Re: Minutes - WS-Addressing Test Suite TF call 8 Dec 2005

See inline ...

paul.downey@bt.com wrote:
> Arun,
> 
> 
>>With this, can I use WS-I monitor to log my SOAP messages and then use 
>>the supplied stylesheet to transform the monitor log into the required 
>>format ?
> 
> 
> yup, stylesheet here:
> http://www.w3.org/2002/ws/addr/testsuite/logs/ws-i/ws-i.xsl
> 
> example output here:
> http://www.w3.org/2002/ws/addr/testsuite/logs/ws-i/ws-i.xsl
> 
> 
>>I understand there could be multiple approaches but is the following one 
>>a correct approach ?
> 
> 
> What's not 'correct' about it?
I think this is correct and wanted to validate before I embark upon it :)

> 
> 
>>As a test developer, I focus on writing my tests (whether I start from 
>>WSDL or not is an implementation detail). WS-I monitor can be used to 
>>capture SOAP logs. ws-i.xsl will transform the captured monitor log into 
>>another XML document understood by observer.xsl. Finally, observer.xsl 
>>will parse this newly created XML log file and report which tests are 
>>passing and failing.
> 
> 
> Yes, except we don't require the WS-I monitor to capture the logs, you
> can use tcpdump, SOAPscope, YATT, tcptrace, Ethereal, libpcap, whatever 
> suites you, and if your service or client captures messages then you can 
> generate a log from that. 
If I use some other tool to capture the logs, then I need to write my 
own stylesheet to convert the logs to the format required by observer. 
Is that right ?

> 
> 
>>What are the other alternatives to generate pass/fail results ?
> 
> 
> You can check the messages by hand, if you prefer. If you have an 
> alternative proposal for CR testing, then I suggest you put a proposal
> to the Working Group, and we can discuss it.
I'm fine with the existing proposal.

> 
> 
>>>Arun's coordination concern will be addressed by having at least one
>>>public endpoint for people to check their implementations prior to the event.
> 
> 
>>Not sure how this addresse my concern. I was more interested in finding 
>>out how will the implementers communicate with each other. In debugging 
>>a test case failure, I'm guessing we'll be talking specifics about our 
>>products and platforms and some of us may not like to share the 
>>information publicly. Sun's product is all open sourced so I've no 
>>inhibitions in talking publicly about it but that may not be the case 
>>for all participants.
> 
> 
> right, 'will' was a little stronger that I might have typed during working
> hours, sorry I meant to say 'should be to a large extent'. 
> 
> I do take your point about debugging, in particular informally before the
> event, so the alternatives of hooking up via direct email, via the list
> or IRC still stand. At the interop "event" itself there will be ample
> chance for coordination F2F, by phone, Zakim, IRC, IM, etc.
> 
> 
>>In the WS-I Sample Apps WG, we created an alias only for the 
>>implementers since the overall WG alias had approx 200 members. Are we 
>>going to do something similar here as well ?
> 
> 
> Well we do have this public list for discussion of testing. 
> 
> I'm wondering what more you are asking for - another list, just for 
> implementers? Would the list be public, archived, W3C member only,
> WG member only or a controlled, private subscription list just for 
> "implementers"? I suggest talking to Hugo and see what's possible.
So all the technical discussions are expected to happen on this email 
alias. Other than that, there might be some other 1-1 discussions to 
resolve interop bugs. Is that the idea ?

-Arun
> 
> Paul
> 

-- 
got Web Services ?
Download Java Web Services Developer Pack from
http://java.sun.com/webservices

Received on Wednesday, 14 December 2005 04:45:49 UTC