[Sorry for the duplicate message, I forgot to send the message to commons-dev instead of [EMAIL PROTECTED]

[EMAIL PROTECTED] wrote:
 > Hi Edwin,
 >
 > I guess this really isn't a development list, so we should probably find
 > some other forum at some point; but hopefully this won't bother
others too
 > much.

OK then how about if we move this thread to [EMAIL PROTECTED]

> Great work here; this looks really robust! My only concern is that,
if the
> context class loader has been set, there's no fallback to the bootstrap
> classloader in case the thread's classloader can't find the class. I
know
> of environments that rely on this kind of behaviour, so I'm hoping you'll
> be willing to change FactoryFinder to allow it. To help, here's a patch
> that I *think* would do the trick (though I haven't tested it very
> thoroughly): (See attached file: FactoryFinder-diff.txt)


Thanks for the feedback.  I've incorporated code that will fallback to
the current classloader if a provider cannot be found using the context
classloader, which if I recall is what your patch does.  (The current
classloader is different from the bootstrap classloader which is
represented as a 'null' value in many implementations.  The java
classloading API is a bit messy unfortunately. :-( )  Let me know if
this works for you.

-Edwin




Reply via email to