[self-interest] Aspect oriented and self

Steve Dekorte steve at dekorte.com
Wed Aug 21 13:13:15 UTC 2002


On Wednesday, August 21, 2002, at 01:14  AM, Thorsten Dittmar wrote:
> For example, Self has mixins and dynamic inheritance with this two
> mechanisms you can solve a lot of the AOP stuff.

Yes, I've thought the same thing. Of the AOP descriptions I've read, it 
all sounds like just adding some extra parent links in Self would 
accomplish the same thing. I once asked one of the PARC guys working on 
AspectJ why they didn't use a more dynamic language instead of trying to 
put these features into a language that doesn't want to be dynamic. He 
said that they can get funded if they use Java.

I've also wondered what AOP has to do with the original (excellent) 
paper that seemed to coin the term.(I think Jecel posted it here a while 
back). That paper focused on the problem of how abstractions make 
assumptions, so no given abstraction will be well suited for all 
problems. The examples given were the hardware abstractions many 
operating systems provide. Dynamic language features allow for greater 
flexibility and therefore less dependence on initial assumptions, so I 
can see why tinkering with some dynamic features is related to the 
problem - but if your goal is to address the abstraction assumption 
problem, I don't see how the one feature AspectJ adds to Java makes a 
significant contribution to the solution. Why not go farther and say 
"there's all sorts of other assumptions that inflexible systems force 
upon your abstractions, so let's see what we can do with highly flexible 
systems"?

Sorry for the rant. That's been building up for a while. :-]

Cheers,
Steve





More information about the Self-interest mailing list