On Jan 8, 2010, at 1:52 PM, Vladimir Blagojevic wrote:
On 2010-01-04, at 1:02 PM, Mircea Markus wrote:
>
> As long as the CM are created through TestCacheManagersFactory, this should will soon
not be an issue:
https://jira.jboss.org/jira/browse/ISPN-314
It still is an issue when tests do not kill cache managers, is it not!
After
https://jira.jboss.org/jira/browse/ISPN-314 will be implemented this should no longer be
an issue.
However, we can still leverage the fact that we use TestCacheManagersFactory throughout.
We can track all DefaultCacheManager(s) we create from TestCacheManagersFactory. We can
use hashmap keyed by test class name
stop right here! :)
this is tricky with TestNG when it comes to inheritance.
BaseTest {
testMethod() {
TestCacheManagersFactory.getCacheManager();
}
}
OtherTest extends BasteTest {
}
when running OtherTest.testMethod, the TestCacheManagersFactory.getCacheManager will
appear as being called from BaseTest rather than the actual class which is OtherTest. This
is why, in JBC , we pass the actual .class as a parameter to
TestCacheManagersFactory.getCacheManager
and a value which is a list of all created DefaultCacheManager (in
all test methods of a test class). Since lifecycle of DefaultCacheManager never spans
across test classes we can find these cache manager and reclaim them in
AbstractInfinispanTest @afterClass method.
_______________________________________________
infinispan-dev mailing list
infinispan-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/infinispan-dev