Bryan Pendleton wrote:
I will try to see if the problem reproduces without your 1080 fix.
Hi Sunitha,
Yes, the problem reproduces without your fix. The output files are a bit
different, but the symptom is the same: the test mysteriously exits
without
producing all the required output, and no other messages are produced.
I think in a way this is good news, because it indicates that your
DERBY-1080 changes are not implicated in the problem.
How should we proceed?
Thanks Bryan for trying this out without the 1080 fix.
So,
-- DERBY-1080 does not cause the strange test exit in testSecMec.java.
-- I do think that the networkserver.shutdown and how it affects streams
needs to be investigated further.
Will the below be reasonable ways to proceed ?
-- this fix (derby1080.2.diff.txt ) can be committed now as is.
-- In the testSecMec test, I dont see why shutdown should throw an
exception. I can change the test to not do this switching of streams,
so hopefully test will not exit. If this seems ok, I can submit a test
only patch for this.
-- open a jira for further investigation of networkserver.shutdown and
how it affects streams.
Thanks,
Sunitha.
