Re: Anyone on ll1-discuss besides spammers?

[Steve Dekorte <address@hidden>]

On Apr 13, 2004, at 2:16 PM, Johan Ovlinger wrote:
> The implementation often requires some sort of hook. However, in the 
> same way object orientation is more than just vtable management, aop 
> is more than just the interception mechanism.
>
> The fundamental building blocks of AOP is rather the mechanism by 
> which we specify what points to use as hooks (pointcuts to specify 
> join-points) and how those points should be modified (the advice).
>
> Hooks on a methods are just one instance of that (albeit the most 
> popular).  Other examples are:
>
> 1. Program structuring conventions (such as the law of demeter)
> 2. Real time constraints
> 3. Invariant mainenance
>
> The ultimate goal is to be able to
>
> A capture each concern in some unit of program structre (call it a 
> module)
> B compose these units into a runnable program
> C be able to understand each unit separately and then compose that 
> understanding into an understanding of the big picture.
>
> The problem is that getting all three at once is a bit tricky. The 
> abilities that appear to be need in order to enable A seem to  make C 
> very difficult.

I'm afraid I don't follow you. Could you describe it in the more 
concrete terms of implementation? For example, how do we determine if a 
given programming language supports AOP?

-- Steve