (which contains jUDDI specific data classes). We have a requirement to
> >> send the entire Holder class over RMI, but Holder is not Serializable. I
> >> think this is a short coming of the implementation.
> >>
> >> http://svn.apache.org/repos/asf/geronimo/specs/trunk/geronimo-jaxws_2.2_
> >>spec/src/main/java/javax/xml/ws/Holder.java
> >>
> >> If you agree can this be fixed? Or do you believe that the spec dictates
> >> this?
> >>
> >> Thx,
> >>
> >> --Kurt
> >>
> >> kstam at apache.org

--
Daniel Kulp
dkulp@...
http://www.dankulp.com/blog

[Kurt T Stam-4]

Thx for the backgroud info Dan :(

Isn't it great, the also decided to make the class 'final'? So I can't
even make a wrapper.

I've send a request to jsr-224-public@.... Hoping for the best..

--Kurt


Daniel Kulp wrote:

> On Tue June 30 2009 11:38:35 am Kurt T Stam wrote:
>  
>> Thanks for going to test this Jarek.
>>
>> It sounds like I would need to being this up with JCP-224 then.
>>    
>
> I'm pretty sure you won't like the answer.   :-)
>
> Basically, the jcp224/jaxws folks (and the jaxb folks), at one point long ago,
> specifically decided against making things "Serializable" since the
> Serializable symantics are not honored by the XML serialization.   (example:
> field marks transient would still be written to XML)
>
> Anyway, for the most part, it was a conscious decision.   While that decision
> may or may not apply to the Holder objects, it wouldn't surprise me at all if
> they say something along those lines.
>
> In anycase, it's definitely not something that would change for JAX-WS
> 2.0/2.1/2.2 so you would be looking at something at LEAST a year away,
> probably more, which probably doesn't help you now.
>
> Dan
>
>
>  
>> --Kurt
>>
>> Jarek Gawor wrote:
>>    
>>> Kurt,
>>>
>>> I'm pretty sure that's what the spec mandates. I can test this out to
>>> see if it breaks the TCK signature tests but even if it doesn't you
>>> might still run into that problem on Java 6 or later since the
>>> javax.xml.ws API are now provided by the JVM. However, I think you
>>> could override these JVM API by placing the Geronimo spec jar into the
>>> JVM's lib/endorsed directory.
>>>
>>> I'll test this out and let you know.
>>>
>>> Jarek
>>>
>>> On Mon, Jun 29, 2009 at 7:56 PM, Kurt T Stam<kurt.stam@...> wrote:
>>>      
>>>> Hi guys,
>>>>
>>>> On the jUDDI project one of our APIs uses the javax.ws.xml.Holder class
>>>> (which contains jUDDI specific data classes). We have a requirement to
>>>> send the entire Holder class over RMI, but Holder is not Serializable. I
>>>> think this is a short coming of the implementation.
>>>>
>>>> http://svn.apache.org/repos/asf/geronimo/specs/trunk/geronimo-jaxws_2.2_
>>>> spec/src/main/java/javax/xml/ws/Holder.java
>>>>
>>>> If you agree can this be fixed? Or do you believe that the spec dictates
>>>> this?
>>>>
>>>> Thx,
>>>>
>>>> --Kurt
>>>>
>>>> kstam at apache.org
>>>>        
>
>