Re: [aspectj-users] Exposing the target object of a constructor call po

[Andy Clement <andrew.clement@xxxxxxxxx>]

Hello Henrique,

Sorry I didn't answer your private email to me yet, my inbox is a mess
- but this kind of question is always better on the mailing list (as
more than just me might answer).

Is this not sufficient for you?

after() returning(Foo f): call(Foo.new(..)) {
  System.out.println("Created a foo "+f);
}

Andy

2010/3/26 Henrique Rebêlo <hemr@xxxxxxxxxxx>:
> Hi all,
>
>  I was wondering whether we can expose the target object of a call pointcut.
>
>  Consider the following simple class:
>
> class C {
>
>    public C () {}
>    ...
> }
>
> along with the simple call pointcut and an after returning advice that uses
> such a pointcut:
>
> pointcut callConst(C c) : call (public C.new()) && target(c);
>
> after (C c) returning() : callConst(c){
>
>     //advice code
> }
>
> The call pointcut callConst() will match any call to C's constructor but
> only when it does not include the target designator. Without the target
> designator we cannot expose the target object.
>
> So, in the above example, the call is not matched! If we change the call to
> execution pointcut it matches normally (with context exposure )!
>
> I want to know the main problem to have accessible the target object in a
> call pointcut that is used in a after returning advice. As I said, we can
> expose the object If we use a  after returning advice with execution
> pointcut. But, according to the AspectJ semantics, after returning call will
> occurs just after the method finishes its execution, so I ask again, Why we
> cannot use context exposure during a call? (at this time the, the
> constructor already instantiated the object!!). Is there a reason for that?
> or just a language limitation...
>
> I just want to have interesting discussions about the AspectJ semantics!
>
> cheers,
> Henrique
>
> --
> ...............................................................................................................................
> Henrique Rebelo
> http://www.cin.ufpe.br/~hemr
> Informatics Center, UFPE, Brazil
>
> _______________________________________________
> aspectj-users mailing list
> aspectj-users@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/aspectj-users
>
>