[cddlm] third pass at jun's slides

[Ayla Debora Dantas de Souza - Projeto Ourgrid]

Hi,

I've updated Steve's slides with our current status.
Regards,
  Ayla

Steve Loughran wrote:

>
> hello all
>
> I've updated the slides with the latest test results from my system, 
> and tweaked some other things
>  -put in current interop test results and URL to full report
>  -some textual commentary in the notes page expands on the details.
>  -added '$Vendor Test Results' on each test result slides for 
> $Vendor={hp,ucfg,nec}
>  -made the TODO page an HP TODO page that lists things I have to do
>  -made the issues page purely one about communal testing issues, 
> things that would be of interest to anyone doing interop testing
>
> 1. we need better access to each other's logs. Otherwise you have to 
> try and get the other teams to rummage in their
>
> I am actually publishing a log dir:
> http://deployapi.iseran.com:8080/logs/
>
> However, on the public endpoint, log4J is not logging their, even 
> though it does on my staging system. Some research needed there, but a 
> fix should go in before the end of the week, CVS-problems 
> notwithstanding.
>
> 2. Failing tests cannot distinguish why tests failed, because we dont 
> have standard fault codes. Example: imagine the test to submit a CDL 
> doc with an incomplete non-lazy reference. This must not enter the 
> running state, so initalize must fail. But without an error code test, 
> you cannot distinguish "failed with invalid reference" from "failed 
> because the XML parser is broken". These generates false positives. 
> There is no fix for this, other than standard fault codes.
>
>
> 3. having continuous integration servers redeploying the servers and 
> rerunning the tests all the time would help the other teams. For 
> example, if my work box ran the hp client interop tests every half 
> hour, the published results would always be up to date, and the cycle 
> time for anyone testing their server would be much shorter.
>
> -steve
>