On Jan 15, 2008 9:37 AM, Stuart McCulloch <[EMAIL PROTECTED]> wrote: > On 15/01/2008, Karl Pauls <[EMAIL PROTECTED]> wrote: > > > > > I am hoping that if I add my library jars in system classpath and set > > > bootdelegation="*", the library jar's content will be automatically > > > available to other bundles even if I don't set > > > org.osgi.framework.system.packages property. Is this not going to work? > > > > No. It doesn't work like this. You still need to make the packages > > your bundles explicitly import available form the system bundle (i.e., > > add them to the system.packages property). The bootdelegation is > > needed only for the case where you want to make packages available to > > bundles _without_ them needing to import them. However, if they import > > them then it doesn't work. In other words, this is not a bug in Felix > > but you will have to adjust your assumptions. > > > > What happens in your case is that the bootdelegation=* effectively > > short-cuts the whole resolving process and makes Felix use the > > bootclassloader _all_ the time. That one, however, doesn't know about > > classes that come from a bundle -- hence, the CNFE. As Felix > > Meschberger mentions, do try to avoid using the bootdelegation feature > > unless your really need it - it's not a convenience feature but a last > > resort. > > > but the spec does not say the search should end if a non java.* > bootdelegation > class can't be found in the parent classloader, which is what Felix does - > it treats > all packages in bootdelegation the same. > > the earlier part of the spec explicitly says the search ends at the parent > for java.* > packages, but makes no mention of stopping there for non java.* packages - > this > suggests to me that it should continue and search imports rather than > stopping > at the parent > > this appears to be the behaviour in the reference implementation and a > couple > of other frameworks I've tested. Regardless of the merits of bootdelegation, > I still > think this is something we need to fix (or clarify in the spec).
After re-reading the spec, I happen to agree. We should continue normally in case we can not load a class from the bootdelegation. I'll look into it. regards, Karl > regards, > > > > Karl > > > > > Thanks, > > > Sahoo > > > > > > --------------------------------------------------------------------- > > > To unsubscribe, e-mail: [EMAIL PROTECTED] > > > For additional commands, e-mail: [EMAIL PROTECTED] > > > > > > > > > > > > > > -- > > Karl Pauls > > [EMAIL PROTECTED] > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: [EMAIL PROTECTED] > > For additional commands, e-mail: [EMAIL PROTECTED] > > > > > > > -- > Cheers, Stuart > -- Karl Pauls [EMAIL PROTECTED] --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
