[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [aspectj-users] args() evaluation and around advice interaction
|
In what sense is a pointer to this on the stack
(a) implementation-dependent or (b) an argument? I don't
see the similarity to advice execution on this point.
I agree that it would be better to track the programmer's
model (i.e., the args declared in the source code), but
right now that's not the bytecode model (as it isn't in
the case of inner-class constructors). In general
we'd like to avoid special cases and depending on
our compiling the code being woven, so we're sticking
to the bytecode model.
But it's worth posting a bug because in the case of
advice execution, we have a some confidence that we
compiled the code, so it might be a reasonable
enhancement request for a special case, depending on
how people might implement things like around advice
inlining and closing over proceed statements. I'm
(happily) not involved in those decisions, so I
can't comment.
i.e.: email is talk, but bugs mean action!
Wes
Devi Prasad wrote:
Hello Wes
Thanks for the feedback. I didn't know that advice
execution is a variation point for different
implementations.
However, I feel it would still be a better idea to
keep out implementation dependent types (such as
AroundClosure) from interfering with args(). If it
were not so, wouldn't it break orthogonality with
args() matching for method execution, where we do not
explicitly specify (again implementation dependent)
object reference ("this" reference)? Is advice
execution, looked this way, any different from
method execution?
Thanks again.
__________________________________
Do you Yahoo!?
Yahoo! Photos: High-quality 4x6 digital prints for 25ยข
http://photos.yahoo.com/ph/print_splash
_______________________________________________
aspectj-users mailing list
aspectj-users@xxxxxxxxxxx
http://dev.eclipse.org/mailman/listinfo/aspectj-users
