Re: SOAP implementation for Vienna & SOAP details

[Michael Joya <mike.joya@xmlglobal.com>]

Krishna Sankar wrote:

>         1.      User Apache/SOAP and capture the XML document via a facade pattern and
> then process it as you wish.
>         2.      With Axis, one can register handlers and then the handlers will get the
> document and a context.
>         3.      Process the XML document from an HTTP POST
> 
>         Options 2 and 3 are more flexible, but #1 might be faster. I don't know if
> anyone has done #1 with an ebXML TRP 0.98  document yet. I am working on #2
> and #3 ;-0. As the Axis project is still in it's infancy, #3 gives us some
> flexibility.

>         On a related note, do you plan to show reliable messaging as well as some
> security ? If so, #3 looks like a good choice. You need to validate the XML
> Payload and get the required elements from the DOM.

  I'm doing 3. I will bring source code with me to Vienna so that I am flexible to changes at the last minute. I have some issues to raise though. They regard namespaces and DOM:

  The prefixes "SOAP-ENV:" and "eb:" are used to denote namespace scopes in the header document. Although it is technically valid to use any prefix you like, (ie: "soapenv:",  "Eb:") some DOM implementations tend to choke on the ability to properly correlate the prefix to the namespace. (See Apache Xerces, IBM XML4J, ) This isn't a problem with implementation. This is a problem with DOM.

   Would it be fair to agree that the prefixes must be "SOAP-ENV:" and "eb:" as given in the spec? I've had to make this temporary comprimise where I hardcode the prefix into the tagname itself. I suspsect other DOM implementors must do the same. If anyone knows of a web resource detailing another way around this while still using DOM, please send it to the list.



-- 
// Michael Joya
// XML Global Research and Development
// 1818 Cornwall Ave. Suite 9
// Vancouver, Canada
// 604-717-1100x230