Hi, when I've changed to the new released version (in fact a SNAPSHOT) of Muse and I try to compile the generated code for the portal and system services, even do not changing anything in the generated code, I get an error indicating that the "DeploymentFaultException" file is missing (it was not generated). I saw that the "ComponentFaultException" was generated. I think that maybe there is a problem with the deployment-api.wsdl, in the part: <wsdl:message name="DeploymentFault"> <wsdl:part name="fault" element="cmp:ComponentFault"/> </wsdl:message> When I look at the component-model.xsd file cmp:ComponentFault element, I see that its type is cmp:DeploymentFaultType. I'm trying to solve it in several ways, but with my changes I get other exceptions. Does someone have an idea? Thanks, Ayla
Ayla Debora Dantas de Souza - Projeto Ourgrid wrote:
Hi,
when I've changed to the new released version (in fact a SNAPSHOT) of Muse and I try to compile the generated code for the portal and system services, even do not changing anything in the generated code, I get an error indicating that the "DeploymentFaultException" file is missing (it was not generated). I saw that the "ComponentFaultException" was generated. I think that maybe there is a problem with the deployment-api.wsdl, in the part:
<wsdl:message name="DeploymentFault"> <wsdl:part name="fault" element="cmp:ComponentFault"/> </wsdl:message>
When I look at the component-model.xsd file cmp:ComponentFault element, I see that its type is cmp:DeploymentFaultType.
I'm trying to solve it in several ways, but with my changes I get other exceptions. Does someone have an idea?
I think you are in the wonderful world of WSDL2Java generated stuff. Ed and I wrote paper critiquing that very thing ( http://www.hpl.hp.com/techreports/2005/HPL-2005-83.html ) the short term solution is to raise it with the muse mail list; point them at the XSD/WSDL and see if it is a bug on their side. I suspect they could do with more test WSDL/XSD and so to get ours in there would be a good bit of integration. Long term (2 weeks plus), well, we need to do a new SOAP stack, dont we?
Steve Loughran wrote:
Ayla Debora Dantas de Souza - Projeto Ourgrid wrote:
Hi,
when I've changed to the new released version (in fact a SNAPSHOT) of Muse and I try to compile the generated code for the portal and system services, even do not changing anything in the generated code, I get an error indicating that the "DeploymentFaultException" file is missing (it was not generated). I saw that the "ComponentFaultException" was generated. I think that maybe there is a problem with the deployment-api.wsdl, in the part:
<wsdl:message name="DeploymentFault"> <wsdl:part name="fault" element="cmp:ComponentFault"/> </wsdl:message>
When I look at the component-model.xsd file cmp:ComponentFault element, I see that its type is cmp:DeploymentFaultType.
I'm trying to solve it in several ways, but with my changes I get other exceptions. Does someone have an idea?
I think you are in the wonderful world of WSDL2Java generated stuff. Ed and I wrote paper critiquing that very thing ( http://www.hpl.hp.com/techreports/2005/HPL-2005-83.html )
the short term solution is to raise it with the muse mail list; point them at the XSD/WSDL and see if it is a bug on their side. I suspect they could do with more test WSDL/XSD and so to get ours in there would be a good bit of integration.
Long term (2 weeks plus), well, we need to do a new SOAP stack, dont we?
Hi Steve!!! We all read your paper and agree with it!! :-) But, Muse assumes we start with XML, and generates the Java code that you hook the business logic. So, we feel that this much less of a problem than the Java to XML mapping of JAX-RPC. Isn't this so? Is there anyhing we are missing? []s Walfredo
Walfredo Cirne wrote:
Steve Loughran wrote:
Long term (2 weeks plus), well, we need to do a new SOAP stack, dont we?
Hi Steve!!!
We all read your paper and agree with it!! :-) But, Muse assumes we start with XML, and generates the Java code that you hook the business logic. So, we feel that this much less of a problem than the Java to XML mapping of JAX-RPC. Isn't this so? Is there anyhing we are missing?
yes, the bit where I explained how the generated code is still inordinately brittle and of limited value or quality. Have a look in smartfrog CVS at the CDL implementation (core/extras/cdl/). I'm using Xom and have my own classes to represent the various xml elements we get. But they are the elements the Xom code returns when it parses the document...the parser generates what looks like a normal Xom tree, but it has our custom classes in, classes that contain the CDL-specific logic. And they are also xpath addressible, so document.getPath("/cdl:configuration/tns:component/") works to resolve things and return something I can cast to my PropertyList and work with. -steve
Hi, So, do you think Muse is a dead end? I created by hand the "DeploymentFaultException" based on the "ComponentFaultException" which was generated and the compilation and deployment with Muse worked out. It may be a problem with Muse, but I thought it was strange to see a message called "DeploymentFault" pointing to an element called "ComponentFault" which was of the type "DeploymentFaultType". That's why I'm not sure that this is not a problem with the specification. Does anyone see any problem in this part of the specification? Regarding Muse use, I really appreciate your opinion about it. This latest version seems to be better, but now I would have to learn how to communicate with the deployed services and insert code into the generated code which will actually be the deployment engine. Steve gave me the advice of looking at their source code, as Muse documentation is not complete. Ayla Steve Loughran wrote:
Ayla Debora Dantas de Souza - Projeto Ourgrid wrote:
Hi,
when I've changed to the new released version (in fact a SNAPSHOT) of Muse and I try to compile the generated code for the portal and system services, even do not changing anything in the generated code, I get an error indicating that the "DeploymentFaultException" file is missing (it was not generated). I saw that the "ComponentFaultException" was generated. I think that maybe there is a problem with the deployment-api.wsdl, in the part:
<wsdl:message name="DeploymentFault"> <wsdl:part name="fault" element="cmp:ComponentFault"/> </wsdl:message>
When I look at the component-model.xsd file cmp:ComponentFault element, I see that its type is cmp:DeploymentFaultType.
I'm trying to solve it in several ways, but with my changes I get other exceptions. Does someone have an idea?
I think you are in the wonderful world of WSDL2Java generated stuff. Ed and I wrote paper critiquing that very thing ( http://www.hpl.hp.com/techreports/2005/HPL-2005-83.html )
the short term solution is to raise it with the muse mail list; point them at the XSD/WSDL and see if it is a bug on their side. I suspect they could do with more test WSDL/XSD and so to get ours in there would be a good bit of integration.
Long term (2 weeks plus), well, we need to do a new SOAP stack, dont we?
Ayla Debora Dantas de Souza - Projeto Ourgrid wrote:
Hi,
So, do you think Muse is a dead end?
I created by hand the "DeploymentFaultException" based on the "ComponentFaultException" which was generated and the compilation and deployment with Muse worked out. It may be a problem with Muse, but I thought it was strange to see a message called "DeploymentFault" pointing to an element called "ComponentFault" which was of the type "DeploymentFaultType". That's why I'm not sure that this is not a problem with the specification. Does anyone see any problem in this part of the specification?
Regarding Muse use, I really appreciate your opinion about it. This latest version seems to be better, but now I would have to learn how to communicate with the deployed services and insert code into the generated code which will actually be the deployment engine. Steve gave me the advice of looking at their source code, as Muse documentation is not complete.
Ayla
I dont think muse is a dead end, i havent used it and have no opinions on that. My issue is that underneath it (and globus, and lots of other things) are built on JAXRPC, and that is what I think is the wrong underpinning. A lot of the suffering in muse is a consequence of that underlying architecture, not of muse's design itself. -steve
participants (3)
-
Ayla Debora Dantas de Souza - Projeto Ourgrid -
Steve Loughran -
Walfredo Cirne