Hi Richard,

I had just noticed some small differences which I first thought might
have been due to a different policy being used. 
It's not the correct mailing list, but since you're here - is the
following a bug?

Index: RequiredBundleImpl.java
===================================================================
--- RequiredBundleImpl.java     (revision 886835)
+++ RequiredBundleImpl.java     (working copy)
@@ -74,7 +74,7 @@
             IModule[] dependents = ((ModuleImpl)
modules[modIdx]).getDependentRequirers();
             for (int depIdx = 0; (dependents != null) && (depIdx <
dependents.length); depIdx++)
             {
-                moduleList.add(dependents[modIdx]);
+                moduleList.add(dependents[depIdx]);
             }

Thanks,
Alan

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Richard S. Hall
Sent: 04 December 2009 15:11
To: OSGi Developer Mail List
Subject: Re: [osgi-dev] RequiredBundle clarification - PackageAdmin spec

On 12/4/09 10:06, Alan Keane wrote:
> Hi Tom,
> Thanks for the reply.. makes sense
> I was just a bit confused by the statement -/* *'if an exported 
> package/ /of the bundle it references has been updated or removed'/ by
> refreshPackages() - along with some differences I have noticed between

> felix and equinox.

Feel free to bring the differences to our attention, we love to debate
these issues. :-)

-> richard

> Thanks,
> Alan
>
> ----------------------------------------------------------------------
> --
> *From:* [email protected] 
> [mailto:[email protected]] *On Behalf Of *Thomas Watson
> *Sent:* 04 December 2009 13:16
> *To:* OSGi Developer Mail List
> *Subject:* Re: [osgi-dev] RequiredBundle clarification - PackageAdmin 
> spec
>
> The RequiredBundle#getRequiringBundles() is only intended to return 
> the bundles wired through Require-Bundle constraints. A RequiredBundle

> object is removal pending as soon as the bundle content it represents 
> has been updated or uninstalled. The same rule for removal pending 
> applies to ExportedPackage objects. As soon as the bundle content 
> which exports the package has been updated or uninstalled then the 
> ExportedPackage becomes stale. This is related to the class loader 
> which loads the bundle content. If this class loader becomes marked 
> for removal because of an update or uninstall then all exported 
> packages available for load from that bundle class loader must become 
> stale as well as the RequiredBundle object.
>
> Hope that helps.
>
> Tom
>
>
>
> Inactive hide details for "Alan Keane" ---12/04/2009 04:09:26 AM---Hi,

> Apologies if this is clearly stated somewhere but I coul"Alan Keane"
> ---12/04/2009 04:09:26 AM---Hi, Apologies if this is clearly stated 
> somewhere but I couldn't find an explicit statement in the spec..
>
>
> From:         
> "Alan Keane" <[email protected]>
>
> To:   
> <[email protected]>
>
> Date:         
> 12/04/2009 04:09 AM
>
> Subject:      
> [osgi-dev] RequiredBundle clarification - PackageAdmin spec
>
> ----------------------------------------------------------------------
> --
>
>
>
> Hi,
>
> Apologies if this is clearly stated somewhere but I couldn't find an 
> explicit statement in the spec..
> I just want to clarify something in relation to the
> RequiredBundle#getRequiringBundles() method. Should this method return

> all bundles that
>
> are currently wired to the underlying bundle as a result of both 
> Require-Bundle and explicit Import-Package statements?
>
> >From the spec:
> "The information about a required bundle provided by this object may 
> change. A RequiredBundle object becomes stale if an exported package 
> of the bundle it references has been updated or removed as a result of

> calling PackageAdmin.refreshPackages()). If this object becomes stale,

> its
> getSymbolicName()
> and getVersion() methods continue to return their original values,
> isRemovalPending() returns true, and getBundle() and
> getRequiringBundles() return null."
>
> To me, this implies that wiring as a result of explicit Import-Package

> statements should also be considered.
> If not, then should the RequiredBundle object only become stale when 
> the bundle it represents no longer exports any packages..
>
> As long as it continues to export 1+ packages then any bundle 
> previously wired as a result of Require-Bundle would remain a valid 
> requiring bundle.
>
> Thanks,
> Alan _______________________________________________
> OSGi Developer Mail List
> [email protected]
> https://mail.osgi.org/mailman/listinfo/osgi-dev
>
>
> _______________________________________________
> OSGi Developer Mail List
> [email protected]
> https://mail.osgi.org/mailman/listinfo/osgi-dev
_______________________________________________
OSGi Developer Mail List
[email protected]
https://mail.osgi.org/mailman/listinfo/osgi-dev

_______________________________________________
OSGi Developer Mail List
[email protected]
https://mail.osgi.org/mailman/listinfo/osgi-dev

Reply via email to