> I agree, this would be a good thing to have. I don't know 
> that setHeaders is the right place as this could be used in a 
> REST binding too.
> 
> I would outline two other goals as we figure something out:
> 1. Get rid of Message.getAttachmentMimeType 2. Unify the 
> mechanism with the Attachment class's methods for getting headers
> 
> As SAAJ always scares me a little (for instance MimeHeaders 
> uses a Vector according to the constructor?), I would be more 
> prone to use using a Map and putting it in the Message - 
> message.get(MIME_HEADERS) 

Yep, a named property of type Map<Sting, List<String>> makes sense.

> and then also declaring a 
> getMimeHeaders() on Attachment. What do you think?

What would be a typical use case for specifying *different* MIME headers
on an attachement versus the associated message?

/Eoghan
 
> - Dan
> 
> Glynn, Eoghan wrote:
> 
> >Folks,
> >
> >In the old SOAP binding we used
> >javax.xml.soap.SOAPMessage.getMimeHeaders() as a bucket for 
> >transport-specific headers, so for example the WS-A layer could 
> >propogate the wsa:Action to the transport as the SOAPAction header.
> >
> >We no longer seem to do this in the new SOAP binding as a matter of 
> >course. The closest analogue I can find is in the 
> AttachmentTest unit 
> >test, but this hijacks the HTTP_REQUEST_HEADERS property as 
> opposed to 
> >using a separate SOAP-specific headers map.
> >
> >Is it an oversight that we don't current do something like 
> >org.apache.cxf.soap.SoapMessage.setHeaders(MimeHeaders.class,
> >mimeHeaders) in one of the SOAP interceptors?
> >
> >Cheers,
> >Eoghan
> >
> >
> >  
> >
> 
> 
> --
> Dan Diephouse
> (616) 971-2053
> Envoi Solutions LLC
> http://netzooid.com
> 
> 
> 

Reply via email to