Bryan Pendleton wrote:
Meanwhile, I noticed in John's message earlier today that he has
been able to build a specialized test to stress test this area
of the system:
> I created a smaller stand-alone application "inspired by" the
> really bad parts of the DOTS test
John, is it worth incorporating this test code into the regression suite?
Maybe. I have not studied the details of Deepa's jdbcapi/derbyStress.java
test yet, in which the lack of closing PreparedStatements is being tested (I
think). My test may not be too different, but:
I think my test puts more stress on the Derby language/compilation system and
hence the permanent generation part of the heap than Deepa's repro. I will
upload my test to Jira soon, so that you can take a look.
--
John
