[aosd-discuss] Decoupling of Pointcut and Advice concepts

[Eddy Truyen]

In general I think the subject of the discussion should not about
decoupling pointcut and advice concepts from each other, but how to
achieve a clean separation between 
(1) programming aspects as components with well defined interfaces

(2) deploying aspects (possibly by a third party)

Thus lifting the discussion to concepts of component-based development.
This is actually what we did in Lasagne, see for
example http://www.cs.kuleuven.ac.be/~eddy/PUBLICATIONS/icse2001.pdf.

I think the challenge lies in trying to increase the expressive power of
join point models as much as possible without disabling desirable software
engineering properties that already exist in component-based development.

Cu

Eddy.


On Mon, 16 Feb 2004, Daniel Lohmann wrote:

> We are talking a lot about the principle of separation of concerns. It 
> is a good idea, to *do* separate things that *do not* belong togehter. 
> However, the reverse consequence, that it is a good idea to *do not* 
> separate things that *do* belong together, is also true! I frequently 
> found myself in a situation where I am writing aspect code, that is 
> intendent to interfer with only a well defined set of classes and 
> methods. These aspects are simply not intendent to be reused.
> 
> It therefore depends a lot on the achieved goal if decoupling or 
> coupling aspects and pointcuts is the way to go. IMHO the way followed 
> by AspectJ and AspectC++ is a pragmatic approach, because it offers both 
> options.
> 
> (Beeing a C++ fan I simply regret that a language can have too much 
> expressive power. Don't mutilate your toolkit - train your sofware 
> developers to use it appropriately! ;-) )
> 
> 
> Daniel
> 
> P.S. Christopher, I am looking forward to the next version of your C++ 
> article
> 
> 
> 
> __________________________________________________
> AOSD Discuss mailing list    -    discuss at aosd.net
> To unsubscribe go to http://aosd.net
>