Rickard �berg wrote:
>
> Hey
>
> Imre Kifor wrote:
> > Assuming that DGC is available, the client GC notifies the server GC that it
> > doesn't use our object anymore. However, the server GC still cannot release
> > the object since there are local references to it (remember that at
> > least the server is holding on to it). Hence DGC is not enough for
> > reclaiming our object.
>
> But since the server is the *only* local referrer you know that if DGC
> tells you it's unreferenced, you can reclaim it. All clients must (?)
> use remote references for talking to it.
>
> > Java 2 solves the problem very nicely by introducing "weak" references.
> > Using Java 2, the server can hold on to our object with weak references thus
> > not restricting the GC. Hence the GC can reclaim the object after all (both
> > remote and local) strong references are released.
>
> How will this relate to the upcoming EJB spec. release where RMI/IIOP
> will be mandated (since RMI/IIOP doesn't support DGC!!!!)??
DGC is a non-issue if you take activation into account. You can encapsulate
any info you want in an IOR, which in the case of an EJBObject is
the home and the primary key. Since the EJBObject can be activated any
time a request comes in, and since all required state is part of the
IOR, you don't need the EJBObject to sit around occupying memory.
Further, since the server reserves the right to deactivate this object
at will, there's no need for GC or for weak references or whatever.
This is fast, extremely scaleable, and doesn't *require* Java2.
-Sriram
BEA/Weblogic
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff EJB-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
