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