[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
RE: [aspectj-users] when final is not
|
I don't like the idea of adding Xlint warnings. I'd rather use declare
warning to explicitly add such warnings, and even encourage the community to
contribute such to a library of common rules.
I definitely don't want Xlint warnings for advice that affects non-public
methods that aren't accessible to the aspect, and I think advice on final
methods is in the same category. I personally don't like adding Xlint
warnings for ITD's on final classes, but I think there's a reasonable
argument in favor of them.
Ron
On Wed, 14 Jan 2004 16:58:29 -0500, Echlin Harmer, Elizabeth wrote
> I agree with Ron that the ability to specify the constraints you
> want on your aspect using isFinal() or @privileged is a better way
> to go than to limit the aspect compiler.
>
> And yes Yes YES an Xlint option, because we know that the mistake the
> programmer is likely to make is not to forget to declare error if the
> pointcut unintentionally hits one of the known final classes, but to
> fail to realize that there is a final class somewhere down in the
> depths of the system that might be affected by the aspect, and so
> not even think about the possibility that such a constraint might be
> needed.
>
> Finally, I agree with Gregor that the implementation is not the
> aspect, and we want to watch what unconscious assumptions and mental
> models we are using for thinking and writing about aspects.
>
> However, I also think it is possible to say 'the aspect modified my final
> class' while thinking 'the aspect modified _the externally visible behaviour
> of the part of the system implemented within_ my final class'. As
> long as we are careful about what we are thinking, and are confident
> that our listeners won't be confused or misled, IMHO the concise
> form is acceptable.
>
> Elizabeth
Ron Bodkin
Chief Technology Officer
New Aspects
m: (415) 509-2895