Is the testcase ProxyExamples#testReferenceProxyWithInheritance
a known failure on the 1.0.x branch at the moment?

I'm currently looking into a fix for dynamic proxies when using
per-thread metadata profiles - a follow-up on a fix I did for
CollectionProxy-instances a long time ago, see:
http://mail-archives.apache.org/mod_mbox/db-ojb-dev/200410.mbox/[EMAIL 
PROTECTED]

I have not actually tried to rollback my changes and see if they
are responsible but the are so isolated to per-thread mode so
I am pretty sure they are not...

testReferenceProxyWithInheritance fails because the proxied unloaded
instances are only instanceof Ownable, not the specific instanceof
checked by the following asserts in the testcase:
325:assertTrue(owned1 instanceof Ownable2);
326:assertTrue(owned2 instanceof Ownable1);

There is a comment in the testcase:
323:// This should be true according the docs of class Proxy and because
324:// the concrete classes are Onwable2Impl/Onwable1Impl

However, are not the comments and asserts wrong? The proxies are
instanceof Ownable until they are materialized, when the instances
are indeed instanceof the more specific interfaces.

If I change the testcase lines 325-326 to this:
 assertTrue(owned1 instanceof Ownable);
 assertTrue(owned2 instanceof Ownable);
 assertTrue(ProxyHelper.getRealObject(owned1) instanceof Ownable2);
 assertTrue(ProxyHelper.getRealObject(owned2) instanceof Ownable1);
...it passes.

Any comments on this? I am thinking wrong and the old testcase code
correct? (In that case something in proxy generation code needs to
be fixed, since the test currently fails.)

Regards,
 Martin


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to