Good to know!  I think Ginsu predates my awareness of Squeak, so I am 
not familiar with it.

Based on what you said, if I could find Ginsu, it might be an 
interesting place for me to start so that I can *add* aspect semantics 
to it.  If by "aspect semantics" you meant the advice and pointcuts, 
that's precisely what I'm interested in: the ability to modify the code 
of an existing method to support my package  (aspect) better, but by 
maintaining a logical separation between the original code and the code 
that I've added.  Further, aspect semantics not only allow me to do 
that, but others as well, in a matter that cleanly avoids many kinds of 
change conflicts, as multiple kinds of advice can affect a single 
method without necessarily causing a conflict.  That's our problem 
today with changesets and packages: there can only be 1 version of a 
method (or class) installed in the image, and there is no merging 
possibility.  That is, the definition of a method in one changeset 
cannot be merged safely with the definition of a method in another 
changset.  AOP techniques open the possibility if making that 
possible--if one is willing to add a new set of techniques to their 
toolbelt.

Thanks for the info about Ginsu!

On Nov 9, 2003, at 3:03 AM, ducasse wrote:

> Hi phil
>
> if we remove the aspect semantics, you have a code model like Ginsu 
> that covers packages, classes, methods....like monticello.
> With Ginsu the loading of a package was installing declarative 
> elements that are objects that the browsers uses in Class, 
> CompiledMethods and so on....
>
> So the idea is the same.
>
> Stef
>
>
>