12:33 p.m.
Hibernate Search test suite used to run in about 2 mins. Now it runs
in a solid 7 minutes. Can we get that number back?
It seems like maven is pausing between tests or something like that.
10:35 a.m.
On Sun, 23 Aug 2009 12:33:05 +0200, Emmanuel Bernard
<emmanuel(a)hibernate.org> wrote:
Hibernate Search test suite used to run in about 2 mins. Now it runs
in a solid 7 minutes. Can we get that number back?
The problem is that in forkMode=once some of the tests are failing -
LazyCollectionsUpdatingTest
and EmbeddedTest. The tests themselves are ok. The problem is some
interdependency between previous tests.
For the sake of a consistent build (including a stable Hudson build) we
set the fork mode to 'perTest'.
--Hardy
1:18 p.m.
Any hint on why this is failing? The most sensitive tests are the perf
tests and the JMS ones. The others should be fixable wo too much effort.
On 24 août 09, at 10:35, Hardy Ferentschik wrote:
On Sun, 23 Aug 2009 12:33:05 +0200, Emmanuel Bernard
<emmanuel(a)hibernate.org
> wrote:
> Hibernate Search test suite used to run in about 2 mins. Now it runs
> in a solid 7 minutes. Can we get that number back?
The problem is that in forkMode=once some of the tests are failing -
LazyCollectionsUpdatingTest
and EmbeddedTest. The tests themselves are ok. The problem is some
interdependency between previous tests.
For the sake of a consistent build (including a stable Hudson build)
we set the fork mode to 'perTest'.
--Hardy
4:16 p.m.
If tests rely on the state from previous tests, that's a bad setup. The
issue we have seen is that the tests can be run in different orders
based on the jvm used.
On Mon, 2009-08-24 at 13:18 +0200, Emmanuel Bernard wrote:
Any hint on why this is failing? The most sensitive tests are the
perf
--
Steve Ebersole <steve(a)hibernate.org>
Hibernate.org
4:37 p.m.
Also note that JUnit explicitly does not support this. TestNG *can* but
you have to explicitly tell it about the dependency from one test to the
next.
On Mon, 2009-08-24 at 09:16 -0500, Steve Ebersole wrote:
If tests rely on the state from previous tests, that's a bad
setup. The
issue we have seen is that the tests can be run in different orders
based on the jvm used.
On Mon, 2009-08-24 at 13:18 +0200, Emmanuel Bernard wrote:
> Any hint on why this is failing? The most sensitive tests are the
> perf
--
Steve Ebersole <steve(a)hibernate.org>
Hibernate.org
5638
days inactive
5639
days old
4 comments
3 participants
participants (3)
-
Emmanuel Bernard -
Hardy Ferentschik -
Steve Ebersole