Hi Willem, OK, I understand it that you also do not like the side effects :-). I created CAMEL-8410 on top of CAMEL-8402 that removes the side effects and implements a StreamCache for CxfPayload objects. As a consequence streams are used directly, but if you have stream caching you are supposed to be able to read from the CachedCxfPayload as often as you want...
What do you think about this? Best regards Stephan -----Original Message----- From: Willem Jiang [mailto:willem.ji...@gmail.com] Sent: Donnerstag, 26. Februar 2015 09:46 To: dev@camel.apache.org Subject: Re: StreamCache for CxfPayload Using the Source object in CxfPayload is get a better performance, but it could cause some trouble if you want to reread the Payload message again. I agree with you we need avoid the satiation that converted object have some side effects on CxfPayload body. I will apply the patch shortly. -- Willem Jiang Red Hat, Inc. Web: http://www.redhat.com Blog: http://willemjiang.blogspot.com (English) http://jnn.iteye.com (Chinese) Twitter: willemjiang Weibo: 姜宁willem On February 25, 2015 at 5:12:19 PM, Siano, Stephan (stephan.si...@sap.com) wrote: > Hi, > > The CxfPayload data type (from camel-cxf) contains some stream elements but > does not > support stream caching. > > I am working on a contribution to provide stream caching for that. During > that if found > some strange things in the CxfPayloadConverter: > If the CxfPayloadConverter converts a CxfPayload into something else, it > might change > the original CxfPayload as a side effect. This is likely done in order to > make the CxfPayload > re-readable (because otherwise the Source containing the message body may get > lost > after processing). The code also contains a bug: if the body source is e.g. > converted > into a SAXSource the body source will be set to that SAXSource even if was > set to a (re-readable) > DOMSource before. I created CAMEL-8402 for that (the patch just removes two > lines of > code). > > In my optinion this side-effect of the type converter should not be there at > all. Do you > think it is ok if I remove the mechanism altogether when introducing a > StreamCache for > CxfPayload? The StreamCache will make sure that the CxfPayload is re-readable > without > these hacks. > > Best regards > Stephan >
