Hi Gregor:
I am sorry I will miss your presentation at the DARPA
PI meeting.

I would like to provide you with viewgraphs 
for aspectual methods that are fully implemented
and that work together with AspectJ (we tried it).
Aspectual methods are modular units (Java methods)
of crosscutting implementation expressed in terms
of the DJ package.
Aspectual methods can be well explained in terms
of pointcuts and advice.

I will send you a brief paper tomorrow to give you some
background. I will also send you a couple of viewgraphs
on aspectual methods and DJ from which you are free
to choose.

I think this will not overlap with what Mira provides
which goes beyond aspectual methods.

Improvements to AspectJ:

Aspectual Methods:
Pointcut reduction, pointcuts based on class graph connectivity

Collaborations/Adapters:
Modeling each aspect as a unit with its own role model.
Mapping support from roles to classes.

Please let me know if we should only focus on
collaborations and adapters.

-- Karl

From mira@piserver.pi.informatik.tu-darmstadt.de  Wed Jan 31 05:14:43 2001
Reply-To: <mezini@informatik.tu-darmstadt.de>
From: "Mira Mezini" <mira@piserver.pi.informatik.tu-darmstadt.de>
To: "'Gregor Kiczales'" <gregor@cs.ubc.ca>, <lieber@ccs.neu.edu>
Cc: <wand@ccs.neu.edu>, <mira@ccs.neu.edu>
Subject: AW: slide or 2 re adaptors etc.
Date: Wed, 31 Jan 2001 11:15:27 +0100

Hi Gregor,

Sure, I can try that. I mean, I don't think that the message
would be "the mechanisms you and Mira have worked on can do 
that AspectJ can't". It will more likely be smth like 
"... can do probably a bit better (what ever this means) than 
in AspectJ"

Anyway, I will try to come up with something and hope you are 
willing to interact so that we really find out what this thing
is, if there is one. When do you need the 1 or 2 slides?


Mira


> Karl (and Mira),
> 
> As you've seen, Doug and Helen roped me into doing a little
> talk at the upcoming PI meeting.  Helen's idea was that I
> should share some of what we have learned about AOP with
> some of the newer projects that maybe don't have as much
> background.  My concern is to do that in a way that is
> useful and doesn't step on (too many) toes.
> 
> I'd really like to include a slide or two about what the mechanisms
> you and Mira have worked on can do that AspectJ can't.  The
> context for the slide would be "here's something we haven't
> been able to get into AspectJ", but which others have worked
> out.  If I make this all fit together right, I will have made
> it clear that one of the basic themes of AOP research is
> mechanisms for identifying crosscutting that are: intuitive,
> clear, powerful, composable etc.  So I would point to your stuff
> as being more one good example of more powerful kinds of crosscutting
> mechanisms.
> 
> In order to do this I would need a couple of slides that have
> a really intuitive example of what you can say that AspectJ can't.
> 
> Would you be willing to help me come up with such slides?  (To be
> honest, I should say that 'help' means 'do most of the work to
> produce such slides' because believe it or not I am even more
> busy than usual.)
> 
> Gregor
> 
> PS.  Mira, this is about the DARPA program that we are all in.