cxf-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Eric Johnson <>
Subject Re: Question for CXF developers - level of support for W3C SOAP/JMS specification
Date Mon, 18 Apr 2011 22:22:57 GMT
Hi Daniel,

Thanks for the very quick response. Sorry, I've been a little swamped. 
My best attempt at answers follows.

On 4/5/11 7:45 AM, Daniel Kulp wrote:
> On Monday 04 April 2011 5:32:49 PM Eric Johnson wrote:
>> I've a question for the CXF developers. To quickly introduce myself, I'm
>> the chair of the W3C WG for SOAP/JMS [1].
>> We're at the point where we want to declare a "Proposed Recommendation"
>> (PR) for what is currently our "working draft" [2]. For those
>> unfamiliar, that's the step before declaring something an actual W3C
>> official "Recommendation." [4]
>> To achieve this milestone, we simply need to have two implementations of
>> the specification. More specifically, we need two implementations of
>> each and every normative statement of the specification - even those
>> normative statements that are optional.
> That's interesting.     If you don't mind my asking, what's the other
> implementation?   Not sure if you can say that or not.   We've obviously done
> quite a bit of testing CXF<--->CXF, but not really any interop testing.
> Thus, it would be good to really try some interop testing with another
> implementation.

I believe that the IBM WebSphere Application Server, version 8, has this 
capability. Currently in beta I think, but I think if you crawl around 
their site you can find the download.

>>    From work I've done looking at the source code and the samples, it
>> appears that CXF falls into that category. Of particular concern, we're
>> curious about the WSDL extension elements that we've defined, in part
>> because some of the vendors we've talked to have indicated that they
>> will not be supporting such extension elements.
> I know we have some tests that stick the extensions all over the place in the
> wsdl.   While I personsonally don't aggree with some of the places they are
> allowed, I know the spec does allow for them so the tests are valid.

>> In any case, I'm looking for some sort of public statement from the CXF
>> developers about each the normative statements ("assertions") [3] from
>> the spec, and whether each is covered by the CXF implementation. Since
>> the specification has changed slightly since the last working draft -
>> mostly to clarify the assertions, and fix some oversights - it would
>> actually be useful to know about CXF with respect to our latest working
>> copy, and its assertions [5], and that either or both will do.
>> Can anyone comment?
> Honestly, Peter Easton might know a bit more since the last patch in this area
> came from him.   See:

I've CC'd him on this email.
> Looking at the test suite itself:

Oh, excellent - I hadn't actually noticed where those were in the source 
tree. Looks like they moved slightly:

> I do see some tests that are missing from the official suite:
> test0007
> test0015
> test0016
> test0017
> test0018
> test0019
> test0020
> test1005
> What I DON'T know is if the features tested by those tests are not implemented
> in CXF or just not tested or possibly tested as part of one of the other
> tests.    If anyone would like to fill those in, I'd be happy to review and
> apply any patches.

That, in and of itself is useful - some of those tests don't matter, 
because they don't address portions of the specification that are 
normative (we "demoted" the WSDL 2.0 support, because we didn't 
anticipate two interoperable versions.)

I did a cross-reference on all those tests, and it looks like the one 
we're missing is Protocol-2070



> Dan
>> Thanks!
>> -Eric Johnson
>> [1]
>> [2]
>> [3]
>> [4]
>> [5]
>> pe=text/html;charset=utf-8#assertionsummary

View raw message