Why wouldn't it be invalidated when EMF is closed by the container?
-marina
Laird Nelson wrote:
> On Tue, Sep 27, 2011 at 8:58 PM, Laird Nelson <ljnelson_at_gmail.com
> <mailto:ljnelson_at_gmail.com>> wrote:
>
> File this one under working theory at the moment.
>
>
> Modified working theory. :-)
>
> What I am seeing is that the effects of one test method's
> operation on the database are visible--through
> EclipseLink-supplied EntityManagers only, not in the database--by
> other test methods.
>
>
> This turns out to be due to the well-documented EclipseLink session
> cache, which I'm actually fairly comfortable with.
>
> But here's the remaining problem. I would like--somehow, preferably
> from "outside" embedded Glassfish, but still in the same VM--to
> programmatically invalidate the session cache.
>
> Is there some kind of static EclipseLink method to do this, or an
> obscure asadmin command? Perhaps some kind of JMX invocation?
>
> I am aware that I can not use the session cache at all, by making use
> of the <property name="eclipselink.cache.shared.default" value="false"
> /> persistence.xml property. But that turns it off entirely.
>
> I am also aware that on a query by query basis (at least on the
> server) I can instruct a query not to use the shared cache. But for a
> variety of reasons I don't want to do that.
>
> What I want to do is, well, empty the cache in between my test
> methods. Given that (a) the cache can be invalidated and (b) I have
> an entire in-memory Glassfish instance at my disposal, it seems like I
> should be able to call some static method somewhere to do this, or to
> at least get a handle on the current EclipseLink session by rooting
> around in the Glassfish object itself
> (http://embedded-glassfish.java.net/nonav/apidocs/org/glassfish/embeddable/GlassFish.html).
>
> Any ideas?
>
> Thanks,
> Laird
>
> --
> http://about.me/lairdnelson
>