On 29/01/2013 11:17, Andrei Shakirin (JIRA) wrote:
However, I have the following concerns:
1. the 1_0_X branch has the same test that's working flawlessly with JDK 1.6
and 1.7: why? Is this due to the modification made to make tests re-runnable?
The same test works even in trunk under 1.6 and under some other conditions,
but is it reliable way to write integration tests? Basically it SHOULDN'T work
anywhere and runs successfully only occasionally, see [1].
Problem is that some tests can be easily broken by change absolutely not
related with the test itself, like it happens with Task Service implementation
(see jira attachment).
I already have seen the same effect for Notification object [2].
Fine.
2. does it make sense to change something in the core only to have easier and
more reliable tests?
I don't see it as only test related problem. If user configures and uses H2 DB,
is there guarantee that it cannot happens during normal project using as well?
3. I think it will not be so straightforward to make all necessary adjustments
to content.xml in order to cope with new id generation
Could you elaborate this? Which issues are you expecting? Do you see other ways
to fix the problem?
Sorry, I am not following this thread enough to be able to see if there
are other ways to fix; anyway, you (and Fabio) believe so, and for me
it's fine, then.
What I mean above is that I am expecting it will take some time to
adjust the content.xml with ids generated differently: nothing serious,
only time-consuming. But maybe I'm wrong...
[1]
http://syncope-dev.1063484.n5.nabble.com/Question-Problem-with-test-UsertTestITCase-issueSYNCOPE279-tt5712389.html
[2]
http://syncope-dev.1063484.n5.nabble.com/Persistence-id-generation-strategy-TABLE-vs-AUTO-td5711813.html
--
Francesco Chicchiriccò
ASF Member, Apache Syncope PMC chair, Apache Cocoon PMC Member
http://people.apache.org/~ilgrosso/
