perfsonar development work

[Maciej Glowiak <>]

Hi Michael,

I think using ant should be necessary during your functional tests 
because we use ant for compilation and running as well as for configuration.

I think it should be a way how functional tests are being done.
So I suggest to use ant especially that you have problems with SOAPClient.

There are targets called "ls-client-run-query" "ls-client-run-register" 
"ls-client-run-deregister" which run SOAPClient directly and can be 
modified for your purposes.

I corrected "success" codes in trunk. Now they're as described in 
documentation.

The problem with no metadataId, you described, is possible. Perhaps LS 
could be more intelligent here. If there were only one metadata and one 
data without identifiers it could assume they should be combined 
together. But in fact, this case is wrong XML.

Maciej

[CC: to the mailing list, just for group information]

Michael Michalis wrote:
> Good morning Maciej,
>
> Thanks for the clarification, I imagined you read something else than the
> one I had written. I have no experience in ant (although I will probably
> need it for coordinating all the tests), so at that point I'm using the
> SOAPClient class, which I can use in iteration in order to send a large
> number of requests. I'll give it a try and refresh the libraries and the
> code. 
> Now the problem with the result codes exist also in the Register and
> keepalive requests where upon successful execution of the request, the
> result code is always result.success and not success.ls.* . 
>
> I'm a little bit lost about some possible bugs in previous requests, since a
> don't now if you already fixed them due to my SOAPClient problem. When I get
> that fixed I'll provide you, the possible bugs or issues I encountered.
> One of my old comments about the LSRegister request which I think is
> important is this:
>
> I believe there is a conflict when the id attribute is missing from the
> metadata element. Because although the handler assigns a new id to the
> metadata element, the data element could contain a metadataIdRef attribute
> which doesn't have the same value as the assigned id attribute. So in this
> case there is no match between the metadata and data element and thus no
> triggering, so the response should be erroneous.
>
> It's not quite a bug or a problem just a remark.
>
>
> -Michalis         
>
> -----Original Message-----
> From: Maciej Glowiak [mailto:] 
> Sent: Tuesday, November 07, 2006 5:44 PM
> To: Michael Michalis
> Subject: Re: LS Functional Tests
>
> Michael,
>
> Sorry for confusion. I read "LSRegister" instead of "LSDeregister" and 
> just looked into the code without checking your requests.
>
> I investigated the problem again today, this time I took LSDeregister...
>
> So:
>
>
> Michael Michalis wrote:
> > Hi Maciej,
> >
> > I m not sure I'm following your answer. I'm referring to the LSDeregister
> > request. My question was why is the response about metadata when the
> request
> > is missing the data element, or the metadataIdRef attribute. The result
> > codes that I think are wrong are the ones that state the successful
> > execution of the request i.e success.ls.register. The service returns only
> a
> > result.success for every type of request. 
>
> 1. LSDeregister always returned result.success, which was, as you said,
>     wrong.
>
>     Now there are two options: "success.ls.deregister" or
>     "error.ls.key_not_found" if key was not found in DB.
>
>
> > Also could you comment on the
> > service behaviour when a non valid or no key is inserted in the request?
>
> Now it should return "error.ls.key_not_found"
>
>
> > Also about the LSUpdate request, as I said in previous email, there seems
> to
> > be a problem of compatibility between the SOAPClient class I'm using to
> feed
> > the requests and the new version of the service you placed on svn. They
> can
> > not communicate with each other(I found that quite painfully when eclipse
> > crashed).
>
> Why don't you use ant targets?
>
> ----------------------------------------------------------------------------
> -----------
> root@reed:/projects/sonar/perfsonar/ant#
>  ant ls-run-client-deregister
>
> Buildfile: build.xml
>
> ls-run-client-deregister:
>       [java] End point: 
> http://reed.man.poznan.pl:8080/axis/services/LookupService
>       [java] Request file: 
> /projects/sonar/perfsonar/schema/example-instances/sonar/LS/test/LSDeregiste
> rRequest.xml
>       [java] Response file: 
> /projects/sonar/perfsonar/schema/example-instances/sonar/LS/test/LSDeregiste
> rResponse.xml
>       [java] time taken :0.179 secs
>       [java] Client exiting
>
> BUILD SUCCESSFUL
> Total time: 2 seconds
> ----------------------------------------------------------------------------
> -----------
>
> and then
>
> ----------------------------------------------------------------------------
> -----------
> root@reed:/projects/sonar/perfsonar/ant# cat 
> /projects/sonar/perfsonar/schema/example-instances/sonar/LS/test/LSDeregiste
> rResponse.xml
>
> <?xml version="1.0" encoding="UTF-8"?>
> <nmwg:message id="msg1_resp" messageIdRef="msg1"
>      type="LSDeregisterResponse" 
> xmlns:nmwg="http://ggf.org/ns/nmwg/base/2.0/";>
>      <nmwg:metadata id="resultCodeMetadata">
>          <nmwg:eventType>error.ls.key_not_found</nmwg:eventType>
>      </nmwg:metadata>
>      <nmwg:data id="resultDescriptionData_for_resultCodeMetadata" 
> metadataIdRef="resultCodeMetadata">
>          <nmwgr:datum 
> xmlns:nmwgr="http://ggf.org/ns/nmwg/result/2.0/";>There were no metadata 
> and data for key 
> [http://not_existing_deregister_key.net:8080/axis/services/MA]</nmwgr:datum>
>      </nmwg:data>
> </nmwg:message>
> ----------------------------------------------------------------------------
> -----------
>
> SOAPClient works fine for me. Maybe you have old classes or old JAR 
> libraries (Axis)? Try to checkout trunk from SVN once again and run "ant 
> libs-xmlls" in order to download required libraries.
>
> > So I can not test the changes and fixes you made. In order to
> > continue the rest of the tests I had to convert back to the latest
> snapshot.
> > Finally I think there are some more issues unanswered about the LSRegister
> > and LSUpdate requests. I'll try to find those points and send them back to
> > you, later today or tomorrow. Thanks for your help.
>
> That's no problem. Sorry once again for answering not the right "topic" :)
>
> If there is something left to be answered, please send me it once again. 
>   I had some problems with my thunderbird and am not sure if I didn't 
> miss something.
>
> Maciej


--

 --------------------------------------------------------------------
| Maciej Glowiak        Network Research and Development            ||
| 

     Poznan Supercomputing and Networking Center ||
| (+48 61) 858 2024     http://monstera.man.poznan.pl/              ||
 ====================================================================