cxf-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Willem Jiang <willem.ji...@gmail.com>
Subject Re: CXF continuation enhancement
Date Wed, 08 Sep 2010 14:02:50 GMT
Hi Dan,

Daniel Kulp wrote:
> On Thursday 02 September 2010 9:32:46 am Willem Jiang wrote:
>> Hi
>>
>> I just did some work to let the camel-cxf leverage the CXF continuation
>> framework which you did. I just found current CXF continuation which
>> suspend is implemented by throw the runtime exception. This
>> implementation has a shortcoming which cannot call the other framework's
>> async API after continuation suspend is called as Jetty7 does.
>>
>> This will introduce a situation that continuation resume will be called
>> before the continuation suspend is called if the famework's async API is
>> finally using the thread to do the job.
>>
>> So I'm thinking of some enhancement could be possible to let CXF
>> continuation leverage Jetty 7 new continuation API.
>>
>> Any idea?

I created a JIRA[1] and submit a patch for it. Please review it :)

> 
> While upgrading things to Jetty 7, this is something I was thinking about a 
> bit more as well.    I actually wanted to add a method like:
> 
> public void resume(Object r);
> 
> method onto the continuation.   The endpoint (or background thread or 
> whatever) could call that directly (the Object being the thing that would 
> normally be returned from the method).  The runtime could immediately just 
> grab the associated chain and pick up where it left off and not do the restart 
> thing.   The new Servlet 3 API's and the new things in Jetty 7 seem to allow 
> that.   It could potentially make the continuations stuff a lot easier to work 
> with.
Current we just make the interceptor chain resume from where we pause 
it. I don't know that you want.
I think the blow code is enough.
  continuation.setObject(obj);
  continuation.resume()

> 
> I actually would like to change everything to not bother with the exception 
> either.   The endpoint would call "suspend" and if it returns  true (meaning 
> the request was suspended), just return null.  If false, it knows it needs to 
> do the work synchronously.   It's much closer to how the Servlet 3 things 
> work.   Unfortunately, that would completely change the semantics of the API 
> and would require some good docs for the migration guide.
I don't think this is a good idea, if we can't support to the 
continuation API from the transport level, we simple don't let the user 
can get the ContinuationProvider from the message context.

It could be much easier for user to use :)

[1]https://issues.apache.org/jira/browse/CXF-2982

Willem

Mime
View raw message