Hi Monosij,
See comments inline.
Simon
Monosij Dutta-Roy wrote:
hi Ant, Simon -
I wanted to get back to you regards your answers on creating the webapp
with multiple components.
I hope you remember my situation with the webapp running multiple
contributions that I contacted you last week about.
---------------------------
As per Ant's suggestion I moved to Tuscany version -- > 2.0-Beta2
And also updated namespace as: -->
xmlns="http://docs.oasis-open.org/ns/opencsa/sca/200912";
Also should this change to 2.0 or some
other? xmlns:tuscany="http://tuscany.apache.org/xmlns/sca/1.1";
1.1 should be fine.
However now the basic webapp while deploys in Jetty - gives an error on
invocation:
Problem accessing /qmAppSCA01/.
Reason: java.lang.IllegalArgumentException: Business interface
org.rd.qm.impl.QueryServiceImpl is not compatible with
org.rd.qm.QueryService
I have not changed the interface and classes at all from my previous
version which used to be invoked fine by the JSP. Again this is the
basic Tuscany webapp template created by Maven.
---------------------------
I have the @Remotable on the interface org.rd.qm.QueryService as below:
@Remotable
public interface QueryService {
String getPatients();
}
and the impl:
public class QueryServiceImpl implements QueryService {
public String getPatients() {
return "Got Patient!";
}
}
---------------------------
The qmAppSCA01.composite is attached.
Again - this basic webapp configuration was working fine in M4 which I
was running before. Am I missing something?
>
I think you may need to add the annotation @Service(QueryService.class)
to the QueryServiceImpl class. This changed in the OASIS SCA specs
some time ago, and it may be that this change wasn't in 2.0-M4 but is
in 2.0-Beta2.
---------------------------
Also regards my initial queries ANT ELDER wrote:
There is a way to do this since 2.0-M5 by putting the contributions into
a folder name WEB-INF/sca-contributions. There is a sample which
demonstrates this that we have just added to the trunk samples, so its
pom.xml shows what you need to do, see
https://svn.apache.org/repos/asf/tuscany/sca-java-2.x/trunk/samples/getting-started/helloworld-webapp/
-----
I haven't gotten there yet.
However SIMON NASH wrote:
My understanding is that the JSP invokes QueryService correctly, but
QueryService can't invoke QueryController because QueryController isn't
part of the same webapp. To fix this, you could expose QueryController
using an interoperable binding such as binding.ws <http://binding.ws>,
and configure the reference in QueryService to also use binding.ws
<http://binding.ws> with an endpoint URI that matches the deployment URI
for QueryController. You would also need to use a separate Tuscany
domain (not part of the webapp) to load and deploy QueryController.
In Tuscany 2.x (not 1.x) there is an SCA client API which allows you to
invoke SCA services running in a different SCA domain. This is intended
for use from non-SCA code, such as a servlet running in a webapp. As
you are using some SCA code within the webapp (though only as a UI front
end) I think it's better to use an interoperable binding to invoke
QueryController rather than using the SCA client API.
---------------------------
Clarification seeking on which is the way to go - with my interpretation as:
ANT's statement says that with 2.0-Beta2 I should be able to do multiple
components / composites for a webapp as long as I put them in the
WEB-INF/sca-contributions folder. So it should work just by putting it
there and I intend to try it out as per the example.
However, I am interpreting SIMON's statement as a best practice - and
will be necessary to run the QueryController in a scalable manner such
as multiple nodes. So in this approach the QueryController has nothing
to do with the webapp but will be invoked by the webapp as needed
through the binding.ws <http://binding.ws> - and this leaves me free to
change webapp framework / UI without worrying about backend logic.
However just for now to get things running and make sure configs and
such are set correctly I can get by using the approach Ant suggests.
However this approach won't be scalable in a way SCA is meant to scale -
but will be good for the prototyping phase - including set up
db interconnectivity through DAS and such - and then when ready figure
out the deployment / scaling aspect in running across multiple nodes put
in the ws.bindings and deployment scenarios.
Please let me know if my interpretation is correct.
---------------------------
Yes, that's a very good summary and it's exactly correct.
Simon
Again thanks for all your help and suggestions.
monosij
