Mailing-List: contact dev-help@felix.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@felix.apache.org
Received-SPF: pass (nike.apache.org: domain of gnodet@gmail.com designates
 66.249.92.172 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=message-id:date:from:to:subject:in-reply-to:mime-version:content-type:references;
        b=dU8uW488TreFNMh2DBD9jZ6GvvrJe0RnEgaG3JdHAWquO3DbE8YhAGeAXbLYyQT4p3hS13L+ZdG2CIQ0gXHx4hvQ4JxtTcmoS6EDz/jmE+01Dd7K2SUMhqlQefbYUFPRYnQghmHHZtSOesQVgV/Q7C3WSoHIQ5IDcQZZ2R8ps8M=
Message-ID: <b23ecedc0712200830m601766fs28ecdaac0c9ea515@mail.gmail.com>
Date: Thu, 20 Dec 2007 17:30:09 +0100
From: "Guillaume Nodet" <gnodet@gmail.com>
To: dev@felix.apache.org
Subject: Re: Framework error with ClassNotFoundException
In-Reply-To: <476A9260.6000403@ungoverned.org>
MIME-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_Part_1634_6772296.1198168209999"
References: <b23ecedc0712200704x30a2b0b5g147de448d313b563@mail.gmail.com>
	 <b23ecedc0712200729u1225f288n5d635a07b87585d4@mail.gmail.com>
	 <476A9260.6000403@ungoverned.org>

------=_Part_1634_6772296.1198168209999
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

I've raised a JIRA and attached a patch:
  https://issues.apache.org/jira/browse/FELIX-441

On Dec 20, 2007 5:03 PM, Richard S. Hall <heavy@ungoverned.org> wrote:

> Guillaume Nodet wrote:
> > I think the problem is in the R4SearchPolicyCore, line 345.
> > When the bundle can not be resolved, it throws a ClassNotFoundException.
> > I think it should rethrow the ResolveException, and the Felix#loadClass
> > method could catch it, fire the framework event and hide it by sending a
> > ClassNotFoundException.
> > If there's no problem, I'll provide a patch asap.
> >
>
> No, that won't work for the reason stated in the comment in the search
> policy. Further, ResolveException is a checked exception created in the
> searchpolicy package, which is a specific implementation of a search
> policy for the module layer. The module layer doesn't know about
> ResolveExceptions and thus IModule.getClass() does not throw such an
> exception.
>
> I guess the proper way to achieve what you want would be for
> loadBundleClass() to try to resolve the bundle before calling
> IModule.getClass(), that way it can catch the resolve exception and only
> report the error if necessary.
>
> -> richard
>
> >
> > On Dec 20, 2007 4:04 PM, Guillaume Nodet <gnodet@gmail.com> wrote:
> >
> >
> >> When a class can not be loaded, the o.a.f.framework.Felix class sends a
> >> framework event with an ERROR level.
> >> It can be very verbose in some cases: for example when using
> spring-osgi,
> >> for each spring bean defined, spring tries to load a class names
> >> XxxBeanInfo, where Xxx is the implemenation class name of the loaded
> spring
> >> bean.
> >> The piece of code is the loadBundleClass method, lines 1452 of the
> Felix
> >> class:
> >>
> >>     protected Class loadBundleClass(FelixBundle bundle, String name)
> >> throws ClassNotFoundException
> >>     {
> >>         Class clazz = bundle.getInfo
> ().getCurrentModule().getClass(name);
> >>         if (clazz == null)
> >>         {
> >>             // Throw exception.
> >>             ClassNotFoundException ex = new
> ClassNotFoundException(name);
> >>
> >>             // The spec says we must fire a framework error.
> >>             fireFrameworkEvent(
> >>                 FrameworkEvent.ERROR, bundle,
> >>                 new BundleException(ex.getMessage()));
> >>
> >>             throw ex;
> >>         }
> >>         return clazz;
> >>     }
> >>
> >> Given the comment, I've tried to find what the spec says.  I think the
> >> comment refers to the Bundle.loadClass methods.  Quoting the spec:
> >>
> >> If this bundle cannot be resolved, a Framework event of type Frame-
> >> workEvent.ERROR is fired containing a BundleException with details of
> the
> >> reason this bundle could not be resolved. This method must then throw a
> >> ClassNotFoundException.
> >>
> >> My understanding is that the framework event must be fired when the
> bundle
> >> can not be *resolved*, not each time a class can not be found.  I
> suppose
> >> such a case could only happen if the bundle has been installed but not
> yet
> >> resolved and someone is using the loadClass method.  Can someone please
> >> share thoughts on this ?
> >>
> >> --
> >> Cheers,
> >> Guillaume Nodet
> >> ------------------------
> >> Blog: http://gnodet.blogspot.com/
> >>
> >
> >
> >
> >
> >
>


-- 
Cheers,
Guillaume Nodet
------------------------
Blog: http://gnodet.blogspot.com/

------=_Part_1634_6772296.1198168209999--