tag:blogger.com,1999:blog-5763764290649132593.post8906769093186801959..comments2009-11-11T09:52:46.284+01:00Jonas Bandihttp://www.blogger.com/profile/00990537252799084615blog@jonasbandi.netBlogger2125tag:blogger.com,1999:blog-5763764290649132593.post-12225015646516350932009-08-18T00:31:19.622+02:002009-08-18T00:31:19.622+02:00@Laird<br /><br />Sorry, I cannot reproduce your statement.<br /><br />If I call<br /><br />p.put(&quot;openejb.embedded.initialcontext.close&quot;, &quot;destroy&quot;);<br /><br />before creating my InitialContext. And then call<br /><br />context.close();<br /><br />at the end of each test.<br />Then the persistence-unit gets deployed for each test. If you then have your persistence-provider configured to newly create the schema at deployment, you should not have any data surviving between two tests ...Jonas Bandihttp://www.blogger.com/profile/00990537252799084615noreply@blogger.comtag:blogger.com,1999:blog-5763764290649132593.post-74856409839391947652009-08-17T19:45:53.011+02:002009-08-17T19:45:53.011+02:00Hi; excellent article.<br /><br />Have you observed that the OpenJPA provider is not restarted when you do this? I&#39;m trying to work around this.<br /><br />That is, although a new EJB container is started for each test, the underlying JPA provider (OpenJPA) survives Context.close(). This means that data inserted into an in-memory H2 database, for example, can be seen by two test methods. I haven&#39;t found a way to solve this problem.Laird Nelsonhttp://www.blogger.com/profile/07905132403257139697noreply@blogger.com