Scott T Weaver wrote:
No. And you get new type of errors. I've tried that before and it didn't help. I haven't got a stack trace lying around anymore but you can easily test it out.On Fri, 2004-05-21 at 17:39, David Sean Taylor wrote:
On May 21, 2004, at 2:23 PM, Scott T Weaver wrote:
This is fixed by removing the proxy="dynamic" from the class descriptor. I would suggest using proxies only in collection descriptors for now, which works fine. We can address the dynamic proxies/classloading issues at a later date as that is less of an issue than getting all sub projects using the same release of OJB.
Scott,
From your top-posting style response, Im left wondering if you are responding to my issue or David Le Strat's.
Are you saying that we need to remove the proxy=dynamic in order to fix the RowReader error
or are the errors one and the same?
Sorry about that. Yes, it looks like that when OJB generates the proxy class (RC6) it must be putting a RowReader object in the proxy instance (need to verify that though). I have a funny feeling that if you put the OJB jar in the shared/lib and kept proxy=dynamic the ClassNotFound would go away.
p.s.
In regards to dynamic proxies. Since, TMK, we are not seeing any issues
with performance and since they tend to cause more problems then they
solve, I propose we limit the use of dynamic proxies until we are
positive they work 100% of the time in all situations.
Yes Im 100% with you on that -- lets not push any of OJB's advanced features that are not well tested in test cases
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
