dev@jax-ws.java.net

Re: Possible bugs in MtomEncoder?

From: Vivek Pandey <Vivek.Pandey_at_Sun.COM>
Date: Thu, 23 Feb 2006 09:19:54 -0800

Paul Sandoz wrote:
> Vivek Pandey wrote:
>> Paul Sandoz wrote:
>>
>>> Hi,
>>>
>>> Should the MtomEncoder assume that the charset is UTF-8 for the SOAP
>>> message? What if we want to create UTF-16 encoded XML SOAP messages?
>>>
>> I dont think so. We dont have programming model to allow encoding
>> anything other than utf-8.
>
> Not yet :-) I see that UTF-8 is a hard-coded assumption throughout the
> code :-(
>
Probably you're talking about chaset parameter on Content-Type, where
else? One place I can think of is that the XMLStreamWriter should be
created using specified encoding using OutputStremaWriter using
scpecifed charset.
>
>> This is ok even from policy assertion point of view where it says
>> andpoint should be able to digest utf8,utf16BE and utf16LE. As far as
>> sending is concerned we can keep sending utf8.
>>
>
> I am thinking from the perspective of Japanese or Korean clients that
> want to produce SOAP messages encoded in UTF-16 for efficiency reasons.
>
> It seems like something that could be configurable via client policy
> on the binding.
>
>
AFAIK, there is no such policy that tells what encoding to use to send
the message. I agree it will be good to have one. I know in indigo you
can to it using some configuration setting in.

-vivek.

>>> Should the Content-Type header of the SOAP 1.2 part include the
>>> 'action' parameter? See the example here [1], where 'action' is
>>> included in both cases. Not sure it is required (the spec does not
>>> say) and it certainly makes it easier if it is not.
>>>
>> Good catch. Spec says that[1], though indigo doesnt emit it. I wil
>> make this change anyway. . The action should go as parameter on to
>> Multipart/Related header and also on the Content-Type mime header of
>> the root part.
>>
>
> OK.
>
> Paul.
>

-- 
Vivek Pandey
Web Services Standards and Technologies
Sun Microsystems Inc.