perfsonar development work

[ulisses <>]

Hi Jason

On 2006-08-27 22:15:42, Jason Zurawski wrote:
[...]
> As I said in the previous e-mail I see no errors when doing a simple 
> conversion using the Handler class and the toDOM() routine, but I am 
> curious to see what method you are using that invokes these problems.

In my performAction method, I only return a message.


> >>* Have you defined all possible message sequences for your service?
> >>    
> >
> >it is an stateless service for now, do you refer to this?
> >
> >  
> 
> 
> By message sequences Loukik means what the purpose of the request and 
> the response are, and what they general format should be.  I don't 
> believe you have ever roughed out an 'expected' sequence of exchanges to 
> this service (i.e. a request document format and the resulting response 
> document format).  

for now, it's only one request:

        - "please dump me the entire topology" 

and one reply:

        - "just dump all elements"
        
that are the current requirements

> Even though you have already started the physical 
> service, it would be nice for me to see what your request and responses 
> would look like.  

future developments can include:

- Retrieval specific elements of the topology, that is do partial download of 
the database

        From 
http://wiki.perfsonar.net/jra1-wiki/index.php/Topology_Service_resources # 
Required interactions
        #  Retrieval of topology information: Clients need to have the 
possibility to retrieve network topology information. It needs to be 
discussed which granularity is needed (whole networks, a component and all 
neighbors, etc).

- Register at the Tops in order to be notified when the topology or an 
specific element of the topology has been updated

        From 
http://wiki.perfsonar.net/jra1-wiki/index.php/Topology_Service_resources # 
Required interactions
        # Client notification: It might be interesting for clients to 
register themselves at a ToS and to be informed about changes in the topology 
automatically. 


> That way we could decided how many data elements you 
> would need, and if your metadata blocks need to be blank or not. 

please tell me what steps I have to do.

> >>* Are these message sequences agreed upon by NMWG (i.e Martin and Jason)
> >>    
> >
> >maybe you are refering to the xml format/schema used?
> >
> >The following is a document with a more complete xml, including references 
> >within the document,
> >even Jason told me I should use 
> >org.ggf.ns.nmwg.topology.l3.v3_0.Interface, currently I'm using
> >org.ggf.ns.nmwg.topology.base.v3_0.Interface which seems to does the trick.
> >
> >  
> 
> Yes, you are using the topology elements from that particular schema, 
> but the actual exchanges that this service will have between client 
> applications or other services is very important as well.  Take a look 
> at the (many) examples of exchanges that exist for the LS, or the 
> various MAs.  Each request example is followed by an appropriate 
> response instance document. 

I don't know what that means in my case. Could you tell me an specific 
example.

> >>* If the above two points are done, are you sure your service is 
> >>producing xml messages which are conforming to the agreements you might 
> >>have had with NMWG?
> >>    
> >
> >I think I don't understand what you mean with message sequences, do you 
> >refer to xml schema?
> >
> >  
> 
> Message sequence would be TOPOQueryRequest -->  TOPOQueryResponse or 
> something similar.  Once the actual messages are rouged out, then a 
> specific schema (beyond the general topology elements) could be created 
> just for your service. 

ok, that is what I understand with a sequence, but because now I only have a 
single message an reply I was confused.

well, It is the first time I develop with java/web services but maybe the 
following 
could be an starting point:

        ---> means request
        <--- means reply

full database download (being implemented)

        ---> TOPSDownloadDBRequest
        <--- TOPSDownloadDBResponse

partial database download, I understand that is downloading nodes or links 
that 
matches some criteria.

        ---> TopsDownloadNodesRequest
        ---> TopsDownloadLinksRequest

        <--- in both cases the output could be the same as in 
TOPSDownloadDBResponse
             but only showing the matching elements.

register for database change, entire, nodes or links database respectively.

        ---> TopsRegisterDBModification
        ---> TopsRegisterNodesModification
        ---> TopsRegisterLinks Modification

        the client should include an URL for which the topology should issue 
a GET
        method if the modification is done. The client, upon a GET method 
request, should
        issue a TOPSDownloadDBRequest, TopsDownloadNodesRequest or 
TopsDownloadLinksRequest

        <--- The reply would be an absolute time reference which the register 
is value.

does this looks right for you?

        Ulisses

PD: I have to leave the office now.

> >>Finally, if all above are ok, what is the error message being produced 
> >>(look at catalina.out if you are testing it on a deployed service)
> >>    
> >
> >I think that I already asked you previously if it was possible force the 
> >XML to DOM parser to tell in which line it is the error?
> >
> >The output is just:
> >
> ><?xml version="1.0" encoding="UTF-8"?>
> ><nmwg:message xmlns:nmwg="http://ggf.org/ns/nmwg/base/2.0/"; 
> >id="resultCodeMessage">
> >   <nmwg:metadata id="resultCodeMetadata">
> >      <nmwg:eventType>error.common.parse_error</nmwg:eventType>
> >   </nmwg:metadata>
> >   <nmwg:data id="resultDescriptionData_for_resultCodeMetadata" 
> >   metadataIdRef="resultCodeMetadata">
> >      <nmwg:datum value="Cannot convert Message to DOM" />
> >   </nmwg:data>
> ></nmwg:message>
> >
> >I don't see any XML/DOM errors in the tomcat logs
> >  
> 
> I see that you already filled a bug on the issue of 'line numbers' for 
> the output errors.  I will see what I can do on that front, but perhaps 
> someone that knows a bit better could clue us in as to which service 
> component would generate the 'error.common.parse_error' to begin with?  
> It does not get thrown in the base GGF classes, so I am guessing it is 
> related to either a utility call or something in the axis libs? 
> 
> -jason
>