On 9 Jul 2010, at 16:36, Sanne Grinovero wrote:
2010/7/9 Manik Surtani <manik(a)jboss.org>:
>
> On 9 Jul 2010, at 08:45, Mircea Markus wrote:
>
>>> Carrying over flags is the generic fix, since it covers both the case of a tx
being present and not. So this should be done anyway, but we should also look at why a
new context is created if there is no ongoing tx.
>> If there's no ongoing tx then BatchingInterceptor creates one and then
invokes next interceptor, so that all interceptors in the stack will be executed in a
transactional context. This is where a new context (TxInvocationContext) is created, as
the existing context is not transactional: NonTxInvocationContext.
>
> Ah ok, that makes sense. In any case flags should be copied across.
>
thanks for the clarifications on the new context; so I created ISPN-535
I'd like to commit this one-liner fix + test, still I'm starting the unit tests
and a variable number of tests is failing. I remember this was the
case some time ago, no consistency yet?
I'm afraid of committing this w/o someone confirming this is still normal :/
If the inconsistent tests are in the distribution package, then this is expected. Those
tests are flaky and fail occasionally; I need to look at refactoring them. As long as the
other tests run clean you should be good.
Cheers
Manik
--
Manik Surtani
manik(a)jboss.org
Lead, Infinispan
Lead, JBoss Cache
http://www.infinispan.org
http://www.jbosscache.org