2009/11/26 Guo Du <mrdu...@gmail.com> > On Thu, Nov 26, 2009 at 1:28 PM, Ivan <xhh...@gmail.com> wrote: > > But what does "A resolved exporter must be preferred over an unresolved > > exporter." mean ? :-( > It means if both bound A(installed) and B(resolved) export package P > for same version V, framework will get the P from B and leave A in > installed status. > > > In the doc, it says that the framework will try to resolve the bundle > once > > it is installed, and it will also try to resolve the bundle in installed > It's not clear to say "framework will try to resolve the bundle once > it is installed". > This resolve may means parse the bundle header. But it's definite not > to put the bundle in resolved status after installed. > > > As the spec says, the classloader for the bundle could be created once > it > > is resolved, so why does the framework need to start the bundle if its > > export packages are wired to other bundles ? Is it the expected behavior > by > > the spec ? > Yes. Start the bundle to resolved status is the way to mark bundle as > IN USE. The flow are: > Resolved -> Create classloader -> Load class for others bundles > > So no start/resolved, no class available to be consumed. >
So I suppose the word "start" here means that its status would be "resolved", not "active". Right ? Many thanks ! > -Guo > > --------------------------------------------------------------------- > To unsubscribe, e-mail: users-unsubscr...@felix.apache.org > For additional commands, e-mail: users-h...@felix.apache.org > > -- Ivan
