RE: breaking changes in 4.0 for CDI (was: RE: [cdt-debug-dev] Lis t of P

[John Cortell <john.cortell@xxxxxxxxxxxxx>]

At 01:45 PM 10/23/2006, Alain Magloire wrote:


> > From: cdt-debug-dev-bounces@xxxxxxxxxxx [mailto:cdt-debug-dev-
> > bounces@xxxxxxxxxxx] On Behalf Of Mikhail Khodjaiants
> > Sent: Wednesday, October 18, 2006 1:31 PM
> > To: CDT Debug developers list
> > Subject: RE: breaking changes in 4.0 for CDI (was: RE: [cdt-debug-dev]
> > List of Plan items)
> >
> > > For example adding IAdaptable to ICDIObject and ICDISessionObject are
> > breaking changes.
> >
> > There are many ways to minimize the impact of this type of changes.
> > For example, introduce new ICDIObject2 and ICDISessionObject2 interfaces
> > and add two new abstract classes:
> >
> > abstract class AbstractObject implements ICDIObject, ICDIObject2
> >
> > abstract class AbstractSessionObject implements ICDISessionObject,
> > ICDISessionObject2
> >
> > Implementors can use theses classes as base classes for their hierrachy.
> > It's not a very good solution, but it covers most of the use cases.
> >
> > I agree, 4.0 is a major release, but we at least should try to make the
> > transition easier.
> >
>
> It seems to be the same i.e. the client must still do instanceof:
>
>  if (obj1 instanceof ICDIObject2) {
>         ICDIObject2 obj2 = (ICDIObject2)obj1;
>         MyClass clz = obj2.getAdapter(MyClass.class);
>  }
>
> So still a lot of instanceof for little gain, IMHO.
>
> So, why the hesitation?  I would prefer to fix things right at the root and
> a 4.0 release is a good candidate, IMHO?
>
> Any other votes/takes on this?

I agree. You end up with a new instanceof in order to avoid another.

John