[cddlm] Notes from the meeting on 5/31/2006

[Steve Loughran]

Milojicic, Dejan S (HP Labs) wrote:
> Attendees: Satish (NEC), Flavio, Guilherme (UFCG), Dejan (HP)
> 
> Agenda:
> 
> 1. Update: 
> 
> UFCG: no significant progress with component model because they missing
> files from others, from everyone. ACTION item: get files from Satish,
> Steve, and Stuart, so that UFCG can make some progress. (Satish: some of
> these cases, such as switch and delegate still haven't been implemented
> in the parser, so right now he does not have these files, will probably
> work on these this week and send them to Flavio. Others (STUART AND
> STEVE): PLEASE REPLY BY EMAIL WHEN YOU WILL PROVIDE THE FILES). Flavio
> will explicitly request these files. Guilherme testing against the NEC
> endpoint broke again, not sure what is it about. The interval was 2
> minutes. Satish requested to get the sample tests and possibly trace.
> Satish was testing from inside and did not get problems. 

I cant reach the NEC endpoint from behind my firewall; can anyone else 
see it?

I think the proxy server here is complicating the reporting, because 
when I telnet from a public site I get a connection but no response

 > telnet cddlm.nec-labs.com 9090
Trying 138.15.110.100...
Connected to cddlm.nec-labs.com.
Escape character is '^]'.
GET /cddlm/services/portal HTTP/1.0


...and I can sit there indefinately waiting for a reply. Indeed, my test 
client does sit there indefinately, which is a bug somewhere in apache 
httpclient.

Actually, the UFCG epr isnt reachable either, even when I ssh to a site 
beyond the firewall and try and telnet to port 8080

 > uname -a
FreeBSD minotaur.apache.org 6.1-RELEASE FreeBSD 6.1-RELEASE #0: Thu May 
11 11:50:25 PDT 2006 
root at minotaur.apache.org:/usr/obj/usr/src/sys/SMP-turbo  i386

 > telnet cddlm.lsd.ufcg.edu.br 8080
Trying 150.165.15.37...
telnet: connect to address 150.165.15.37: Connection refused
telnet: Unable to connect to remote host


> 
> NEC: Wrt UFCG, he did deployment test using CDL that Flavio sent a few
> weeks ago, is trying to reproduce the problems and it is failing but
> still does not understand what it is. The problem is with not being able
> to deployment components on each other's components. Deployment APIs are
> 75% on HP endpoint, but not wrt component model. Asked Steve with how to
> resolve the issue with him not being able to talk to other endpoints,
> but still didn't get response.

1. I need to be able to talk to the endpoints

2. I can then use axis tcpmon to get a trace record of the message that 
is sent and the fault that is sent back.

3. I can send the record of the message to the other endpoint to see if 
they can replicate the problem. If we can find out what line is throwing 
the real fault, its possible to track down the underlying cause, which 
is possible/probable that muse and I have different misinterpretations 
of the WSRF and WSA specifications. I will have to workaround whatever 
they are, regardless of whose fault it is.

> 
> 2. We received feedback on the CDL spec and Jun should take action on
> it. Jun, can you please let us know approximately when you will be able
> to address the questions raised.

Is this the ongoing discussion about importing into namespaces?

> 
> 3. EVERYONE: PLEASE TRY TO ATTEND NEXT MEETING. WE ARE CLOSE TO OUR
> MILESTONE TO COMPLETE TESTING SO THAT WE CAN START WRITING INFORMAL
> DOCUMENT SUMMARIZING INTEROPERABILITY!
> 

I've been on holiday.

-steve