Fwd: [aosd-discuss] "AOP considered harmful"

[Sean Gilbertson]

     Perhaps it would also solve my problem in trying to compile AOP  
Java code in Xcode (OS X, for Cocoa): for one thing, some join points  
are ineffective, most notably (for me), any interaction between Cocoa  
and my Interface Builder-instantiated controller class.  The  
constructor, and basically any call, can not be captured, because (I  
think) the code that calls the methods cannot be modified for  
whatever reason (perhaps it has something to do with an Objective-C/ 
Java bridge?), even though, as I understand it, AspectJ can affect  
change on compiled Java code.  Also, I have an issue with Aspect code  
not being woven into my other classes, if I've only modified the  
aspect.  That might be an environment issue, though.

On Apr 25, 2005, at 2:52 PM, Eric Tanter wrote:



>> Perhaps debugging can be more complicated - I don't feel qualified  
>> to comment, since I haven't used AOP enough - but can it really be  
>> more complicated or confusing than OOP debugging?.
>>
>>
>>
>
> I guess this is a problem of _current_ AOP implementations rather  
> than AOP itself.
> As of today, the Java VM is not "aspect aware", and debuggers are  
> usually pretty tied to the VM (eg via its debugging interface).  
> Code is preprocessed in order for aspects to apply. I'm pretty  
> convinced that in the end AOP needs to be supported by the main  
> language (not an extension based on transformation), that is also,  
> the VM. And then debugging of AO programs should be as hard -oups,  
> sorry, as easy- as debugging of OO programs.
>
> -- Eric
>
>
>
>
>