Hi Sanjiva,
On 2 Apr 2006, at 8:38, Sanjiva Weerawarana wrote:
On Sun, 2006-04-02 at 08:31 +0100, Michel.Drescher (JIRA) wrote:
[ http://issues.apache.org/jira/browse/AXIS2-540?
page=comments#action_12372822 ]
Michel.Drescher commented on AXIS2-540:
---------------------------------------
Still, the address property of the EPR is NOT determining the
transport
protocol, whether WSDL is involved or not. THis is just the wrong
way
to do it, even if I would program against the API.
I guess I don't understand ... in *most* cases, the address
property of
the EPR *does* determine the protocol and the physical address as
well.
In some cases (like having an explicit proxy of some sort) it
doesn't -
and we have a property that can be set to indicate that address.
The fact that an EPR is by about 99% a URL does not mean you can
extract its scheme, see that it is "http" and just go along with this.
The EPR's address property is a URI, and as such, just a string with
a defined formatting. This string, however, is just an identifier
with no additional sematics (as opposed to a URL, which incorporates
a scheme/transport id, authority, etc.) An example for such
identifying strings are XML namespaces, which are in overwhelming
majority URLs (because they offer intrinsic uniqueness).
But still, they are just identifier, nothing more.
As an example, WSRF uses WS-Addressing to identify the resource to
which to send the messages. It explicitly does *not* mention any
transport layer, as WSRF is itself transport agnostic. This way, WSRF
allows you to model your Web Service(s) so that you either offer one
Web Service per Resource, or one Web Service for all your resources.
Does this make it clearer, what I mean?
Cheers,
Michel
