Well, I guess I cannot really help you with that one, but thanks for
coming back to us and letting us know about your progress.

Regards
Werner

Stevo Slavić wrote:
> Hello Werner,
> 
> Yes, I have, thank you. Tho, another one popped in. To use
> CommonsHttpMessageSender, Spring-WS requires commons-httpclient 3.x, but
> as Hippo CMS Repository Java Adapter, used to access Hippo CMS
> Repository, already has dependency to commons-httpclient 2.x (and is not
> compatible with 3.x), and as this CMS is currently irreplaceable part of
> the system being built, I had to revert back to using
> HttpUrlConnectionMessageSender.
> 
> Anyway, using Axiom instead of Saaj was key to gaining a lot better
> performance for handling big messages. Castor does it's job well in
> either case.
> 
> Regards,
> Stevo.
> 
> Werner Guttmann wrote:
>> In other words, you have resolved your issue ?
>>
>> Werner
>>
>> Stevo Slavić wrote:
>>  
>>> Hello Werner,
>>>
>>> I've tuned Spring-WS, by replacing message factory and message sender
>>> implementation, from SaajSoapMessageFactory to AxiomSoapMessageFactory,
>>> and from HttpUrlConnectionMessageSender to CommonsHttpMessageSender, and
>>> experienced a dramatic performance increase with big message payloads,
>>> from ~10min down to 10-15sec.
>>>
>>> Regards,
>>> Stevo.
>>>
>>> Werner Guttmann wrote:
>>>    
>>>> So the difference only occurs when spring-ws is added to the technology
>>>> stack, correct. If that's the case, feel free to supply us with the
>>>> complete code (incl. instructions how to replay the problem=, and we'll
>>>> be having a look. Please try to be as minimal as possible, though.
>>>>
>>>> Regards
>>>> Werner
>>>>
>>>> Stevo Slavić wrote:
>>>>  
>>>>      
>>>>> Hello Werner,
>>>>>
>>>>> Thank you for reply. I've created a test, generating 4000 instances
>>>>> and
>>>>> tried to marshall and unmarshall them, and it was all done under a
>>>>> second! This brings me to another problem, finding what else is
>>>>> causing
>>>>> such a long time to process. I'm using Castor Marshaller to
>>>>> marshall and
>>>>> unmarshall parameters of a web service method, and same
>>>>> marshaller/unmarshaller is configured to be used by Spring WS when
>>>>> making web service calls.
>>>>>
>>>>> Regards,
>>>>> Stevo.
>>>>>
>>>>> Werner Guttmann wrote:
>>>>>           
>>>>>> Without a bit of code to look at, providing you with hints/answers
>>>>>> will
>>>>>> be difficult. Would you be willing to provide us with a test case ?
>>>>>>
>>>>>> Regards
>>>>>> Werner
>>>>>>
>>>>>> Stevo Slavić wrote:
>>>>>>  
>>>>>>               
>>>>>>> Hello all,
>>>>>>>
>>>>>>> It took Castor 10 minutes and 22 seconds to unmarshall 3835
>>>>>>> occurrences
>>>>>>> of element named "person" in an XML document with structure as in
>>>>>>> following example:
>>>>>>>
>>>>>>> <persons>
>>>>>>>  <person>
>>>>>>>    <user>
>>>>>>>      <username>[EMAIL PROTECTED]</username>
>>>>>>>      <password>aPassword</password>
>>>>>>>      <enabled>true</enabled>
>>>>>>>      <activated>true</activated>
>>>>>>>      <activationCode>anActivationCode</activationCode>
>>>>>>>    </user>
>>>>>>>  </person>
>>>>>>>  <person>
>>>>>>>    ...
>>>>>>>  </person>
>>>>>>>  ...
>>>>>>> </persons>
>>>>>>>
>>>>>>>
>>>>>>> A "person" can also have a "profile" element which is more complex
>>>>>>> than
>>>>>>> "user" element, while both "user" and "profile" are optional. XML
>>>>>>> document that was unmarshalled for every "person" had only "user" as
>>>>>>> subelements. It looks simple to me and I expected it to be
>>>>>>> unmarshalled
>>>>>>> in far less time. Such my expectations were based on the fact
>>>>>>> that it
>>>>>>> took 10 seconds on side that marshalled object tree into XML, but
>>>>>>> using
>>>>>>> Axis 1.
>>>>>>>
>>>>>>> Castor (un)marshaller was not used directly but through Spring-OXM
>>>>>>> module, all within a Spring framework based web application,
>>>>>>> deployed on
>>>>>>> a laptop with Intel T7500 2.2GHz CPU and 4GB of memory, Vista x64
>>>>>>> with
>>>>>>> Java 6 VM and Tomcat 6 tweaked to take 1280MB of memory and use
>>>>>>> parallel
>>>>>>> garbage collector.
>>>>>>>
>>>>>>> I'm looking for answers why it took Castor so long to complete, and
>>>>>>> how
>>>>>>> should I tweak it to perform a lot better? Any suggestions are
>>>>>>> appreciated. Thanks in advance!
>>>>>>>
>>>>>>>
>>>>>>> Regards,
>>>>>>> Stevo.
>>>>>>>
>>>>>>> ---------------------------------------------------------------------
>>>>>>>
>>>>>>> To unsubscribe from this list, please visit:
>>>>>>>
>>>>>>>    http://xircles.codehaus.org/manage_email
>>>>>>>
>>>>>>>
>>>>>>>                         
>>>>>> ---------------------------------------------------------------------
>>>>>> To unsubscribe from this list, please visit:
>>>>>>
>>>>>>     http://xircles.codehaus.org/manage_email
>>>>>>
>>>>>>
>>>>>>
>>>>>>                   
>>>> ---------------------------------------------------------------------
>>>> To unsubscribe from this list, please visit:
>>>>
>>>>     http://xircles.codehaus.org/manage_email
>>>>
>>>>
>>>>
>>>>         
>>
>> ---------------------------------------------------------------------
>> To unsubscribe from this list, please visit:
>>
>>     http://xircles.codehaus.org/manage_email
>>
>>
>>
>>   
> 

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email


Reply via email to